Skip to content

Connectors API Reference

This page provides API reference documentation for the Connectors module, which contains classes and functions for creating complex connectivity patterns in BMTK networks.

Utility Functions

bmtool.connectors.num_prop(ratio, N)

Calculate numbers of total N in proportion to ratio.

Parameters:

ratio : array-like Proportions to distribute N across. N : int Total number to distribute.

Returns:

numpy.ndarray Array of integers that sum to N, proportionally distributed according to ratio.

Source code in bmtool/connectors.py 
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
def num_prop(ratio, N):
    """
    Calculate numbers of total N in proportion to ratio.

    Parameters:
    -----------
    ratio : array-like
        Proportions to distribute N across.
    N : int
        Total number to distribute.

    Returns:
    --------
    numpy.ndarray
        Array of integers that sum to N, proportionally distributed according to ratio.
    """
    ratio = np.asarray(ratio)
    p = np.cumsum(np.insert(ratio.ravel(), 0, 0))  # cumulative proportion
    return np.diff(np.round(N / p[-1] * p).astype(int)).reshape(ratio.shape)

bmtool.connectors.decision(prob, size=None, rng=None)

Make random decision(s) based on input probability.

Parameters:

prob : float Probability threshold between 0 and 1. size : int or tuple, optional Size of the output array. If None, a single decision is returned. rng : numpy.random.Generator, optional Random number generator. If None, uses the default generator.

Returns:

bool or numpy.ndarray Boolean result(s) of the random decision(s). True if the random number is less than prob, False otherwise.

Source code in bmtool/connectors.py 
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
def decision(prob, size=None, rng=None):
    """
    Make random decision(s) based on input probability.

    Parameters:
    -----------
    prob : float
        Probability threshold between 0 and 1.
    size : int or tuple, optional
        Size of the output array. If None, a single decision is returned.
    rng : numpy.random.Generator, optional
        Random number generator. If None, uses the default generator.

    Returns:
    --------
    bool or numpy.ndarray
        Boolean result(s) of the random decision(s). True if the random number
        is less than prob, False otherwise.
    """
    rng = _get_default_rng() if rng is None else rng
    return rng.random(size) < prob

bmtool.connectors.decisions(prob, rng=None)

Make multiple random decisions based on input probabilities.

Parameters:

prob : array-like Array of probability thresholds between 0 and 1. rng : numpy.random.Generator, optional Random number generator. If None, uses the default generator.

Returns:

numpy.ndarray Boolean array with the same shape as prob, containing results of the random decisions.

Source code in bmtool/connectors.py 
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
def decisions(prob, rng=None):
    """
    Make multiple random decisions based on input probabilities.

    Parameters:
    -----------
    prob : array-like
        Array of probability thresholds between 0 and 1.
    rng : numpy.random.Generator, optional
        Random number generator. If None, uses the default generator.

    Returns:
    --------
    numpy.ndarray
        Boolean array with the same shape as prob, containing results of
        the random decisions.
    """
    prob = np.asarray(prob)
    rng = _get_default_rng() if rng is None else rng
    return rng.random(prob.shape) < prob

bmtool.connectors.euclid_dist(p1, p2)

Euclidean distance between two points p1, p2: Coordinates in numpy array

Source code in bmtool/connectors.py 
 94
 95
 96
 97
 98
 99
100
def euclid_dist(p1, p2):
    """
    Euclidean distance between two points
    p1, p2: Coordinates in numpy array
    """
    dvec = np.asarray(p1) - np.asarray(p2)
    return (dvec @ dvec) ** 0.5

bmtool.connectors.spherical_dist(node1, node2)

Spherical distance between two input nodes

Source code in bmtool/connectors.py 
103
104
105
def spherical_dist(node1, node2):
    """Spherical distance between two input nodes"""
    return euclid_dist(node1["positions"], node2["positions"]).item()

bmtool.connectors.cylindrical_dist_z(node1, node2)

Cylindircal distance between two input nodes (ignoring z-axis)

Source code in bmtool/connectors.py 
108
109
110
def cylindrical_dist_z(node1, node2):
    """Cylindircal distance between two input nodes (ignoring z-axis)"""
    return euclid_dist(node1["positions"][:2], node2["positions"][:2]).item()

Probability Functions

bmtool.connectors.ProbabilityFunction

Bases: ABC

Abstract base class for connection probability function

Source code in bmtool/connectors.py 
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
class ProbabilityFunction(ABC):
    """Abstract base class for connection probability function"""

    @abstractmethod
    def probability(self, *arg, **kwargs):
        """Allow numpy array input and return probability in numpy array"""
        return NotImplemented

    @abstractmethod
    def __call__(self, *arg, **kwargs):
        """Return probability within [0, 1] for single input"""
        return NotImplemented

    @abstractmethod
    def decisions(self, *arg, **kwargs):
        """Return bool array of decisions according probability"""
        return NotImplemented

__call__(*arg, **kwargs) abstractmethod

Return probability within [0, 1] for single input

Source code in bmtool/connectors.py 
122
123
124
125
@abstractmethod
def __call__(self, *arg, **kwargs):
    """Return probability within [0, 1] for single input"""
    return NotImplemented

decisions(*arg, **kwargs) abstractmethod

Return bool array of decisions according probability

Source code in bmtool/connectors.py 
127
128
129
130
@abstractmethod
def decisions(self, *arg, **kwargs):
    """Return bool array of decisions according probability"""
    return NotImplemented

probability(*arg, **kwargs) abstractmethod

Allow numpy array input and return probability in numpy array

Source code in bmtool/connectors.py 
117
118
119
120
@abstractmethod
def probability(self, *arg, **kwargs):
    """Allow numpy array input and return probability in numpy array"""
    return NotImplemented

bmtool.connectors.DistantDependentProbability

Bases: ProbabilityFunction

Base class for distance dependent probability

Source code in bmtool/connectors.py 
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
class DistantDependentProbability(ProbabilityFunction):
    """Base class for distance dependent probability"""

    def __init__(self, min_dist=0.0, max_dist=np.inf):
        assert min_dist >= 0 and min_dist < max_dist
        self.min_dist, self.max_dist = min_dist, max_dist

    def __call__(self, dist, *arg, **kwargs):
        """Return probability for single distance input"""
        if dist >= self.min_dist and dist <= self.max_dist:
            return self.probability(dist)
        else:
            return 0.0

    def decisions(self, dist, rng=None):
        """Return bool array of decisions given distance array"""
        dist = np.asarray(dist)
        dec = np.zeros(dist.shape, dtype=bool)
        mask = (dist >= self.min_dist) & (dist <= self.max_dist)
        dist = dist[mask]
        prob = np.empty(dist.shape)
        prob[:] = self.probability(dist)
        dec[mask] = decisions(prob, rng=rng)
        return dec

__call__(dist, *arg, **kwargs)

Return probability for single distance input

Source code in bmtool/connectors.py 
140
141
142
143
144
145
def __call__(self, dist, *arg, **kwargs):
    """Return probability for single distance input"""
    if dist >= self.min_dist and dist <= self.max_dist:
        return self.probability(dist)
    else:
        return 0.0

decisions(dist, rng=None)

Return bool array of decisions given distance array

Source code in bmtool/connectors.py 
147
148
149
150
151
152
153
154
155
156
def decisions(self, dist, rng=None):
    """Return bool array of decisions given distance array"""
    dist = np.asarray(dist)
    dec = np.zeros(dist.shape, dtype=bool)
    mask = (dist >= self.min_dist) & (dist <= self.max_dist)
    dist = dist[mask]
    prob = np.empty(dist.shape)
    prob[:] = self.probability(dist)
    dec[mask] = decisions(prob, rng=rng)
    return dec

bmtool.connectors.UniformInRange

Bases: DistantDependentProbability

Constant probability within a distance range

Source code in bmtool/connectors.py 
159
160
161
162
163
164
165
166
167
168
169
class UniformInRange(DistantDependentProbability):
    """Constant probability within a distance range"""

    def __init__(self, p=0.0, min_dist=0.0, max_dist=np.inf):
        super().__init__(min_dist=min_dist, max_dist=max_dist)
        self.p = np.array(p)
        assert self.p.size == 1
        assert p >= 0.0 and p <= 1.0

    def probability(self, dist):
        return self.p

bmtool.connectors.gaussian(x, mean=0.0, stdev=1.0, pmax=NORM_COEF)

Gaussian function. Default is the PDF of standard normal distribution

Source code in bmtool/connectors.py 
175
176
177
178
def gaussian(x, mean=0.0, stdev=1.0, pmax=NORM_COEF):
    """Gaussian function. Default is the PDF of standard normal distribution"""
    x = (x - mean) / stdev
    return pmax * np.exp(-x * x / 2)

bmtool.connectors.GaussianDropoff

Bases: DistantDependentProbability

Connection probability class that follows a Gaussian function of distance.

This class calculates connection probabilities using a Gaussian function of the distance between cells, with options for spherical or cylindrical metrics.

Parameters:

mean : float, optional Mean parameter of the Gaussian function, typically 0 for peak at origin. stdev : float, optional Standard deviation parameter controlling the width of the Gaussian. min_dist : float, optional Minimum distance for connections. Below this distance, probability is zero. max_dist : float, optional Maximum distance for connections. Above this distance, probability is zero. pmax : float, optional Maximum probability value at the peak of the Gaussian function. ptotal : float, optional Overall connection probability within the specified distance range. If provided, pmax is calculated to achieve this overall probability. ptotal_dist_range : tuple, optional Distance range (min_dist, max_dist) for calculating pmax when ptotal is provided. dist_type : str, optional Distance metric to use, either 'spherical' (default) or 'cylindrical'.

Notes:

When ptotal is specified, the maximum probability (pmax) is calculated to achieve the desired overall connection probability within the specified distance range, assuming homogeneous cell density.

Source code in bmtool/connectors.py 
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
class GaussianDropoff(DistantDependentProbability):
    """
    Connection probability class that follows a Gaussian function of distance.

    This class calculates connection probabilities using a Gaussian function
    of the distance between cells, with options for spherical or cylindrical metrics.

    Parameters:
    -----------
    mean : float, optional
        Mean parameter of the Gaussian function, typically 0 for peak at origin.
    stdev : float, optional
        Standard deviation parameter controlling the width of the Gaussian.
    min_dist : float, optional
        Minimum distance for connections. Below this distance, probability is zero.
    max_dist : float, optional
        Maximum distance for connections. Above this distance, probability is zero.
    pmax : float, optional
        Maximum probability value at the peak of the Gaussian function.
    ptotal : float, optional
        Overall connection probability within the specified distance range.
        If provided, pmax is calculated to achieve this overall probability.
    ptotal_dist_range : tuple, optional
        Distance range (min_dist, max_dist) for calculating pmax when ptotal is provided.
    dist_type : str, optional
        Distance metric to use, either 'spherical' (default) or 'cylindrical'.

    Notes:
    ------
    When ptotal is specified, the maximum probability (pmax) is calculated to achieve
    the desired overall connection probability within the specified distance range,
    assuming homogeneous cell density.
    """

    def __init__(
        self,
        mean=0.0,
        stdev=1.0,
        min_dist=0.0,
        max_dist=np.inf,
        pmax=1,
        ptotal=None,
        ptotal_dist_range=None,
        dist_type="spherical",
    ):
        super().__init__(min_dist=min_dist, max_dist=max_dist)
        self.mean, self.stdev = mean, stdev
        self.ptotal = ptotal
        self.ptotal_dist_range = (
            (min_dist, max_dist) if ptotal_dist_range is None else ptotal_dist_range
        )
        self.dist_type = dist_type if dist_type in ["cylindrical"] else "spherical"
        self.pmax = pmax if ptotal is None else self.calc_pmax_from_ptotal()
        self.set_probability_func()

    def calc_pmax_from_ptotal(self):
        """
        Calculate the pmax value such that the expected overall connection
        probability to all possible targets within the distance range [r1, r2]=
        `ptotal_dist_range` equals ptotal, assuming homogeneous cell density.
        That is, integral_r1^r2 {g(r)p(r)dr} = ptotal, where g is the Gaussian
        function with pmax, p(r) is the cell density per unit distance at r
        normalized by total cell number within the distance range.
        For cylindrical distance, p(r) = 2 * r / (r2^2 - r1^2)
        For spherical distance, p(r) = 3 * r^2 / (r2^3 - r1^3)
        The solution has a closed form except that te error function erf is in
        the expression, but only when resulting pmax <= 1.

        Caveat: When the calculated pmax > 1, the actual overall probability
        will be lower than expected and all cells within certain distance will
        be always connected. This usually happens when the distance range is
        set too wide. Because a large population will be included for
        evaluating ptotal, and there will be a significant drop in the Gaussian
        function as distance gets further. So, a large pmax will be required to
        achieve the desired ptotal.
        """
        mu, sig = self.mean, self.stdev
        r1, r2 = self.ptotal_dist_range[:2]
        x1, x2 = (r1 - mu) / sig, (r2 - mu) / sig  # normalized distance
        if self.dist_type == "cylindrical":
            dr = r2**2 - r1**2

            def F(x):
                f1 = sig * mu / NORM_COEF * erf(x / 2**0.5)
                f2 = -2 * sig * sig * gaussian(x, pmax=1.0)
                return f1 + f2
        else:
            dr = r2**3 - r1**3

            def F(x):
                f1 = 1.5 * sig * (sig**2 + mu**2) / NORM_COEF * erf(x / 2**0.5)
                f2 = -3 * sig * sig * (2 * mu + sig * x) * gaussian(x, pmax=1.0)
                return f1 + f2

        return self.ptotal * dr / (F(x2) - F(x1))

    def probability(self):
        pass  # to be set up in set_probability_func()

    def compute_ptotal_integral(self, num_points=1000):
        """
        Compute the total integrated probability by numerically integrating
        the Gaussian probability function over the ptotal_dist_range.

        This is useful when ptotal was not explicitly set and needs to be
        calculated from the distribution parameters (pmax, stdev, etc.).

        Parameters:
        -----------
        num_points : int, optional
            Number of points to use for numerical integration (default: 1000).

        Returns:
        --------
        float
            The integrated total probability over the ptotal_dist_range.

        Notes:
        ------
        For cylindrical distance: integrates p(r) * 2Ï€r over the range,
        normalized by the total cylindrical area.
        For spherical distance: integrates p(r) * 4πr² over the range,
        normalized by the total spherical volume.
        """
        from scipy.integrate import trapz

        # Handle both range objects and tuples
        if isinstance(self.ptotal_dist_range, range):
            r1 = self.ptotal_dist_range.start if self.ptotal_dist_range.start is not None else 0
            r2 = self.ptotal_dist_range.stop
        else:
            r1, r2 = self.ptotal_dist_range[0], self.ptotal_dist_range[1]

        distances = np.linspace(r1, r2, num_points)
        probabilities = self.probability(distances)

        if self.dist_type == "cylindrical":
            # For cylindrical: weight by 2Ï€r (density at radius r in 2D)
            weights = 2 * np.pi * distances
            # Normalize by total cylindrical area in the range
            total_area = np.pi * (r2**2 - r1**2)
            integrand = probabilities * weights / total_area
        else:  # spherical
            # For spherical: weight by 4πr² (density at radius r in 3D)
            weights = 4 * np.pi * distances ** 2
            # Normalize by total spherical volume in the range
            total_volume = (4/3) * np.pi * (r2**3 - r1**3)
            integrand = probabilities * weights / total_volume

        ptotal = trapz(integrand, distances)
        return ptotal

    def set_probability_func(self):
        """Set up function for calculating probability"""
        keys = ["mean", "stdev", "pmax"]
        kwargs = {key: getattr(self, key) for key in keys}
        probability = partial(gaussian, **kwargs)

        # Verify maximum probability
        # (is not self.pmax if self.mean outside distance range)
        bounds = (self.min_dist, min(self.max_dist, 1e9))
        pmax = (
            self.pmax
            if self.mean >= bounds[0] and self.mean <= bounds[1]
            else probability(np.asarray(bounds)).max()
        )
        if pmax > 1:
            d = minimize_scalar(
                lambda x: (probability(x) - 1) ** 2, method="bounded", bounds=bounds
            ).x
            warn = (
                "\nWarning: Maximum probability=%.3f is greater than 1. "
                "Probability crosses 1 at distance %.3g.\n"
            ) % (pmax, d)
            if self.ptotal is not None:
                warn += " ptotal may not be reached."
            print(warn, flush=True)
            self.probability = lambda dist: np.fmin(probability(dist), 1.0)
        else:
            self.probability = probability

calc_pmax_from_ptotal()

Calculate the pmax value such that the expected overall connection probability to all possible targets within the distance range [r1, r2]= ptotal_dist_range equals ptotal, assuming homogeneous cell density. That is, integral_r1^r2 {g(r)p(r)dr} = ptotal, where g is the Gaussian function with pmax, p(r) is the cell density per unit distance at r normalized by total cell number within the distance range. For cylindrical distance, p(r) = 2 * r / (r2^2 - r1^2) For spherical distance, p(r) = 3 * r^2 / (r2^3 - r1^3) The solution has a closed form except that te error function erf is in the expression, but only when resulting pmax <= 1.

Caveat: When the calculated pmax > 1, the actual overall probability will be lower than expected and all cells within certain distance will be always connected. This usually happens when the distance range is set too wide. Because a large population will be included for evaluating ptotal, and there will be a significant drop in the Gaussian function as distance gets further. So, a large pmax will be required to achieve the desired ptotal.

Source code in bmtool/connectors.py 
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
def calc_pmax_from_ptotal(self):
    """
    Calculate the pmax value such that the expected overall connection
    probability to all possible targets within the distance range [r1, r2]=
    `ptotal_dist_range` equals ptotal, assuming homogeneous cell density.
    That is, integral_r1^r2 {g(r)p(r)dr} = ptotal, where g is the Gaussian
    function with pmax, p(r) is the cell density per unit distance at r
    normalized by total cell number within the distance range.
    For cylindrical distance, p(r) = 2 * r / (r2^2 - r1^2)
    For spherical distance, p(r) = 3 * r^2 / (r2^3 - r1^3)
    The solution has a closed form except that te error function erf is in
    the expression, but only when resulting pmax <= 1.

    Caveat: When the calculated pmax > 1, the actual overall probability
    will be lower than expected and all cells within certain distance will
    be always connected. This usually happens when the distance range is
    set too wide. Because a large population will be included for
    evaluating ptotal, and there will be a significant drop in the Gaussian
    function as distance gets further. So, a large pmax will be required to
    achieve the desired ptotal.
    """
    mu, sig = self.mean, self.stdev
    r1, r2 = self.ptotal_dist_range[:2]
    x1, x2 = (r1 - mu) / sig, (r2 - mu) / sig  # normalized distance
    if self.dist_type == "cylindrical":
        dr = r2**2 - r1**2

        def F(x):
            f1 = sig * mu / NORM_COEF * erf(x / 2**0.5)
            f2 = -2 * sig * sig * gaussian(x, pmax=1.0)
            return f1 + f2
    else:
        dr = r2**3 - r1**3

        def F(x):
            f1 = 1.5 * sig * (sig**2 + mu**2) / NORM_COEF * erf(x / 2**0.5)
            f2 = -3 * sig * sig * (2 * mu + sig * x) * gaussian(x, pmax=1.0)
            return f1 + f2

    return self.ptotal * dr / (F(x2) - F(x1))

compute_ptotal_integral(num_points=1000)

Compute the total integrated probability by numerically integrating the Gaussian probability function over the ptotal_dist_range.

This is useful when ptotal was not explicitly set and needs to be calculated from the distribution parameters (pmax, stdev, etc.).

Parameters:

num_points : int, optional Number of points to use for numerical integration (default: 1000).

Returns:

float The integrated total probability over the ptotal_dist_range.

Notes:

For cylindrical distance: integrates p(r) * 2πr over the range, normalized by the total cylindrical area. For spherical distance: integrates p(r) * 4πr² over the range, normalized by the total spherical volume.

Source code in bmtool/connectors.py 
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
def compute_ptotal_integral(self, num_points=1000):
    """
    Compute the total integrated probability by numerically integrating
    the Gaussian probability function over the ptotal_dist_range.

    This is useful when ptotal was not explicitly set and needs to be
    calculated from the distribution parameters (pmax, stdev, etc.).

    Parameters:
    -----------
    num_points : int, optional
        Number of points to use for numerical integration (default: 1000).

    Returns:
    --------
    float
        The integrated total probability over the ptotal_dist_range.

    Notes:
    ------
    For cylindrical distance: integrates p(r) * 2Ï€r over the range,
    normalized by the total cylindrical area.
    For spherical distance: integrates p(r) * 4πr² over the range,
    normalized by the total spherical volume.
    """
    from scipy.integrate import trapz

    # Handle both range objects and tuples
    if isinstance(self.ptotal_dist_range, range):
        r1 = self.ptotal_dist_range.start if self.ptotal_dist_range.start is not None else 0
        r2 = self.ptotal_dist_range.stop
    else:
        r1, r2 = self.ptotal_dist_range[0], self.ptotal_dist_range[1]

    distances = np.linspace(r1, r2, num_points)
    probabilities = self.probability(distances)

    if self.dist_type == "cylindrical":
        # For cylindrical: weight by 2Ï€r (density at radius r in 2D)
        weights = 2 * np.pi * distances
        # Normalize by total cylindrical area in the range
        total_area = np.pi * (r2**2 - r1**2)
        integrand = probabilities * weights / total_area
    else:  # spherical
        # For spherical: weight by 4πr² (density at radius r in 3D)
        weights = 4 * np.pi * distances ** 2
        # Normalize by total spherical volume in the range
        total_volume = (4/3) * np.pi * (r2**3 - r1**3)
        integrand = probabilities * weights / total_volume

    ptotal = trapz(integrand, distances)
    return ptotal

set_probability_func()

Set up function for calculating probability

Source code in bmtool/connectors.py 
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
def set_probability_func(self):
    """Set up function for calculating probability"""
    keys = ["mean", "stdev", "pmax"]
    kwargs = {key: getattr(self, key) for key in keys}
    probability = partial(gaussian, **kwargs)

    # Verify maximum probability
    # (is not self.pmax if self.mean outside distance range)
    bounds = (self.min_dist, min(self.max_dist, 1e9))
    pmax = (
        self.pmax
        if self.mean >= bounds[0] and self.mean <= bounds[1]
        else probability(np.asarray(bounds)).max()
    )
    if pmax > 1:
        d = minimize_scalar(
            lambda x: (probability(x) - 1) ** 2, method="bounded", bounds=bounds
        ).x
        warn = (
            "\nWarning: Maximum probability=%.3f is greater than 1. "
            "Probability crosses 1 at distance %.3g.\n"
        ) % (pmax, d)
        if self.ptotal is not None:
            warn += " ptotal may not be reached."
        print(warn, flush=True)
        self.probability = lambda dist: np.fmin(probability(dist), 1.0)
    else:
        self.probability = probability

bmtool.connectors.NormalizedReciprocalRate

Bases: ProbabilityFunction

Reciprocal connection probability given normalized reciprocal rate. Normalized reciprocal rate is defined as the ratio between the reciprocal connection probability and the connection probability for a randomly connected network where the two unidirectional connections between any pair of neurons are independent. NRR = pr / (p0 * p1)

Parameters: NRR: a constant or distance dependent function for normalized reciprocal rate. When being a function, it should be accept vectorized input. Returns: A callable object that returns the probability value.

Source code in bmtool/connectors.py 
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
class NormalizedReciprocalRate(ProbabilityFunction):
    """Reciprocal connection probability given normalized reciprocal rate.
    Normalized reciprocal rate is defined as the ratio between the reciprocal
    connection probability and the connection probability for a randomly
    connected network where the two unidirectional connections between any pair
    of neurons are independent. NRR = pr / (p0 * p1)

    Parameters:
        NRR: a constant or distance dependent function for normalized reciprocal
            rate. When being a function, it should be accept vectorized input.
    Returns:
        A callable object that returns the probability value.
    """

    def __init__(self, NRR=1.0):
        self.NRR = NRR if callable(NRR) else lambda *x: NRR

    def probability(self, dist, p0, p1):
        """Allow numpy array input and return probability in numpy array"""
        return p0 * p1 * self.NRR(dist)

    def __call__(self, dist, p0, p1, *arg, **kwargs):
        """Return probability for single distance input"""
        return self.probability(dist, p0, p1)

    def decisions(self, dist, p0, p1, cond=None, rng=None):
        """Return bool array of decisions
        dist: distance (scalar or array). Will be ignored if NRR is constant.
        p0, p1: forward and backward probability (scalar or array)
        cond: A tuple (direction, array of outcomes) representing the condition.
            Conditional probability will be returned if specified. The condition
            event is determined by connection direction (0 for forward, or 1 for
            backward) and outcomes (bool array of whether connection exists).
        rng : numpy.random.Generator, optional
            Random number generator. If None, uses the default generator.
        """
        dist, p0, p1 = map(np.asarray, (dist, p0, p1))
        pr = np.empty(dist.shape)
        pr[:] = self.probability(dist, p0, p1)
        pr = np.clip(pr, a_min=np.fmax(p0 + p1 - 1.0, 0.0), a_max=np.fmin(p0, p1))
        if cond is not None:
            mask = np.asarray(cond[1])
            pr[mask] /= p1 if cond[0] else p0
            pr[~mask] = 0.0
        return decisions(pr, rng=rng)

__call__(dist, p0, p1, *arg, **kwargs)

Return probability for single distance input

Source code in bmtool/connectors.py 
384
385
386
def __call__(self, dist, p0, p1, *arg, **kwargs):
    """Return probability for single distance input"""
    return self.probability(dist, p0, p1)

decisions(dist, p0, p1, cond=None, rng=None)

Return bool array of decisions dist: distance (scalar or array). Will be ignored if NRR is constant. p0, p1: forward and backward probability (scalar or array) cond: A tuple (direction, array of outcomes) representing the condition. Conditional probability will be returned if specified. The condition event is determined by connection direction (0 for forward, or 1 for backward) and outcomes (bool array of whether connection exists). rng : numpy.random.Generator, optional Random number generator. If None, uses the default generator.

Source code in bmtool/connectors.py 
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
def decisions(self, dist, p0, p1, cond=None, rng=None):
    """Return bool array of decisions
    dist: distance (scalar or array). Will be ignored if NRR is constant.
    p0, p1: forward and backward probability (scalar or array)
    cond: A tuple (direction, array of outcomes) representing the condition.
        Conditional probability will be returned if specified. The condition
        event is determined by connection direction (0 for forward, or 1 for
        backward) and outcomes (bool array of whether connection exists).
    rng : numpy.random.Generator, optional
        Random number generator. If None, uses the default generator.
    """
    dist, p0, p1 = map(np.asarray, (dist, p0, p1))
    pr = np.empty(dist.shape)
    pr[:] = self.probability(dist, p0, p1)
    pr = np.clip(pr, a_min=np.fmax(p0 + p1 - 1.0, 0.0), a_max=np.fmin(p0, p1))
    if cond is not None:
        mask = np.asarray(cond[1])
        pr[mask] /= p1 if cond[0] else p0
        pr[~mask] = 0.0
    return decisions(pr, rng=rng)

probability(dist, p0, p1)

Allow numpy array input and return probability in numpy array

Source code in bmtool/connectors.py 
380
381
382
def probability(self, dist, p0, p1):
    """Allow numpy array input and return probability in numpy array"""
    return p0 * p1 * self.NRR(dist)

Connector Base Classes

bmtool.connectors.AbstractConnector

Bases: ABC

Abstract base class for connectors

Source code in bmtool/connectors.py 
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
class AbstractConnector(ABC):
    """Abstract base class for connectors"""

    @abstractmethod
    def setup_nodes(self, source=None, target=None):
        """After network nodes are added to the BMTK network. Pass in the
        Nodepool objects of source and target nodes using this method.
        Must run this before building connections."""
        return NotImplemented

    @abstractmethod
    def edge_params(self, **kwargs):
        """Create the arguments for BMTK add_edges() method including the
        `connection_rule` method."""
        return NotImplemented

    @staticmethod
    def constant_function(val):
        """Convert a constant to a constant function"""

        def constant(*arg):
            return val

        return constant

constant_function(val) staticmethod

Convert a constant to a constant function

Source code in bmtool/connectors.py 
427
428
429
430
431
432
433
434
@staticmethod
def constant_function(val):
    """Convert a constant to a constant function"""

    def constant(*arg):
        return val

    return constant

edge_params(**kwargs) abstractmethod

Create the arguments for BMTK add_edges() method including the connection_rule method.

Source code in bmtool/connectors.py 
421
422
423
424
425
@abstractmethod
def edge_params(self, **kwargs):
    """Create the arguments for BMTK add_edges() method including the
    `connection_rule` method."""
    return NotImplemented

setup_nodes(source=None, target=None) abstractmethod

After network nodes are added to the BMTK network. Pass in the Nodepool objects of source and target nodes using this method. Must run this before building connections.

Source code in bmtool/connectors.py 
414
415
416
417
418
419
@abstractmethod
def setup_nodes(self, source=None, target=None):
    """After network nodes are added to the BMTK network. Pass in the
    Nodepool objects of source and target nodes using this method.
    Must run this before building connections."""
    return NotImplemented

bmtool.connectors.is_same_pop(source, target, quick=False)

Check whether two NodePool objects direct to the same population

Source code in bmtool/connectors.py 
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
def is_same_pop(source, target, quick=False):
    """Check whether two NodePool objects direct to the same population"""
    if quick:
        # Quick check (compare filter conditions)
        same = (
            source.network_name == target.network_name
            and source._NodePool__properties == target._NodePool__properties
        )
    else:
        # Strict check (compare all nodes)
        same = (
            source.network_name == target.network_name
            and len(source) == len(target)
            and all([s.node_id == t.node_id for s, t in zip(source, target)])
        )
    return same

bmtool.connectors.Timer

Bases: object

Source code in bmtool/connectors.py 
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
class Timer(object):
    def __init__(self, unit="sec"):
        if unit == "ms":
            self.scale = 1e3
        elif unit == "us":
            self.scale = 1e6
        elif unit == "min":
            self.scale = 1 / 60
        else:
            self.scale = 1
            unit = "sec"
        self.unit = unit
        self.start()

    def start(self):
        self._start = time.perf_counter()

    def end(self):
        return (time.perf_counter() - self._start) * self.scale

    def report(self, msg="Run time"):
        print((msg + ": %.3f " + self.unit) % self.end(), flush=True)

bmtool.connectors.pr_2_rho(p0, p1, pr)

Calculate correlation coefficient rho given reciprocal probability pr

Source code in bmtool/connectors.py 
480
481
482
483
484
485
def pr_2_rho(p0, p1, pr):
    """Calculate correlation coefficient rho given reciprocal probability pr"""
    for p in (p0, p1):
        assert p > 0 and p < 1
    assert pr >= 0 and pr <= p0 and pr <= p1 and pr >= p0 + p1 - 1
    return (pr - p0 * p1) / (p0 * (1 - p0) * p1 * (1 - p1)) ** 0.5

bmtool.connectors.rho_2_pr(p0, p1, rho)

Calculate reciprocal probability pr given correlation coefficient rho

Source code in bmtool/connectors.py 
488
489
490
491
492
493
494
495
496
497
498
499
500
def rho_2_pr(p0, p1, rho):
    """Calculate reciprocal probability pr given correlation coefficient rho"""
    for p in (p0, p1):
        assert p > 0 and p < 1
    pr = p0 * p1 + rho * (p0 * (1 - p0) * p1 * (1 - p1)) ** 0.5
    if not (pr >= 0 and pr <= p0 and pr <= p1 and pr >= p0 + p1 - 1):
        pr0, pr = pr, np.max((0.0, p0 + p1 - 1, np.min((p0, p1, pr))))
        rho0, rho = rho, (pr - p0 * p1) / (p0 * (1 - p0) * p1 * (1 - p1)) ** 0.5
        print(
            "rho changed from %.3f to %.3f; pr changed from %.3f to %.3f" % (rho0, rho, pr0, pr),
            flush=True,
        )
    return pr

Connector Implementations

bmtool.connectors.ReciprocalConnector

Bases: AbstractConnector

Object for buiilding connections in bmtk network model with reciprocal probability within a single population (or between two populations).

Algorithm: Create random connection for every pair of cells independently, following a bivariate Bernoulli distribution. Each variable is 0 or 1, whether a connection exists in a forward or backward direction. There are four possible outcomes for each pair, no connection, unidirectional connection in two ways, and reciprocal connection. The probability of each outcome forms a contingency table. b a c k w a r d f --------------- o | | 0 | 1 | The total forward connection probability is r |---|-----|-----| p0 = p10 + p11 w | 0 | p00 | p01 | The total backward connection probability is a |---|-----|-----| p1 = p01 + p11 r | 1 | p10 | p11 | The reciprocal connection probability is d --------------- pr = p11 The distribution can be characterized by three parameters, p0, p1, pr. pr = p0 * p1 when two directions are independent. The correlation coefficient rho between the two has a relation with pr as follow. rho = (pr-p0p1) / (p0(1-p0)p1(1-p1))^(1/2) Generating random outcome consists of two steps. First draw random outcome for forward connection with probability p0. Then draw backward outcome following a conditional probability given the forward outcome, represented by p0, p1, and either pr or rho.

Use with BMTK: 1. Create this object with parameters.

    connector = ReciprocalConnector(**parameters)

2. After network nodes are added to the BMTK network. Pass in the
Nodepool objects of source and target nodes using setup_nodes() method.

    source = net.nodes(**source_filter)
    target = net.nodes(**target_filter)
    connector.setup_nodes(source, target)

3. Use edge_params() method to get the arguments for BMTK add_edges()
method including the `connection_rule` method.

    net.add_edges(**connector.edge_params(),
                  **other_source_to_target_edge_properties)

If the source and target are two different populations, do this again
for the backward connections (from target to source population).

    net.add_edges(**connector.edge_params(),
                  **other_target_to_source_edge_properties)

4. When executing net.build(), BMTK uses built-in `one_to_all` iterator
that calls the make_forward_connection() method to build connections
from source to target. If the two are different populations,
`all_to_one` iterator that calls the make_backward_connection() method
is then used to build connections from target to source.
During the initial iteration when make_forward_connection() is called,
the algorithm is run to generate a connection matrix for both forward
and backward connections. In the iterations afterward, it's only
assigning the generated connections in BMTK.

Parameters: p0, p1: Probability of forward and backward connection. It can be a constant or a deterministic function whose value must be within range [0, 1], otherwise incorrect value may occur in the algorithm. When p0, p1 are constant, the connection is homogenous. symmetric_p1: Whether p0 and p1 are identical. When the probabilities are equal for forward and backward connections, set this to True, Argument p1 will be ignored. This is forced to be True when the population is recurrent, i.e., the source and target are the same. This is forced to be False if symmetric_p1_arg is False. p0_arg, p1_arg: Input argument(s) for p0 and p1 function, e.g., p0(p0_arg). It can be a constant or a deterministic function whose input arguments are two node objects in BMTK, e.g., p0_arg(src_node,trg_node), p1_arg(trg_node,src_node). The latter has reversed order since it's for backward connection. They are usually distance between two nodes which is used for distance dependent connection probability, where the order does not matter. When p0 and p1 does not need inputs arguments, set p0_arg and p1_arg to None as so by default. Functions p0 and p1 need to accept one unused positional argument as placeholder, e.g., p0(args), so it does not raise an error when p0(None) is called. symmetric_p1_arg: Whether p0_arg and p1_arg are identical. If this is set to True, argument p1_arg will be ignored. This is forced to be True when the population is recurrent. pr, pr_arg: Probability of reciprocal connection and its first input argument when it is a function, similar to p0, p0_arg, p1, p1_arg. It can be a function when it has an explicit relation with some node properties such as distance. A function pr requires two additional positional arguments p0 and p1 even if they are not used, i.e., pr(pr_arg, p0, p1), just in case pr is dependent on p0 and p1, e.g., when normalized reciprocal rate NRR = pr/(p0p1) is given. When pr_arg is a string, the same value as p1_arg will be used for pr_arg if the string contains '1', e.g., '1', 'p1'. Otherwise, e.g., '', '0', 'p0', p0_arg will be used for pr_arg. Specifying this can avoid recomputing pr_arg when it's given by p0_arg or p1_arg. estimate_rho: Whether estimate rho that result in an overall pr. This is forced to be False if pr is a function or if rho is specified. To estimate rho, all the pairs with possible connections, meaning p0 and p1 are both non-zero for these pairs, are used to estimate a value of rho that will result in an expected number of reciprocal connections with the given pr. Note that pr is not over all pairs of source and target cells but only those has a chance to connect, e.g., for only pair of cells within some distance range. The estimation is done before generating random connections. The values of p0, p0_arg, p1, p1_arg can be cached during estimation of rho and retrieved when generating random connections for performance. dist_range_forward: If specified, when estimating rho, consider only cell pairs whose distance (p0_arg) is within the specified range. dist_range_backward: Similar to dist_range_forward but consider backward distance range (p1_arg) instead. If both are specified, consider only cell pairs whose both distances are within range. If neither is specified, infer valid pairs by non-zero connection probability. rho: The correlation coefficient rho. When specified, do not estimate it but instead use the given value throughout, pr will not be used. In cases where both p0 and p1 are simple functions, i.e., are constant on their support, e.g., function UniformInRange(), the estimation of rho will be equal to pr_2_rho(p0, p1, pr) where p0, p1 are non-zero. Estimation is not necessary. Directly set rho. n_syn0, n_syn1: Number of synapses in the forward and backward connection if connected. It can be a constant or a (deterministic or random) function whose input arguments are two node objects in BMTK like p0_arg, p1_arg. n_syn1 is force to be the same as n_syn0 when the population is recurrent. Warning: The number must not be greater than 255 since it will be converted to uint8 when written into the connection matrix to reduce memory consumption. autapses: Whether to allow connecting a cell to itself. Default: False. This is ignored when the population is not recurrent. quick_pop_check: Whether to use quick method to check if source and target populations are the same. Default: False. Quick method checks only whether filter conditions match. Strict method checks whether all node id's match considering order. cache_data: Whether to cache the values of p0, p0_arg, p1, p1_arg during estimation of rho. This improves performance when estimate_rho is True while not creating a significant overhead in the opposite case. However, it requires large memory allocation as the population size grows. Set it to False if there is a memory issue. rng: Random number generator (numpy.random.Generator) used for stochastic connection decisions in both forward and backward directions. Provide a seeded generator for reproducible connectivity. If None (default), fall back to the module-level default RNG returned by _get_default_rng(). save_report: Whether to save connection report to CSV file. report_name: Filename for connection report. If None, uses the module-level report_name. verbose: Whether show verbose information in console.

Returns: An object that works with BMTK to build edges in a network.

Important attributes: vars: Dictionary that stores part of the original input parameters. source, target: NodePool objects for the source and target populations. recurrent: Whether the source and target populations are the same. callable_set: Set of arguments that are functions but not constants. cache: ConnectorCache object for caching data. conn_mat: Connection matrix stage: Indicator of stage. 0 for forward and 1 for backward connection. conn_prop: List of two dictionaries that stores properties of connected pairs, for forward and backward connections respectively. In each dictionary, each key is the source node id and the value is a dictionary, where each key is the target node id that the source node connects to, and the value is the value of p0_arg or p1_arg. Example: [{sid0: {tid0: p0_arg0, tid1: p0_arg1, ...}, sid1: {...}, sid2: {...}, ... }, {tid2: {sid3: p1_arg0, sid4: p1_arg1, ...}, tid3: {...}, tid4: {...}, ... }] This is useful when properties of edges such as distance is used to determine other edge properties such as delay. So the distance does not need to be calculated repeatedly. The connector can be passed as an argument for the functions that generates additional edge properties, so that they can access the information here.

Source code in bmtool/connectors.py 
 503
 504
 505
 506
 507
 508
 509
 510
 511
 512
 513
 514
 515
 516
 517
 518
 519
 520
 521
 522
 523
 524
 525
 526
 527
 528
 529
 530
 531
 532
 533
 534
 535
 536
 537
 538
 539
 540
 541
 542
 543
 544
 545
 546
 547
 548
 549
 550
 551
 552
 553
 554
 555
 556
 557
 558
 559
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569
 570
 571
 572
 573
 574
 575
 576
 577
 578
 579
 580
 581
 582
 583
 584
 585
 586
 587
 588
 589
 590
 591
 592
 593
 594
 595
 596
 597
 598
 599
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
class ReciprocalConnector(AbstractConnector):
    """
    Object for buiilding connections in bmtk network model with reciprocal
    probability within a single population (or between two populations).

    Algorithm:
        Create random connection for every pair of cells independently,
        following a bivariate Bernoulli distribution. Each variable is 0 or 1,
        whether a connection exists in a forward or backward direction. There
        are four possible outcomes for each pair, no connection, unidirectional
        connection in two ways, and reciprocal connection. The probability of
        each outcome forms a contingency table.
            b a c k w a r d
        f   ---------------
        o  |   |  0  |  1  |  The total forward connection probability is
        r  |---|-----|-----|  p0 = p10 + p11
        w  | 0 | p00 | p01 |  The total backward connection probability is
        a  |---|-----|-----|  p1 = p01 + p11
        r  | 1 | p10 | p11 |  The reciprocal connection probability is
        d   ---------------   pr = p11
        The distribution can be characterized by three parameters, p0, p1, pr.
        pr = p0 * p1 when two directions are independent. The correlation
        coefficient rho between the two has a relation with pr as follow.
        rho = (pr-p0*p1) / (p0*(1-p0)*p1*(1-p1))^(1/2)
        Generating random outcome consists of two steps. First draw random
        outcome for forward connection with probability p0. Then draw backward
        outcome following a conditional probability given the forward outcome,
        represented by p0, p1, and either pr or rho.

    Use with BMTK:
        1. Create this object with parameters.

            connector = ReciprocalConnector(**parameters)

        2. After network nodes are added to the BMTK network. Pass in the
        Nodepool objects of source and target nodes using setup_nodes() method.

            source = net.nodes(**source_filter)
            target = net.nodes(**target_filter)
            connector.setup_nodes(source, target)

        3. Use edge_params() method to get the arguments for BMTK add_edges()
        method including the `connection_rule` method.

            net.add_edges(**connector.edge_params(),
                          **other_source_to_target_edge_properties)

        If the source and target are two different populations, do this again
        for the backward connections (from target to source population).

            net.add_edges(**connector.edge_params(),
                          **other_target_to_source_edge_properties)

        4. When executing net.build(), BMTK uses built-in `one_to_all` iterator
        that calls the make_forward_connection() method to build connections
        from source to target. If the two are different populations,
        `all_to_one` iterator that calls the make_backward_connection() method
        is then used to build connections from target to source.
        During the initial iteration when make_forward_connection() is called,
        the algorithm is run to generate a connection matrix for both forward
        and backward connections. In the iterations afterward, it's only
        assigning the generated connections in BMTK.

    Parameters:
        p0, p1: Probability of forward and backward connection. It can be a
            constant or a deterministic function whose value must be within
            range [0, 1], otherwise incorrect value may occur in the algorithm.
            When p0, p1 are constant, the connection is homogenous.
        symmetric_p1: Whether p0 and p1 are identical. When the probabilities
            are equal for forward and backward connections, set this to True,
            Argument p1 will be ignored. This is forced to be True when the
            population is recurrent, i.e., the source and target are the same.
            This is forced to be False if symmetric_p1_arg is False.
        p0_arg, p1_arg: Input argument(s) for p0 and p1 function, e.g.,
            p0(p0_arg). It can be a constant or a deterministic function whose
            input arguments are two node objects in BMTK, e.g.,
            p0_arg(src_node,trg_node), p1_arg(trg_node,src_node). The latter
            has reversed order since it's for backward connection. They are
            usually distance between two nodes which is used for distance
            dependent connection probability, where the order does not matter.
            When p0 and p1 does not need inputs arguments, set p0_arg and
            p1_arg to None as so by default. Functions p0 and p1 need to accept
            one unused positional argument as placeholder, e.g., p0(*args), so
            it does not raise an error when p0(None) is called.
        symmetric_p1_arg: Whether p0_arg and p1_arg are identical. If this is
            set to True, argument p1_arg will be ignored. This is forced to be
            True when the population is recurrent.
        pr, pr_arg: Probability of reciprocal connection and its first input
            argument when it is a function, similar to p0, p0_arg, p1, p1_arg.
            It can be a function when it has an explicit relation with some node
            properties such as distance. A function pr requires two additional
            positional arguments p0 and p1 even if they are not used, i.e.,
            pr(pr_arg, p0, p1), just in case pr is dependent on p0 and p1, e.g.,
            when normalized reciprocal rate NRR = pr/(p0*p1) is given.
            When pr_arg is a string, the same value as p1_arg will be used for
            pr_arg if the string contains '1', e.g., '1', 'p1'. Otherwise, e.g.,
            '', '0', 'p0', p0_arg will be used for pr_arg. Specifying this can
            avoid recomputing pr_arg when it's given by p0_arg or p1_arg.
        estimate_rho: Whether estimate rho that result in an overall pr. This
            is forced to be False if pr is a function or if rho is specified.
            To estimate rho, all the pairs with possible connections, meaning
            p0 and p1 are both non-zero for these pairs, are used to estimate
            a value of rho that will result in an expected number of reciprocal
            connections with the given pr. Note that pr is not over all pairs
            of source and target cells but only those has a chance to connect,
            e.g., for only pair of cells within some distance range. The
            estimation is done before generating random connections. The values
            of p0, p0_arg, p1, p1_arg can be cached during estimation of rho
            and retrieved when generating random connections for performance.
        dist_range_forward: If specified, when estimating rho, consider only
            cell pairs whose distance (p0_arg) is within the specified range.
        dist_range_backward: Similar to dist_range_forward but consider
            backward distance range (p1_arg) instead. If both are specified,
            consider only cell pairs whose both distances are within range. If
            neither is specified, infer valid pairs by non-zero connection
            probability.
        rho: The correlation coefficient rho. When specified, do not estimate
            it but instead use the given value throughout, pr will not be used.
            In cases where both p0 and p1 are simple functions, i.e., are
            constant on their support, e.g., function UniformInRange(), the
            estimation of rho will be equal to pr_2_rho(p0, p1, pr) where p0,
            p1 are non-zero. Estimation is not necessary. Directly set rho.
        n_syn0, n_syn1: Number of synapses in the forward and backward
            connection if connected. It can be a constant or a (deterministic
            or random) function whose input arguments are two node objects in
            BMTK like p0_arg, p1_arg. n_syn1 is force to be the same as n_syn0
            when the population is recurrent. Warning: The number must not be
            greater than 255 since it will be converted to uint8 when written
            into the connection matrix to reduce memory consumption.
        autapses: Whether to allow connecting a cell to itself. Default: False.
            This is ignored when the population is not recurrent.
        quick_pop_check: Whether to use quick method to check if source and
            target populations are the same. Default: False.
            Quick method checks only whether filter conditions match.
            Strict method checks whether all node id's match considering order.
        cache_data: Whether to cache the values of p0, p0_arg, p1, p1_arg
            during estimation of rho. This improves performance when
            estimate_rho is True while not creating a significant overhead in
            the opposite case. However, it requires large memory allocation
            as the population size grows. Set it to False if there is a memory
            issue.
        rng: Random number generator (`numpy.random.Generator`) used for
            stochastic connection decisions in both forward and backward
            directions. Provide a seeded generator for reproducible
            connectivity. If None (default), fall back to the module-level
            default RNG returned by _get_default_rng().
        save_report: Whether to save connection report to CSV file.
        report_name: Filename for connection report. If None, uses the
            module-level report_name.
        verbose: Whether show verbose information in console.

    Returns:
        An object that works with BMTK to build edges in a network.

    Important attributes:
        vars: Dictionary that stores part of the original input parameters.
        source, target: NodePool objects for the source and target populations.
        recurrent: Whether the source and target populations are the same.
        callable_set: Set of arguments that are functions but not constants.
        cache: ConnectorCache object for caching data.
        conn_mat: Connection matrix
        stage: Indicator of stage. 0 for forward and 1 for backward connection.
        conn_prop: List of two dictionaries that stores properties of connected
            pairs, for forward and backward connections respectively. In each
            dictionary, each key is the source node id and the value is a
            dictionary, where each key is the target node id that the source
            node connects to, and the value is the value of p0_arg or p1_arg.
            Example: [{sid0: {tid0: p0_arg0, tid1: p0_arg1, ...},
                       sid1: {...}, sid2: {...}, ... },
                      {tid2: {sid3: p1_arg0, sid4: p1_arg1, ...},
                       tid3: {...}, tid4: {...}, ... }]
            This is useful when properties of edges such as distance is used to
            determine other edge properties such as delay. So the distance does
            not need to be calculated repeatedly. The connector can be passed
            as an argument for the functions that generates additional edge
            properties, so that they can access the information here.
    """

    def __init__(
        self,
        p0=1.0,
        p1=1.0,
        symmetric_p1=False,
        p0_arg=None,
        p1_arg=None,
        symmetric_p1_arg=False,
        pr=0.0,
        pr_arg=None,
        estimate_rho=True,
        rho=None,
        dist_range_forward=None,
        dist_range_backward=None,
        n_syn0=1,
        n_syn1=1,
        autapses=False,
        quick_pop_check=False,
        cache_data=True,
        verbose=True,
        save_report=True,
        report_name=None,
        rng = None
    ):
        args = locals()
        var_set = ("p0", "p0_arg", "p1", "p1_arg", "pr", "pr_arg", "n_syn0", "n_syn1")
        self.vars = {key: args[key] for key in var_set}

        self.symmetric_p1 = symmetric_p1 and symmetric_p1_arg
        self.symmetric_p1_arg = symmetric_p1_arg

        self.estimate_rho = estimate_rho and not callable(pr) and rho is None
        self.dist_range_forward = dist_range_forward
        self.dist_range_backward = dist_range_backward
        self.rho = rho

        self.autapses = autapses
        self.quick = quick_pop_check
        self.cache = self.ConnectorCache(cache_data and self.estimate_rho)
        self.verbose = verbose
        self.save_report = save_report

        if report_name is None:
            report_name = globals().get("report_name", "default_report.csv")
        self.report_name = report_name

        self.conn_prop = [{}, {}]
        self.stage = 0
        self.iter_count = 0

        self.rng = _get_default_rng() if rng is None else rng

    # *** Two methods executed during bmtk edge creation net.add_edges() ***
    def setup_nodes(self, source=None, target=None):
        """Must run this before building connections"""
        if self.stage:
            # check whether the correct populations
            if (
                source is None
                or target is None
                or not is_same_pop(source, self.target, quick=self.quick)
                or not is_same_pop(target, self.source, quick=self.quick)
            ):
                raise ValueError("Source or target population not consistent.")
            # Skip adding nodes for the backward stage.
            return

        # Update node pools
        self.source = source
        self.target = target
        if self.source is None or len(self.source) == 0:
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(f"{src_str} nodes do not exists")
        if self.target is None or len(self.target) == 0:
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(f"{trg_str} nodes do not exists")

        # Setup nodes
        self.recurrent = is_same_pop(self.source, self.target, quick=self.quick)
        self.source_ids = [s.node_id for s in self.source]
        self.n_source = len(self.source_ids)
        self.source_list = list(self.source)
        if self.recurrent:
            self.target_ids = self.source_ids
            self.n_target = self.n_source
            self.target_list = self.source_list
        else:
            self.target_ids = [t.node_id for t in self.target]
            self.n_target = len(self.target_ids)
            self.target_list = list(self.target)

        # Setup for recurrent connection
        if self.recurrent:
            self.symmetric_p1_arg = True
            self.symmetric_p1 = True
            self.vars["n_syn1"] = self.vars["n_syn0"]
        if self.symmetric_p1_arg:
            self.vars["p1_arg"] = self.vars["p0_arg"]
        if self.symmetric_p1:
            self.vars["p1"] = self.vars["p0"]

    def edge_params(self):
        """Create the arguments for BMTK add_edges() method"""
        if self.stage == 0:
            params = {
                "source": self.source,
                "target": self.target,
                "iterator": "one_to_all",
                "connection_rule": self.make_forward_connection,
            }
        else:
            params = {
                "source": self.target,
                "target": self.source,
                "iterator": "all_to_one",
                "connection_rule": self.make_backward_connection,
            }
        self.stage += 1
        return params

    # *** Methods executed during bmtk network.build() ***
    # *** Helper functions ***
    class ConnectorCache(object):
        def __init__(self, enable=True):
            self.enable = enable
            self._output = {}
            self.cache_dict = {}
            self.set_next_it()
            self.write_mode()

        def cache_output(self, func, func_name, cache=True):
            if self.enable and cache:
                self.cache_dict[func_name] = func
                self._output[func_name] = []
                output = self._output[func_name]

                def writer(*args):
                    val = func(*args)
                    output.append(val)
                    return val

                setattr(self, func_name, writer)
            else:
                setattr(self, func_name, func)

        def write_mode(self):
            for val in self._output.values():
                val.clear()
            self.mode = "write"
            self.iter_count = 0

        def fetch_output(self, func_name, fetch=True):
            output = self._output[func_name]

            if fetch:

                def reader(*args):
                    return output[self.iter_count]

                setattr(self, func_name, reader)
            else:
                setattr(self, func_name, self.cache_dict[func_name])

        def read_mode(self):
            if self.enable and len(self.cache_dict):
                # check whether outputs were written correctly
                output_len = [len(val) for val in self._output.values()]
                # whether any stored and have the same length
                valid = [n for n in output_len if n]
                flag = len(valid) > 0 and all(n == valid[0] for n in valid[1:])
                if flag:
                    for func_name, out_len in zip(self._output, output_len):
                        fetch = out_len > 0
                        if not fetch:
                            print(
                                "\nWarning: Cache did not work properly for " + func_name + "\n",
                                flush=True,
                            )
                        self.fetch_output(func_name, fetch)
                    self.iter_count = 0
                else:
                    # if output not correct, disable and use original function
                    print("\nWarning: Cache did not work properly.\n", flush=True)
                    for func_name in self.cache_dict:
                        self.fetch_output(func_name, False)
                    self.enable = False
            self.mode = "read"

        def set_next_it(self):
            if self.enable:

                def next_it():
                    self.iter_count += 1
            else:

                def next_it():
                    pass

            self.next_it = next_it

    def node_2_idx_input(self, var_func, reverse=False):
        """Convert a function that accept nodes as input
        to accept indices as input"""
        if reverse:

            def idx_2_var(j, i):
                return var_func(self.target_list[j], self.source_list[i])
        else:

            def idx_2_var(i, j):
                return var_func(self.source_list[i], self.target_list[j])

        return idx_2_var

    def iterate_pairs(self):
        """Generate indices of source and target for each case"""
        if self.recurrent:
            if self.autapses:
                for i in range(self.n_source):
                    for j in range(i, self.n_target):
                        yield i, j
            else:
                for i in range(self.n_source - 1):
                    for j in range(i + 1, self.n_target):
                        yield i, j
        else:
            for i in range(self.n_source):
                for j in range(self.n_target):
                    yield i, j

    def calc_pair(self, i, j):
        """Calculate intermediate data that can be cached"""
        cache = self.cache
        # cache = self  # test performance for not using cache
        p0_arg = cache.p0_arg(i, j)
        p1_arg = p0_arg if self.symmetric_p1_arg else cache.p1_arg(j, i)
        p0 = cache.p0(p0_arg)
        p1 = p0 if self.symmetric_p1 else cache.p1(p1_arg)
        return p0_arg, p1_arg, p0, p1

    def setup_conditional_backward_probability(self):
        """Create a function that calculates the conditional probability of
        backward connection given the forward connection outcome 'cond'"""
        # For all cases, assume p0, p1, pr are all within [0, 1] already.
        self.wrong_pr = False
        if self.rho is None:
            # Determine by pr for each pair
            if self.verbose:

                def cond_backward(cond, p0, p1, pr):
                    if p0 > 0:
                        pr_bound = (p0 + p1 - 1, min(p0, p1))
                        # check whether pr within bounds
                        if pr < pr_bound[0] or pr > pr_bound[1]:
                            self.wrong_pr = True
                            pr = min(max(pr, pr_bound[0]), pr_bound[1])
                        return pr / p0 if cond else (p1 - pr) / (1 - p0)
                    else:
                        return p1
            else:

                def cond_backward(cond, p0, p1, pr):
                    if p0 > 0:
                        pr_bound = (p0 + p1 - 1, min(p0, p1))
                        pr = min(max(pr, pr_bound[0]), pr_bound[1])
                        return pr / p0 if cond else (p1 - pr) / (1 - p0)
                    else:
                        return p1
        elif self.rho == 0:
            # Independent case
            def cond_backward(cond, p0, p1, pr):
                return p1
        else:
            # Dependent with fixed correlation coefficient rho
            def cond_backward(cond, p0, p1, pr):
                # Standard deviation of r.v. for p1
                sd = ((1 - p1) * p1) ** 0.5
                # Z-score of random variable for p0
                zs = ((1 - p0) / p0) ** 0.5 if cond else -((p0 / (1 - p0)) ** 0.5)
                return p1 + self.rho * sd * zs

        self.cond_backward = cond_backward

    def add_conn_prop(self, src, trg, prop, stage=0):
        """Store p0_arg and p1_arg for a connected pair"""
        sid = self.source_ids[src]
        tid = self.target_ids[trg]
        conn_dict = self.conn_prop[stage]
        if stage:
            sid, tid = tid, sid  # during backward, from target to source
        trg_dict = conn_dict.setdefault(sid, {})
        trg_dict[tid] = prop

    def get_conn_prop(self, sid, tid):
        """Get stored value given node ids in a connection"""
        return self.conn_prop[self.stage][sid][tid]

    # *** A sequence of major methods executed during build ***
    def setup_variables(self):
        # If pr_arg is string, use the same value as p0_arg or p1_arg
        if isinstance(self.vars["pr_arg"], str):
            pr_arg_func = "p1_arg" if "1" in self.vars["pr_arg"] else "p0_arg"
            self.vars["pr_arg"] = self.vars[pr_arg_func]
        else:
            pr_arg_func = None

        callable_set = set()
        # Make constant variables constant functions
        for name, var in self.vars.items():
            if callable(var):
                callable_set.add(name)  # record callable variables
                setattr(self, name, var)
            else:
                setattr(self, name, self.constant_function(var))
        self.callable_set = callable_set

        # Make callable variables except a few, accept index input instead
        for name in callable_set - {"p0", "p1", "pr"}:
            var = self.vars[name]
            setattr(self, name, self.node_2_idx_input(var, "1" in name))

        # Set up function for pr_arg if use value from p0_arg or p1_arg
        if pr_arg_func is None:
            self._pr_arg = self.pr_arg  # use specified pr_arg
        else:
            self._pr_arg_val = 0.0  # storing current value from p_arg
            p_arg = getattr(self, pr_arg_func)

            def p_arg_4_pr(*args, **kwargs):
                val = p_arg(*args, **kwargs)
                self._pr_arg_val = val
                return val

            setattr(self, pr_arg_func, p_arg_4_pr)

            def pr_arg(self, *arg):
                return self._pr_arg_val

            self._pr_arg = types.MethodType(pr_arg, self)

    def cache_variables(self):
        # Select cacheable attrilbutes
        cache_set = {"p0", "p0_arg", "p1", "p1_arg"}
        if self.symmetric_p1:
            cache_set.remove("p1")
        if self.symmetric_p1_arg:
            cache_set.remove("p1_arg")
        # Output of callable variables will be cached
        # Constant functions will be called from cache but output not cached
        for name in cache_set:
            var = getattr(self, name)
            self.cache.cache_output(var, name, name in self.callable_set)
        if self.verbose and len(self.cache.cache_dict):
            print("Output of %s will be cached." % ", ".join(self.cache.cache_dict), flush=True)

    def setup_dist_range_checker(self):
        # Checker that determines whether to consider a pair for rho estimation
        if self.dist_range_forward is None and self.dist_range_backward is None:

            def checker(var):
                p0, p1 = var[2:]
                return p0 > 0 and p1 > 0
        else:

            def in_range(p_arg, dist_range):
                return p_arg >= dist_range[0] and p_arg <= dist_range[1]

            r0, r1 = self.dist_range_forward, self.dist_range_backward
            if r1 is None:

                def checker(var):
                    return in_range(var[0], r0)
            elif r0 is None:

                def checker(var):
                    return in_range(var[1], r1)
            else:

                def checker(var):
                    return in_range(var[0], r0) and in_range(var[1], r1)

        return checker

    def initialize(self):
        self.setup_variables()
        self.cache_variables()
        # Intialize connection matrix and get nubmer of pairs
        self.end_stage = 0 if self.recurrent else 1
        shape = (self.end_stage + 1, self.n_source, self.n_target)
        self.conn_mat = np.zeros(shape, dtype=np.uint8)  # 1 byte per entry

    def initial_all_to_all(self):
        """The major part of the algorithm run at beginning of BMTK iterator"""
        if self.verbose:
            src_str, trg_str = self.get_nodes_info()
            print(
                "\nStart building connection between: \n  " + src_str + "\n  " + trg_str, flush=True
            )
        self.initialize()
        cache = self.cache  # write mode

        # Estimate pr
        if self.verbose:
            self.timer = Timer()
        if self.estimate_rho:
            dist_range_checker = self.setup_dist_range_checker()
            p0p1_sum = 0.0
            norm_fac_sum = 0.0
            n = 0
            # Make sure each cacheable function runs excatly once per iteration
            for i, j in self.iterate_pairs():
                var = self.calc_pair(i, j)
                valid = dist_range_checker(var)
                if valid:
                    n += 1
                    p0, p1 = var[2:]
                    p0p1_sum += p0 * p1
                    norm_fac_sum += (p0 * (1 - p0) * p1 * (1 - p1)) ** 0.5
            if norm_fac_sum > 0:
                rho = (self.pr() * n - p0p1_sum) / norm_fac_sum
                if abs(rho) > 1:
                    print(
                        "\nWarning: Estimated value of rho=%.3f "
                        "outside the range [-1, 1]." % rho,
                        flush=True,
                    )
                    rho = np.clip(rho, -1, 1).item()
                    print("Force rho to be %.0f.\n" % rho, flush=True)
                elif self.verbose:
                    print("Estimated value of rho=%.3f" % rho, flush=True)
                self.rho = rho
            else:
                self.rho = 0

            if self.verbose:
                self.timer.report("Time for estimating rho")

        # Setup function for calculating conditional backward probability
        self.setup_conditional_backward_probability()

        # Make random connections
        cache.read_mode()
        possible_count = 0 if self.recurrent else np.zeros(3)
        for i, j in self.iterate_pairs():
            p0_arg, p1_arg, p0, p1 = self.calc_pair(i, j)
            # Check whether at all possible and count
            forward = p0 > 0
            backward = p1 > 0
            if self.recurrent:
                possible_count += forward
            else:
                possible_count += [forward, backward, forward and backward]

            # Make random decision
            if forward:
                forward = decision(p0, rng=self.rng)
            if backward:
                pr = self.pr(self._pr_arg(i, j), p0, p1)
                backward = decision(self.cond_backward(forward, p0, p1, pr), rng=self.rng)

            # Make connection
            if forward:
                n_forward = self.n_syn0(i, j)
                self.add_conn_prop(i, j, p0_arg, 0)
                self.conn_mat[0, i, j] = n_forward
            if backward:
                n_backward = self.n_syn1(j, i)
                if self.recurrent:
                    if i != j:
                        self.conn_mat[0, j, i] = n_backward
                        self.add_conn_prop(j, i, p1_arg, 0)
                else:
                    self.conn_mat[1, i, j] = n_backward
                    self.add_conn_prop(i, j, p1_arg, 1)
            self.cache.next_it()
        self.cache.write_mode()  # clear memory
        self.possible_count = possible_count

        if self.verbose:
            self.timer.report("Total time for creating connection matrix")
            if self.wrong_pr:
                print("Warning: Value of 'pr' outside the bounds occurred.\n", flush=True)
            self.connection_number_info()
        if self.save_report:
            self.save_connection_report()

    def make_connection(self):
        """Assign number of synapses per iteration.
        Use iterator one_to_all for forward and all_to_one for backward.
        """
        nsyns = self.conn_mat[self.stage, self.iter_count, :]
        self.iter_count += 1

        # Detect end of iteration
        if self.iter_count == self.n_source:
            self.iter_count = 0
            if self.stage == self.end_stage:
                if self.verbose:
                    self.timer.report("Done! \nTime for building connections")
                self.free_memory()
        return nsyns

    def make_forward_connection(self, source, targets, *args, **kwargs):
        """Function to be called by BMTK iterator for forward connection"""
        # Initialize in the first iteration
        if self.iter_count == 0:
            self.stage = 0
            self.initial_all_to_all()
            if self.verbose:
                print("Assigning forward connections.", flush=True)
                self.timer.start()
        return self.make_connection()

    def make_backward_connection(self, targets, source, *args, **kwargs):
        """Function to be called by BMTK iterator for backward connection"""
        if self.iter_count == 0:
            self.stage = 1
            if self.verbose:
                print("Assigning backward connections.", flush=True)
        return self.make_connection()

    def free_memory(self):
        """Free up memory after connections are built"""
        # Do not clear self.conn_prop if it will be used by conn.add_properties
        variables = ("conn_mat", "source_list", "target_list", "source_ids", "target_ids")
        for var in variables:
            setattr(self, var, None)

    # *** Helper functions for verbose ***
    def get_nodes_info(self):
        """Get strings with source and target population information"""
        source_str = self.source.network_name + ": " + self.source.filter_str
        target_str = self.target.network_name + ": " + self.target.filter_str
        return source_str, target_str

    def connection_number(self):
        """
        Return the number of the following:
        n_conn: connected pairs [forward, (backward,) reciprocal]
        n_poss: possible connections (prob>0) [forward, (backward, reciprocal)]
        n_pair: pairs of cells
        proportion: of connections in possible and total pairs
        """
        conn_mat = self.conn_mat.astype(bool)
        n_conn = np.count_nonzero(conn_mat, axis=(1, 2))
        n_poss = np.array(self.possible_count)
        n_pair = conn_mat.size / 2
        if self.recurrent:
            n_recp = np.count_nonzero(conn_mat[0] & conn_mat[0].T)
            if self.autapses:
                n_recp -= np.count_nonzero(np.diag(conn_mat[0]))
            n_recp //= 2
            n_conn -= n_recp
            n_poss = n_poss[None]
            n_pair += (1 if self.autapses else -1) * self.n_source / 2
        else:
            n_recp = np.count_nonzero(conn_mat[0] & conn_mat[1])
        n_conn = np.append(n_conn, n_recp)
        n_pair = int(n_pair)
        fraction = np.array([n_conn / n_poss, n_conn / n_pair])
        fraction[np.isnan(fraction)] = 0.0
        return n_conn, n_poss, n_pair, fraction

    def connection_number_info(self):
        """Print connection numbers after connections built"""

        def arr2str(a, f):
            return ", ".join([f] * a.size) % tuple(a.tolist())

        n_conn, n_poss, n_pair, fraction = self.connection_number()
        conn_type = "(all, reciprocal)" if self.recurrent else "(forward, backward, reciprocal)"
        print("Numbers of " + conn_type + " connections:", flush=True)
        print("Number of connected pairs: (%s)" % arr2str(n_conn, "%d"), flush=True)
        print("Number of possible connections: (%s)" % arr2str(n_poss, "%d"), flush=True)
        print(
            "Fraction of connected pairs in possible ones: (%s)"
            % arr2str(100 * fraction[0], "%.2f%%"),
            flush=True,
        )
        print("Number of total pairs: %d" % n_pair, flush=True)
        print(
            "Fraction of connected pairs in all pairs: (%s)\n"
            % arr2str(100 * fraction[1], "%.2f%%"),
            flush=True,
        )

    def save_connection_report(self):
        """Save connections into a CSV file to be read from later"""
        src_str, trg_str = self.get_nodes_info()
        n_conn, n_poss, n_pair, fraction = self.connection_number()

        # Extract the population name from source_str and target_str
        data = {
            "Source": [src_str],
            "Target": [trg_str],
            "Percent connectionivity within possible connections": [fraction[0] * 100],
            "Percent connectionivity within all connections": [fraction[1] * 100],
        }
        df = pd.DataFrame(data)

        # Append the data to the CSV file
        try:
            # Check if the file exists by trying to read it
            existing_df = pd.read_csv(self.report_name)
            # If no exception is raised, append without header
            df.to_csv(self.report_name, mode="a", header=False, index=False)
        except FileNotFoundError:
            # If the file does not exist, write with header
            df.to_csv(self.report_name, mode="w", header=True, index=False)

add_conn_prop(src, trg, prop, stage=0)

Store p0_arg and p1_arg for a connected pair

Source code in bmtool/connectors.py 
964
965
966
967
968
969
970
971
972
def add_conn_prop(self, src, trg, prop, stage=0):
    """Store p0_arg and p1_arg for a connected pair"""
    sid = self.source_ids[src]
    tid = self.target_ids[trg]
    conn_dict = self.conn_prop[stage]
    if stage:
        sid, tid = tid, sid  # during backward, from target to source
    trg_dict = conn_dict.setdefault(sid, {})
    trg_dict[tid] = prop

calc_pair(i, j)

Calculate intermediate data that can be cached

Source code in bmtool/connectors.py 
911
912
913
914
915
916
917
918
919
def calc_pair(self, i, j):
    """Calculate intermediate data that can be cached"""
    cache = self.cache
    # cache = self  # test performance for not using cache
    p0_arg = cache.p0_arg(i, j)
    p1_arg = p0_arg if self.symmetric_p1_arg else cache.p1_arg(j, i)
    p0 = cache.p0(p0_arg)
    p1 = p0 if self.symmetric_p1 else cache.p1(p1_arg)
    return p0_arg, p1_arg, p0, p1

connection_number()

Return the number of the following: n_conn: connected pairs [forward, (backward,) reciprocal] n_poss: possible connections (prob>0) [forward, (backward, reciprocal)] n_pair: pairs of cells proportion: of connections in possible and total pairs

Source code in bmtool/connectors.py 
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
def connection_number(self):
    """
    Return the number of the following:
    n_conn: connected pairs [forward, (backward,) reciprocal]
    n_poss: possible connections (prob>0) [forward, (backward, reciprocal)]
    n_pair: pairs of cells
    proportion: of connections in possible and total pairs
    """
    conn_mat = self.conn_mat.astype(bool)
    n_conn = np.count_nonzero(conn_mat, axis=(1, 2))
    n_poss = np.array(self.possible_count)
    n_pair = conn_mat.size / 2
    if self.recurrent:
        n_recp = np.count_nonzero(conn_mat[0] & conn_mat[0].T)
        if self.autapses:
            n_recp -= np.count_nonzero(np.diag(conn_mat[0]))
        n_recp //= 2
        n_conn -= n_recp
        n_poss = n_poss[None]
        n_pair += (1 if self.autapses else -1) * self.n_source / 2
    else:
        n_recp = np.count_nonzero(conn_mat[0] & conn_mat[1])
    n_conn = np.append(n_conn, n_recp)
    n_pair = int(n_pair)
    fraction = np.array([n_conn / n_poss, n_conn / n_pair])
    fraction[np.isnan(fraction)] = 0.0
    return n_conn, n_poss, n_pair, fraction

connection_number_info()

Print connection numbers after connections built

Source code in bmtool/connectors.py 
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
def connection_number_info(self):
    """Print connection numbers after connections built"""

    def arr2str(a, f):
        return ", ".join([f] * a.size) % tuple(a.tolist())

    n_conn, n_poss, n_pair, fraction = self.connection_number()
    conn_type = "(all, reciprocal)" if self.recurrent else "(forward, backward, reciprocal)"
    print("Numbers of " + conn_type + " connections:", flush=True)
    print("Number of connected pairs: (%s)" % arr2str(n_conn, "%d"), flush=True)
    print("Number of possible connections: (%s)" % arr2str(n_poss, "%d"), flush=True)
    print(
        "Fraction of connected pairs in possible ones: (%s)"
        % arr2str(100 * fraction[0], "%.2f%%"),
        flush=True,
    )
    print("Number of total pairs: %d" % n_pair, flush=True)
    print(
        "Fraction of connected pairs in all pairs: (%s)\n"
        % arr2str(100 * fraction[1], "%.2f%%"),
        flush=True,
    )

edge_params()

Create the arguments for BMTK add_edges() method

Source code in bmtool/connectors.py 
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
def edge_params(self):
    """Create the arguments for BMTK add_edges() method"""
    if self.stage == 0:
        params = {
            "source": self.source,
            "target": self.target,
            "iterator": "one_to_all",
            "connection_rule": self.make_forward_connection,
        }
    else:
        params = {
            "source": self.target,
            "target": self.source,
            "iterator": "all_to_one",
            "connection_rule": self.make_backward_connection,
        }
    self.stage += 1
    return params

free_memory()

Free up memory after connections are built

Source code in bmtool/connectors.py 
1202
1203
1204
1205
1206
1207
def free_memory(self):
    """Free up memory after connections are built"""
    # Do not clear self.conn_prop if it will be used by conn.add_properties
    variables = ("conn_mat", "source_list", "target_list", "source_ids", "target_ids")
    for var in variables:
        setattr(self, var, None)

get_conn_prop(sid, tid)

Get stored value given node ids in a connection

Source code in bmtool/connectors.py 
974
975
976
def get_conn_prop(self, sid, tid):
    """Get stored value given node ids in a connection"""
    return self.conn_prop[self.stage][sid][tid]

get_nodes_info()

Get strings with source and target population information

Source code in bmtool/connectors.py 
1210
1211
1212
1213
1214
def get_nodes_info(self):
    """Get strings with source and target population information"""
    source_str = self.source.network_name + ": " + self.source.filter_str
    target_str = self.target.network_name + ": " + self.target.filter_str
    return source_str, target_str

initial_all_to_all()

The major part of the algorithm run at beginning of BMTK iterator

Source code in bmtool/connectors.py 
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
def initial_all_to_all(self):
    """The major part of the algorithm run at beginning of BMTK iterator"""
    if self.verbose:
        src_str, trg_str = self.get_nodes_info()
        print(
            "\nStart building connection between: \n  " + src_str + "\n  " + trg_str, flush=True
        )
    self.initialize()
    cache = self.cache  # write mode

    # Estimate pr
    if self.verbose:
        self.timer = Timer()
    if self.estimate_rho:
        dist_range_checker = self.setup_dist_range_checker()
        p0p1_sum = 0.0
        norm_fac_sum = 0.0
        n = 0
        # Make sure each cacheable function runs excatly once per iteration
        for i, j in self.iterate_pairs():
            var = self.calc_pair(i, j)
            valid = dist_range_checker(var)
            if valid:
                n += 1
                p0, p1 = var[2:]
                p0p1_sum += p0 * p1
                norm_fac_sum += (p0 * (1 - p0) * p1 * (1 - p1)) ** 0.5
        if norm_fac_sum > 0:
            rho = (self.pr() * n - p0p1_sum) / norm_fac_sum
            if abs(rho) > 1:
                print(
                    "\nWarning: Estimated value of rho=%.3f "
                    "outside the range [-1, 1]." % rho,
                    flush=True,
                )
                rho = np.clip(rho, -1, 1).item()
                print("Force rho to be %.0f.\n" % rho, flush=True)
            elif self.verbose:
                print("Estimated value of rho=%.3f" % rho, flush=True)
            self.rho = rho
        else:
            self.rho = 0

        if self.verbose:
            self.timer.report("Time for estimating rho")

    # Setup function for calculating conditional backward probability
    self.setup_conditional_backward_probability()

    # Make random connections
    cache.read_mode()
    possible_count = 0 if self.recurrent else np.zeros(3)
    for i, j in self.iterate_pairs():
        p0_arg, p1_arg, p0, p1 = self.calc_pair(i, j)
        # Check whether at all possible and count
        forward = p0 > 0
        backward = p1 > 0
        if self.recurrent:
            possible_count += forward
        else:
            possible_count += [forward, backward, forward and backward]

        # Make random decision
        if forward:
            forward = decision(p0, rng=self.rng)
        if backward:
            pr = self.pr(self._pr_arg(i, j), p0, p1)
            backward = decision(self.cond_backward(forward, p0, p1, pr), rng=self.rng)

        # Make connection
        if forward:
            n_forward = self.n_syn0(i, j)
            self.add_conn_prop(i, j, p0_arg, 0)
            self.conn_mat[0, i, j] = n_forward
        if backward:
            n_backward = self.n_syn1(j, i)
            if self.recurrent:
                if i != j:
                    self.conn_mat[0, j, i] = n_backward
                    self.add_conn_prop(j, i, p1_arg, 0)
            else:
                self.conn_mat[1, i, j] = n_backward
                self.add_conn_prop(i, j, p1_arg, 1)
        self.cache.next_it()
    self.cache.write_mode()  # clear memory
    self.possible_count = possible_count

    if self.verbose:
        self.timer.report("Total time for creating connection matrix")
        if self.wrong_pr:
            print("Warning: Value of 'pr' outside the bounds occurred.\n", flush=True)
        self.connection_number_info()
    if self.save_report:
        self.save_connection_report()

iterate_pairs()

Generate indices of source and target for each case

Source code in bmtool/connectors.py 
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
def iterate_pairs(self):
    """Generate indices of source and target for each case"""
    if self.recurrent:
        if self.autapses:
            for i in range(self.n_source):
                for j in range(i, self.n_target):
                    yield i, j
        else:
            for i in range(self.n_source - 1):
                for j in range(i + 1, self.n_target):
                    yield i, j
    else:
        for i in range(self.n_source):
            for j in range(self.n_target):
                yield i, j

make_backward_connection(targets, source, *args, **kwargs)

Function to be called by BMTK iterator for backward connection

Source code in bmtool/connectors.py 
1194
1195
1196
1197
1198
1199
1200
def make_backward_connection(self, targets, source, *args, **kwargs):
    """Function to be called by BMTK iterator for backward connection"""
    if self.iter_count == 0:
        self.stage = 1
        if self.verbose:
            print("Assigning backward connections.", flush=True)
    return self.make_connection()

make_connection()

Assign number of synapses per iteration. Use iterator one_to_all for forward and all_to_one for backward.

Source code in bmtool/connectors.py 
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
def make_connection(self):
    """Assign number of synapses per iteration.
    Use iterator one_to_all for forward and all_to_one for backward.
    """
    nsyns = self.conn_mat[self.stage, self.iter_count, :]
    self.iter_count += 1

    # Detect end of iteration
    if self.iter_count == self.n_source:
        self.iter_count = 0
        if self.stage == self.end_stage:
            if self.verbose:
                self.timer.report("Done! \nTime for building connections")
            self.free_memory()
    return nsyns

make_forward_connection(source, targets, *args, **kwargs)

Function to be called by BMTK iterator for forward connection

Source code in bmtool/connectors.py 
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
def make_forward_connection(self, source, targets, *args, **kwargs):
    """Function to be called by BMTK iterator for forward connection"""
    # Initialize in the first iteration
    if self.iter_count == 0:
        self.stage = 0
        self.initial_all_to_all()
        if self.verbose:
            print("Assigning forward connections.", flush=True)
            self.timer.start()
    return self.make_connection()

node_2_idx_input(var_func, reverse=False)

Convert a function that accept nodes as input to accept indices as input

Source code in bmtool/connectors.py 
881
882
883
884
885
886
887
888
889
890
891
892
893
def node_2_idx_input(self, var_func, reverse=False):
    """Convert a function that accept nodes as input
    to accept indices as input"""
    if reverse:

        def idx_2_var(j, i):
            return var_func(self.target_list[j], self.source_list[i])
    else:

        def idx_2_var(i, j):
            return var_func(self.source_list[i], self.target_list[j])

    return idx_2_var

save_connection_report()

Save connections into a CSV file to be read from later

Source code in bmtool/connectors.py 
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
def save_connection_report(self):
    """Save connections into a CSV file to be read from later"""
    src_str, trg_str = self.get_nodes_info()
    n_conn, n_poss, n_pair, fraction = self.connection_number()

    # Extract the population name from source_str and target_str
    data = {
        "Source": [src_str],
        "Target": [trg_str],
        "Percent connectionivity within possible connections": [fraction[0] * 100],
        "Percent connectionivity within all connections": [fraction[1] * 100],
    }
    df = pd.DataFrame(data)

    # Append the data to the CSV file
    try:
        # Check if the file exists by trying to read it
        existing_df = pd.read_csv(self.report_name)
        # If no exception is raised, append without header
        df.to_csv(self.report_name, mode="a", header=False, index=False)
    except FileNotFoundError:
        # If the file does not exist, write with header
        df.to_csv(self.report_name, mode="w", header=True, index=False)

setup_conditional_backward_probability()

Create a function that calculates the conditional probability of backward connection given the forward connection outcome 'cond'

Source code in bmtool/connectors.py 
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
def setup_conditional_backward_probability(self):
    """Create a function that calculates the conditional probability of
    backward connection given the forward connection outcome 'cond'"""
    # For all cases, assume p0, p1, pr are all within [0, 1] already.
    self.wrong_pr = False
    if self.rho is None:
        # Determine by pr for each pair
        if self.verbose:

            def cond_backward(cond, p0, p1, pr):
                if p0 > 0:
                    pr_bound = (p0 + p1 - 1, min(p0, p1))
                    # check whether pr within bounds
                    if pr < pr_bound[0] or pr > pr_bound[1]:
                        self.wrong_pr = True
                        pr = min(max(pr, pr_bound[0]), pr_bound[1])
                    return pr / p0 if cond else (p1 - pr) / (1 - p0)
                else:
                    return p1
        else:

            def cond_backward(cond, p0, p1, pr):
                if p0 > 0:
                    pr_bound = (p0 + p1 - 1, min(p0, p1))
                    pr = min(max(pr, pr_bound[0]), pr_bound[1])
                    return pr / p0 if cond else (p1 - pr) / (1 - p0)
                else:
                    return p1
    elif self.rho == 0:
        # Independent case
        def cond_backward(cond, p0, p1, pr):
            return p1
    else:
        # Dependent with fixed correlation coefficient rho
        def cond_backward(cond, p0, p1, pr):
            # Standard deviation of r.v. for p1
            sd = ((1 - p1) * p1) ** 0.5
            # Z-score of random variable for p0
            zs = ((1 - p0) / p0) ** 0.5 if cond else -((p0 / (1 - p0)) ** 0.5)
            return p1 + self.rho * sd * zs

    self.cond_backward = cond_backward

setup_nodes(source=None, target=None)

Must run this before building connections

Source code in bmtool/connectors.py 
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
def setup_nodes(self, source=None, target=None):
    """Must run this before building connections"""
    if self.stage:
        # check whether the correct populations
        if (
            source is None
            or target is None
            or not is_same_pop(source, self.target, quick=self.quick)
            or not is_same_pop(target, self.source, quick=self.quick)
        ):
            raise ValueError("Source or target population not consistent.")
        # Skip adding nodes for the backward stage.
        return

    # Update node pools
    self.source = source
    self.target = target
    if self.source is None or len(self.source) == 0:
        src_str, trg_str = self.get_nodes_info()
        raise ValueError(f"{src_str} nodes do not exists")
    if self.target is None or len(self.target) == 0:
        src_str, trg_str = self.get_nodes_info()
        raise ValueError(f"{trg_str} nodes do not exists")

    # Setup nodes
    self.recurrent = is_same_pop(self.source, self.target, quick=self.quick)
    self.source_ids = [s.node_id for s in self.source]
    self.n_source = len(self.source_ids)
    self.source_list = list(self.source)
    if self.recurrent:
        self.target_ids = self.source_ids
        self.n_target = self.n_source
        self.target_list = self.source_list
    else:
        self.target_ids = [t.node_id for t in self.target]
        self.n_target = len(self.target_ids)
        self.target_list = list(self.target)

    # Setup for recurrent connection
    if self.recurrent:
        self.symmetric_p1_arg = True
        self.symmetric_p1 = True
        self.vars["n_syn1"] = self.vars["n_syn0"]
    if self.symmetric_p1_arg:
        self.vars["p1_arg"] = self.vars["p0_arg"]
    if self.symmetric_p1:
        self.vars["p1"] = self.vars["p0"]

bmtool.connectors.UnidirectionConnector

Bases: AbstractConnector

Object for building unidirectional connections in bmtk network model with given probability within a single population (or between two populations).

Parameters: p, p_arg: Probability of forward connection and its input argument when it is a function, similar to p0, p0_arg in ReciprocalConnector. It can be a constant or a deterministic function whose value must be within range [0, 1]. When p is constant, the connection is homogenous. n_syn: Number of synapses in the forward connection if connected. It can be a constant or a (deterministic or random) function whose input arguments are two node objects in BMTK like p_arg. rng: Random number generator (numpy.random.Generator) used for stochastic connection decisions. Provide a seeded generator for reproducible connectivity. If None (default), fall back to the module-level default RNG returned by _get_default_rng(). save_report: Whether to save connection report to CSV file. report_name: Filename for connection report. If None, uses the module-level report_name. verbose: Whether show verbose information in console.

Returns: An object that works with BMTK to build edges in a network.

Important attributes: vars: Dictionary that stores part of the original input parameters. source, target: NodePool objects for the source and target populations. conn_prop: A dictionaries that stores properties of connected pairs. Each key is the source node id and the value is a dictionary, where each key is the target node id that the source node connects to, and the value is the value of p_arg. Example: {sid0: {tid0: p_arg0, tid1: p_arg1, ...}, sid1: {...}, sid2: {...}, ... } This is useful in similar manner as in ReciprocalConnector.

Source code in bmtool/connectors.py 
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
class UnidirectionConnector(AbstractConnector):
    """
    Object for building unidirectional connections in bmtk network model with
    given probability within a single population (or between two populations).

    Parameters:
        p, p_arg: Probability of forward connection and its input argument when
            it is a function, similar to p0, p0_arg in ReciprocalConnector. It
            can be a constant or a deterministic function whose value must be
            within range [0, 1]. When p is constant, the connection is
            homogenous.
        n_syn: Number of synapses in the forward connection if connected. It
            can be a constant or a (deterministic or random) function whose
            input arguments are two node objects in BMTK like p_arg.
        rng: Random number generator (`numpy.random.Generator`) used for
            stochastic connection decisions. Provide a seeded generator for
            reproducible connectivity. If None (default), fall back to the
            module-level default RNG returned by _get_default_rng().
        save_report: Whether to save connection report to CSV file.
        report_name: Filename for connection report. If None, uses the
            module-level report_name.
        verbose: Whether show verbose information in console.

    Returns:
        An object that works with BMTK to build edges in a network.

    Important attributes:
        vars: Dictionary that stores part of the original input parameters.
        source, target: NodePool objects for the source and target populations.
        conn_prop: A dictionaries that stores properties of connected pairs.
        Each key is the source node id and the value is a dictionary, where
        each key is the target node id that the source node connects to, and
        the value is the value of p_arg.
            Example: {sid0: {tid0: p_arg0, tid1: p_arg1, ...},
                      sid1: {...}, sid2: {...}, ... }
            This is useful in similar manner as in ReciprocalConnector.
    """

    def __init__(
        self, p=1.0, p_arg=None, n_syn=1, verbose=True, save_report=True, report_name=None, rng=None
    ):
        args = locals()
        var_set = ("p", "p_arg", "n_syn")
        self.vars = {key: args[key] for key in var_set}

        self.verbose = verbose
        self.save_report = save_report
        if report_name is None:
            report_name = globals().get("report_name", "default_report.csv")
        self.report_name = report_name

        self.conn_prop = {}
        self.iter_count = 0

        self.rng = _get_default_rng() if rng is None else rng

    # *** Two methods executed during bmtk edge creation net.add_edges() ***
    def setup_nodes(self, source=None, target=None):
        """Must run this before building connections"""
        # Update node pools
        self.source = source
        self.target = target
        if self.source is None or len(self.source) == 0:
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(f"{src_str} nodes do not exists")
        if self.target is None or len(self.target) == 0:
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(f"{trg_str} nodes do not exists")
        self.n_pair = len(self.source) * len(self.target)

    def edge_params(self):
        """Create the arguments for BMTK add_edges() method"""
        params = {
            "source": self.source,
            "target": self.target,
            "iterator": "one_to_one",
            "connection_rule": self.make_connection,
        }
        return params

    # *** Methods executed during bmtk network.build() ***
    # *** Helper functions ***
    def add_conn_prop(self, sid, tid, prop):
        """Store p0_arg and p1_arg for a connected pair"""
        trg_dict = self.conn_prop.setdefault(sid, {})
        trg_dict[tid] = prop

    def get_conn_prop(self, sid, tid):
        """Get stored value given node ids in a connection"""
        return self.conn_prop[sid][tid]

    def setup_variables(self):
        """Make constant variables constant functions"""
        for name, var in self.vars.items():
            if callable(var):
                setattr(self, name, var)
            else:
                setattr(self, name, self.constant_function(var))

    def initialize(self):
        self.setup_variables()
        self.n_conn = 0
        self.n_poss = 0
        if self.verbose:
            self.timer = Timer()

    def make_connection(self, source, target, *args, **kwargs):
        """Assign number of synapses per iteration using one_to_one iterator"""
        # Initialize in the first iteration
        if self.iter_count == 0:
            self.initialize()
            if self.verbose:
                src_str, trg_str = self.get_nodes_info()
                print(
                    "\nStart building connection \n  from " + src_str + "\n  to " + trg_str,
                    flush=True,
                )

        # Make random connections

        p_arg = self.p_arg(source, target)
        p = self.p(p_arg)
        possible = p > 0
        self.n_poss += possible
        if possible and decision(p, rng=self.rng):
            nsyns = self.n_syn(source, target)
            self.add_conn_prop(source.node_id, target.node_id, p_arg)
            self.n_conn += 1
        else:
            nsyns = 0

        self.iter_count += 1

        # Detect end of iteration
        if self.iter_count == self.n_pair:
            if self.verbose:
                self.connection_number_info()
                self.timer.report("Done! \nTime for building connections")
            if self.save_report:
                self.save_connection_report()

        return nsyns

    # *** Helper functions for verbose ***
    def get_nodes_info(self):
        """Get strings with source and target population information"""
        source_str = self.source.network_name + ": " + self.source.filter_str
        target_str = self.target.network_name + ": " + self.target.filter_str
        return source_str, target_str

    def connection_number_info(self):
        """Print connection numbers after connections built"""
        print("Number of connected pairs: %d" % self.n_conn, flush=True)
        print("Number of possible connections: %d" % self.n_poss, flush=True)
        print(
            "Fraction of connected pairs in possible ones: %.2f%%"
            % (100.0 * self.n_conn / self.n_poss)
            if self.n_poss
            else 0.0
        )
        print("Number of total pairs: %d" % self.n_pair, flush=True)
        print(
            "Fraction of connected pairs in all pairs: %.2f%%\n"
            % (100.0 * self.n_conn / self.n_pair),
            flush=True,
        )

    def save_connection_report(self):
        """Save connections into a CSV file to be read from later"""
        src_str, trg_str = self.get_nodes_info()

        possible_fraction = 100.0 * self.n_conn / self.n_poss
        all_fraction = 100.0 * self.n_conn / self.n_pair

        # Extract the population name from source_str and target_str
        data = {
            "Source": [src_str],
            "Target": [trg_str],
            "Percent connectionivity within possible connections": [possible_fraction],
            "Percent connectionivity within all connections": [all_fraction],
        }
        df = pd.DataFrame(data)

        # Append the data to the CSV file
        try:
            # Check if the file exists by trying to read it
            existing_df = pd.read_csv(self.report_name)
            # If no exception is raised, append without header
            df.to_csv(self.report_name, mode="a", header=False, index=False)
        except FileNotFoundError:
            # If the file does not exist, write with header
            df.to_csv(self.report_name, mode="w", header=True, index=False)

add_conn_prop(sid, tid, prop)

Store p0_arg and p1_arg for a connected pair

Source code in bmtool/connectors.py 
1374
1375
1376
1377
def add_conn_prop(self, sid, tid, prop):
    """Store p0_arg and p1_arg for a connected pair"""
    trg_dict = self.conn_prop.setdefault(sid, {})
    trg_dict[tid] = prop

connection_number_info()

Print connection numbers after connections built

Source code in bmtool/connectors.py 
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
def connection_number_info(self):
    """Print connection numbers after connections built"""
    print("Number of connected pairs: %d" % self.n_conn, flush=True)
    print("Number of possible connections: %d" % self.n_poss, flush=True)
    print(
        "Fraction of connected pairs in possible ones: %.2f%%"
        % (100.0 * self.n_conn / self.n_poss)
        if self.n_poss
        else 0.0
    )
    print("Number of total pairs: %d" % self.n_pair, flush=True)
    print(
        "Fraction of connected pairs in all pairs: %.2f%%\n"
        % (100.0 * self.n_conn / self.n_pair),
        flush=True,
    )

edge_params()

Create the arguments for BMTK add_edges() method

Source code in bmtool/connectors.py 
1362
1363
1364
1365
1366
1367
1368
1369
1370
def edge_params(self):
    """Create the arguments for BMTK add_edges() method"""
    params = {
        "source": self.source,
        "target": self.target,
        "iterator": "one_to_one",
        "connection_rule": self.make_connection,
    }
    return params

get_conn_prop(sid, tid)

Get stored value given node ids in a connection

Source code in bmtool/connectors.py 
1379
1380
1381
def get_conn_prop(self, sid, tid):
    """Get stored value given node ids in a connection"""
    return self.conn_prop[sid][tid]

get_nodes_info()

Get strings with source and target population information

Source code in bmtool/connectors.py 
1436
1437
1438
1439
1440
def get_nodes_info(self):
    """Get strings with source and target population information"""
    source_str = self.source.network_name + ": " + self.source.filter_str
    target_str = self.target.network_name + ": " + self.target.filter_str
    return source_str, target_str

make_connection(source, target, *args, **kwargs)

Assign number of synapses per iteration using one_to_one iterator

Source code in bmtool/connectors.py 
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
def make_connection(self, source, target, *args, **kwargs):
    """Assign number of synapses per iteration using one_to_one iterator"""
    # Initialize in the first iteration
    if self.iter_count == 0:
        self.initialize()
        if self.verbose:
            src_str, trg_str = self.get_nodes_info()
            print(
                "\nStart building connection \n  from " + src_str + "\n  to " + trg_str,
                flush=True,
            )

    # Make random connections

    p_arg = self.p_arg(source, target)
    p = self.p(p_arg)
    possible = p > 0
    self.n_poss += possible
    if possible and decision(p, rng=self.rng):
        nsyns = self.n_syn(source, target)
        self.add_conn_prop(source.node_id, target.node_id, p_arg)
        self.n_conn += 1
    else:
        nsyns = 0

    self.iter_count += 1

    # Detect end of iteration
    if self.iter_count == self.n_pair:
        if self.verbose:
            self.connection_number_info()
            self.timer.report("Done! \nTime for building connections")
        if self.save_report:
            self.save_connection_report()

    return nsyns

save_connection_report()

Save connections into a CSV file to be read from later

Source code in bmtool/connectors.py 
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
def save_connection_report(self):
    """Save connections into a CSV file to be read from later"""
    src_str, trg_str = self.get_nodes_info()

    possible_fraction = 100.0 * self.n_conn / self.n_poss
    all_fraction = 100.0 * self.n_conn / self.n_pair

    # Extract the population name from source_str and target_str
    data = {
        "Source": [src_str],
        "Target": [trg_str],
        "Percent connectionivity within possible connections": [possible_fraction],
        "Percent connectionivity within all connections": [all_fraction],
    }
    df = pd.DataFrame(data)

    # Append the data to the CSV file
    try:
        # Check if the file exists by trying to read it
        existing_df = pd.read_csv(self.report_name)
        # If no exception is raised, append without header
        df.to_csv(self.report_name, mode="a", header=False, index=False)
    except FileNotFoundError:
        # If the file does not exist, write with header
        df.to_csv(self.report_name, mode="w", header=True, index=False)

setup_nodes(source=None, target=None)

Must run this before building connections

Source code in bmtool/connectors.py 
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
def setup_nodes(self, source=None, target=None):
    """Must run this before building connections"""
    # Update node pools
    self.source = source
    self.target = target
    if self.source is None or len(self.source) == 0:
        src_str, trg_str = self.get_nodes_info()
        raise ValueError(f"{src_str} nodes do not exists")
    if self.target is None or len(self.target) == 0:
        src_str, trg_str = self.get_nodes_info()
        raise ValueError(f"{trg_str} nodes do not exists")
    self.n_pair = len(self.source) * len(self.target)

setup_variables()

Make constant variables constant functions

Source code in bmtool/connectors.py 
1383
1384
1385
1386
1387
1388
1389
def setup_variables(self):
    """Make constant variables constant functions"""
    for name, var in self.vars.items():
        if callable(var):
            setattr(self, name, var)
        else:
            setattr(self, name, self.constant_function(var))

bmtool.connectors.GapJunction

Bases: UnidirectionConnector

Object for buiilding gap junction connections in bmtk network model with given probabilities within a single population which is uncorrelated with the recurrent chemical synapses in this population.

Parameters: p, p_arg: Probability of forward connection and its input argument when it is a function, similar to p0, p0_arg in ReciprocalConnector. It can be a constant or a deterministic function whose value must be within range [0, 1]. When p is constant, the connection is homogenous. rng: Random number generator (numpy.random.Generator) used for stochastic gap-junction decisions. Provide a seeded generator for reproducible connectivity. If None (default), fall back to the module-level default RNG returned by _get_default_rng(). save_report: Whether to save connection report to CSV file. report_name: Filename for connection report. If None, uses the module-level report_name. verbose: Whether show verbose information in console.

Returns: An object that works with BMTK to build edges in a network.

Important attributes: Similar to UnidirectionConnector.

Source code in bmtool/connectors.py 
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
class GapJunction(UnidirectionConnector):
    """
    Object for buiilding gap junction connections in bmtk network model with
    given probabilities within a single population which is uncorrelated with
    the recurrent chemical synapses in this population.

    Parameters:
        p, p_arg: Probability of forward connection and its input argument when
            it is a function, similar to p0, p0_arg in ReciprocalConnector. It
            can be a constant or a deterministic function whose value must be
            within range [0, 1]. When p is constant, the connection is
            homogenous.
        rng: Random number generator (`numpy.random.Generator`) used for
            stochastic gap-junction decisions. Provide a seeded generator for
            reproducible connectivity. If None (default), fall back to the
            module-level default RNG returned by _get_default_rng().
        save_report: Whether to save connection report to CSV file.
        report_name: Filename for connection report. If None, uses the
            module-level report_name.
        verbose: Whether show verbose information in console.

    Returns:
        An object that works with BMTK to build edges in a network.

    Important attributes:
        Similar to `UnidirectionConnector`.
    """

    def __init__(self, p=1.0, p_arg=None, verbose=True, save_report=True, report_name=None, rng=None):
        super().__init__(
            p=p, p_arg=p_arg, verbose=verbose, save_report=save_report, report_name=report_name, rng=rng
        )

    def setup_nodes(self, source=None, target=None):
        super().setup_nodes(source=source, target=target)
        if len(self.source) != len(self.target):
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(
                f"Source and target must be the same for "
                f"gap junction. Nodes are {src_str} and {trg_str}"
            )
        self.n_source = len(self.source)

    def make_connection(self, source, target, *args, **kwargs):
        """Assign gap junction per iteration using one_to_one iterator"""
        # Initialize in the first iteration
        if self.iter_count == 0:
            self.initialize()
            if self.verbose:
                src_str, _ = self.get_nodes_info()
                print("\nStart building gap junction \n  in " + src_str, flush=True)

        # Consider each pair only once
        nsyns = 0
        i, j = divmod(self.iter_count, self.n_source)
        if i < j:
            p_arg = self.p_arg(source, target)
            p = self.p(p_arg)
            possible = p > 0
            self.n_poss += possible
            if possible and decision(p, rng=self.rng):
                nsyns = 1
                sid, tid = source.node_id, target.node_id
                self.add_conn_prop(sid, tid, p_arg)
                self.add_conn_prop(tid, sid, p_arg)
                self.n_conn += 1

        self.iter_count += 1

        # Detect end of iteration
        if self.iter_count == self.n_pair:
            if self.verbose:
                self.connection_number_info()
                self.timer.report("Done! \nTime for building connections")
            if self.save_report:
                self.save_connection_report()
        return nsyns

    def connection_number_info(self):
        n_pair = self.n_pair
        self.n_pair = (n_pair - len(self.source)) // 2
        super().connection_number_info()
        self.n_pair = n_pair

    def save_connection_report(self):
        """Save connections into a CSV file to be read from later"""
        src_str, trg_str = self.get_nodes_info()
        n_pair = self.n_pair
        fraction_0 = self.n_conn / self.n_poss if self.n_poss else 0.0
        fraction_1 = self.n_conn / self.n_pair

        # Convert fraction to percentage and prepare data for the DataFrame
        data = {
            "Source": [src_str + "Gap"],
            "Target": [trg_str + "Gap"],
            "Percent connectionivity within possible connections": [fraction_0 * 100],
            "Percent connectionivity within all connections": [fraction_1 * 100],
        }
        df = pd.DataFrame(data)

        # Append the data to the CSV file
        try:
            # Check if the file exists by trying to read it
            existing_df = pd.read_csv(self.report_name)
            # If no exception is raised, append without header
            df.to_csv(self.report_name, mode="a", header=False, index=False)
        except FileNotFoundError:
            # If the file does not exist, write with header
            df.to_csv(self.report_name, mode="w", header=True, index=False)

make_connection(source, target, *args, **kwargs)

Assign gap junction per iteration using one_to_one iterator

Source code in bmtool/connectors.py 
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
def make_connection(self, source, target, *args, **kwargs):
    """Assign gap junction per iteration using one_to_one iterator"""
    # Initialize in the first iteration
    if self.iter_count == 0:
        self.initialize()
        if self.verbose:
            src_str, _ = self.get_nodes_info()
            print("\nStart building gap junction \n  in " + src_str, flush=True)

    # Consider each pair only once
    nsyns = 0
    i, j = divmod(self.iter_count, self.n_source)
    if i < j:
        p_arg = self.p_arg(source, target)
        p = self.p(p_arg)
        possible = p > 0
        self.n_poss += possible
        if possible and decision(p, rng=self.rng):
            nsyns = 1
            sid, tid = source.node_id, target.node_id
            self.add_conn_prop(sid, tid, p_arg)
            self.add_conn_prop(tid, sid, p_arg)
            self.n_conn += 1

    self.iter_count += 1

    # Detect end of iteration
    if self.iter_count == self.n_pair:
        if self.verbose:
            self.connection_number_info()
            self.timer.report("Done! \nTime for building connections")
        if self.save_report:
            self.save_connection_report()
    return nsyns

save_connection_report()

Save connections into a CSV file to be read from later

Source code in bmtool/connectors.py 
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
def save_connection_report(self):
    """Save connections into a CSV file to be read from later"""
    src_str, trg_str = self.get_nodes_info()
    n_pair = self.n_pair
    fraction_0 = self.n_conn / self.n_poss if self.n_poss else 0.0
    fraction_1 = self.n_conn / self.n_pair

    # Convert fraction to percentage and prepare data for the DataFrame
    data = {
        "Source": [src_str + "Gap"],
        "Target": [trg_str + "Gap"],
        "Percent connectionivity within possible connections": [fraction_0 * 100],
        "Percent connectionivity within all connections": [fraction_1 * 100],
    }
    df = pd.DataFrame(data)

    # Append the data to the CSV file
    try:
        # Check if the file exists by trying to read it
        existing_df = pd.read_csv(self.report_name)
        # If no exception is raised, append without header
        df.to_csv(self.report_name, mode="a", header=False, index=False)
    except FileNotFoundError:
        # If the file does not exist, write with header
        df.to_csv(self.report_name, mode="w", header=True, index=False)

bmtool.connectors.CorrelatedGapJunction

Bases: GapJunction

Object for buiilding gap junction connections in bmtk network model with given probabilities within a single population which could be correlated with the recurrent chemical synapses in this population.

Parameters: p_non, p_uni, p_rec: Probabilities of gap junction connection for each pair of cells given the following three conditions of chemical synaptic connections between them, no connection, unidirectional, and reciprocal, respectively. It can be a constant or a deterministic function whose value must be within range [0, 1]. p_arg: Input argument for p_non, p_uni, or p_rec, when any of them is a function, similar to p0_arg, p1_arg in ReciprocalConnector. connector: Connector object used to generate the chemical synapses of within this population, which contains the connection information in its attribute conn_prop. So this connector should have generated the chemical synapses before generating the gap junction. rng: Random number generator (numpy.random.Generator) used for stochastic gap-junction decisions. Provide a seeded generator for reproducible connectivity. If None (default), fall back to the module-level default RNG returned by _get_default_rng(). save_report: Whether to save connection report to CSV file. report_name: Filename for connection report. If None, uses the module-level report_name. verbose: Whether show verbose information in console.

Returns: An object that works with BMTK to build edges in a network.

Important attributes: Similar to UnidirectionConnector.

Source code in bmtool/connectors.py 
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
class CorrelatedGapJunction(GapJunction):
    """
    Object for buiilding gap junction connections in bmtk network model with
    given probabilities within a single population which could be correlated
    with the recurrent chemical synapses in this population.

    Parameters:
        p_non, p_uni, p_rec: Probabilities of gap junction connection for each
            pair of cells given the following three conditions of chemical
            synaptic connections between them, no connection, unidirectional,
            and reciprocal, respectively. It can be a constant or a
            deterministic function whose value must be within range [0, 1].
        p_arg: Input argument for p_non, p_uni, or p_rec, when any of them is a
            function, similar to p0_arg, p1_arg in ReciprocalConnector.
        connector: Connector object used to generate the chemical synapses of
            within this population, which contains the connection information
            in its attribute `conn_prop`. So this connector should have
            generated the chemical synapses before generating the gap junction.
        rng: Random number generator (`numpy.random.Generator`) used for
            stochastic gap-junction decisions. Provide a seeded generator for
            reproducible connectivity. If None (default), fall back to the
            module-level default RNG returned by _get_default_rng().
        save_report: Whether to save connection report to CSV file.
        report_name: Filename for connection report. If None, uses the
            module-level report_name.
        verbose: Whether show verbose information in console.

    Returns:
        An object that works with BMTK to build edges in a network.

    Important attributes:
        Similar to `UnidirectionConnector`.
    """

    def __init__(
        self,
        p_non=1.0,
        p_uni=1.0,
        p_rec=1.0,
        p_arg=None,
        connector=None,
        verbose=True,
        save_report=True,
        report_name=None,
        rng=None,
    ):
        super().__init__(
            p=p_non, p_arg=p_arg, verbose=verbose, save_report=save_report, report_name=None, rng=rng
        )
        self.vars["p_non"] = self.vars.pop("p")
        self.vars["p_uni"] = p_uni
        self.vars["p_rec"] = p_rec
        self.connector = connector
        conn_prop = connector.conn_prop
        if isinstance(conn_prop, list):
            conn_prop = conn_prop[0]
        self.ref_conn_prop = conn_prop

    def conn_exist(self, sid, tid):
        trg_dict = self.ref_conn_prop.get(sid)
        if trg_dict is not None and tid in trg_dict:
            return True, trg_dict[tid]
        else:
            return False, None

    def connection_type(self, sid, tid):
        conn0, prop0 = self.conn_exist(sid, tid)
        conn1, prop1 = self.conn_exist(tid, sid)
        return conn0 + conn1, prop0 if conn0 else prop1

    def initialize(self):
        self.has_p_arg = self.vars["p_arg"] is not None
        if not self.has_p_arg:
            var = self.connector.vars
            self.vars["p_arg"] = var.get("p_arg", var.get("p0_arg", None))
        super().initialize()
        self.ps = [self.p_non, self.p_uni, self.p_rec]

    def make_connection(self, source, target, *args, **kwargs):
        """Assign gap junction per iteration using one_to_one iterator"""
        # Initialize in the first iteration
        if self.iter_count == 0:
            self.initialize()
            if self.verbose:
                src_str, _ = self.get_nodes_info()
                print("\nStart building gap junction \n  in " + src_str, flush=True)

        # Consider each pair only once
        nsyns = 0
        i, j = divmod(self.iter_count, self.n_source)
        if i < j:
            sid, tid = source.node_id, target.node_id
            conn_type, p_arg = self.connection_type(sid, tid)
            if self.has_p_arg or not conn_type:
                p_arg = self.p_arg(source, target)
            p = self.ps[conn_type](p_arg)
            possible = p > 0
            self.n_poss += possible
            if possible and decision(p, rng=self.rng):
                nsyns = 1
                self.add_conn_prop(sid, tid, p_arg)
                self.add_conn_prop(tid, sid, p_arg)
                self.n_conn += 1

        self.iter_count += 1

        # Detect end of iteration
        if self.iter_count == self.n_pair:
            if self.verbose:
                self.connection_number_info()
                self.timer.report("Done! \nTime for building connections")
            if self.save_report:
                self.save_connection_report()
        return nsyns

make_connection(source, target, *args, **kwargs)

Assign gap junction per iteration using one_to_one iterator

Source code in bmtool/connectors.py 
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
def make_connection(self, source, target, *args, **kwargs):
    """Assign gap junction per iteration using one_to_one iterator"""
    # Initialize in the first iteration
    if self.iter_count == 0:
        self.initialize()
        if self.verbose:
            src_str, _ = self.get_nodes_info()
            print("\nStart building gap junction \n  in " + src_str, flush=True)

    # Consider each pair only once
    nsyns = 0
    i, j = divmod(self.iter_count, self.n_source)
    if i < j:
        sid, tid = source.node_id, target.node_id
        conn_type, p_arg = self.connection_type(sid, tid)
        if self.has_p_arg or not conn_type:
            p_arg = self.p_arg(source, target)
        p = self.ps[conn_type](p_arg)
        possible = p > 0
        self.n_poss += possible
        if possible and decision(p, rng=self.rng):
            nsyns = 1
            self.add_conn_prop(sid, tid, p_arg)
            self.add_conn_prop(tid, sid, p_arg)
            self.n_conn += 1

    self.iter_count += 1

    # Detect end of iteration
    if self.iter_count == self.n_pair:
        if self.verbose:
            self.connection_number_info()
            self.timer.report("Done! \nTime for building connections")
        if self.save_report:
            self.save_connection_report()
    return nsyns

bmtool.connectors.OneToOneSequentialConnector

Bases: AbstractConnector

Object for buiilding one to one correspondence connections in bmtk network model with between two populations. One of the population can consist of multiple sub-populations. These sub-populations need to be added sequentially using setup_nodes() and edge_params() methods followed by BMTK add_edges() method. For example, to connect 30 nodes in population A to 30 nodes in populations B1, B2, B3, each with 10 nodes, set up as follows. connector = OneToOneSequentialConnector(parameters) connector.setup_nodes(source=A, target=B1) net.add_edges(connector.edge_params()) connector.setup_nodes(target=B2) net.add_edges(connector.edge_params()) connector.setup_nodes(target=B3) net.add_edges(connector.edge_params()) After BMTK executes net.build(), the first 10 nodes in A will connect one- to-one to the 10 nodes in B1, then the 11 to 20-th nodes to those in B2, finally the 21 to 30-th nodes to those in B3. This connector is useful for creating input drives to a population. Each node in it receives an independent drive from a unique source node.

Parameters: n_syn: Number of synapses in each connection. It accepts only constant for now. partition_source: Whether the source population consists of multiple sub-populations. By default, the source has one population, and the target can have multiple sub-populations. If set to true, the source can have multiple sub-populations and the target has only one population. verbose: Whether show verbose information in console.

Returns: An object that works with BMTK to build edges in a network.

Important attributes: source: NodePool object for the single population. targets: List of NodePool objects for the multiple sub-populations.

Source code in bmtool/connectors.py 
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
class OneToOneSequentialConnector(AbstractConnector):
    """Object for buiilding one to one correspondence connections in bmtk
    network model with between two populations. One of the population can
    consist of multiple sub-populations. These sub-populations need to be added
    sequentially using setup_nodes() and edge_params() methods followed by BMTK
    add_edges() method. For example, to connect 30 nodes in population A to 30
    nodes in populations B1, B2, B3, each with 10 nodes, set up as follows.
        connector = OneToOneSequentialConnector(**parameters)
        connector.setup_nodes(source=A, target=B1)
        net.add_edges(**connector.edge_params())
        connector.setup_nodes(target=B2)
        net.add_edges(**connector.edge_params())
        connector.setup_nodes(target=B3)
        net.add_edges(**connector.edge_params())
    After BMTK executes net.build(), the first 10 nodes in A will connect one-
    to-one to the 10 nodes in B1, then the 11 to 20-th nodes to those in B2,
    finally the 21 to 30-th nodes to those in B3.
    This connector is useful for creating input drives to a population. Each
    node in it receives an independent drive from a unique source node.

    Parameters:
        n_syn: Number of synapses in each connection. It accepts only constant
            for now.
        partition_source: Whether the source population consists of multiple
            sub-populations. By default, the source has one population, and the
            target can have multiple sub-populations. If set to true, the
            source can have multiple sub-populations and the target has only
            one population.
        verbose: Whether show verbose information in console.

    Returns:
        An object that works with BMTK to build edges in a network.

    Important attributes:
        source: NodePool object for the single population.
        targets: List of NodePool objects for the multiple sub-populations.
    """

    def __init__(self, n_syn=1, partition_source=False, verbose=True):
        self.n_syn = int(n_syn)
        self.partition_source = partition_source
        self.verbose = verbose

        self.targets = []
        self.n_source = 0
        self.idx_range = [0]
        self.target_count = 0
        self.iter_count = 0

    # *** Two methods executed during bmtk edge creation net.add_edges() ***
    def setup_nodes(self, source=None, target=None):
        """Must run this before building connections"""
        # Update node pools
        if self.partition_source:
            source, target = target, source
        if self.target_count == 0:
            if source is None or len(source) == 0:
                src_str, trg_str = self.get_nodes_info()
                raise ValueError(
                    (f"{trg_str}" if self.partition_source else f"{src_str}")
                    + " nodes do not exists"
                )
            self.source = source
            self.n_source = len(source)
        if target is None or len(target) == 0:
            raise ValueError(
                ("Source" if self.partition_source else "Target") + " nodes do not exists"
            )

        self.targets.append(target)
        self.idx_range.append(self.idx_range[-1] + len(target))
        self.target_count += 1

        if self.idx_range[-1] > self.n_source:
            if self.partition_source:
                raise ValueError(
                    "Total target populations exceed the source population."
                    if self.partition_source
                    else "Total source populations exceed the target population."
                )

        if self.verbose and self.idx_range[-1] == self.n_source:
            print(
                "All "
                + ("source" if self.partition_source else "target")
                + " population partitions are filled.",
                flush=True,
            )

    def edge_params(self, target_pop_idx=-1):
        """Create the arguments for BMTK add_edges() method"""
        if self.partition_source:
            params = {
                "source": self.targets[target_pop_idx],
                "target": self.source,
                "iterator": "one_to_all",
            }
        else:
            params = {
                "source": self.source,
                "target": self.targets[target_pop_idx],
                "iterator": "all_to_one",
            }
        params["connection_rule"] = self.make_connection
        return params

    # *** Methods executed during bmtk network.build() ***
    def make_connection(self, source, targets, *args, **kwargs):
        """Assign one connection per iteration using all_to_one iterator"""
        # Initialize in the first iteration
        if self.verbose:
            if self.iter_count == 0:
                # Very beginning
                self.target_count = 0
                src_str, trg_str = self.get_nodes_info()
                print(
                    "\nStart building connection "
                    + ("to " if self.partition_source else "from ")
                    + src_str,
                    flush=True,
                )
                self.timer = Timer()

            if self.iter_count == self.idx_range[self.target_count]:
                # Beginning of each target population
                src_str, trg_str = self.get_nodes_info(self.target_count)
                print(
                    ("  %d. " % self.target_count)
                    + ("from " if self.partition_source else "to ")
                    + trg_str,
                    flush=True,
                )
                self.target_count += 1
                self.timer_part = Timer()

        # Make connection
        nsyns = np.zeros(self.n_source, dtype=int)
        nsyns[self.iter_count] = self.n_syn
        self.iter_count += 1

        # Detect end of iteration
        if self.verbose:
            if self.iter_count == self.idx_range[self.target_count]:
                # End of each target population
                self.timer_part.report("    Time for this partition")
            if self.iter_count == self.n_source:
                # Very end
                self.timer.report("Done! \nTime for building connections")
        return nsyns

    # *** Helper functions for verbose ***
    def get_nodes_info(self, target_pop_idx=-1):
        """Get strings with source and target population information"""
        target = self.targets[target_pop_idx]
        source_str = self.source.network_name + ": " + self.source.filter_str
        target_str = target.network_name + ": " + target.filter_str
        return source_str, target_str

edge_params(target_pop_idx=-1)

Create the arguments for BMTK add_edges() method

Source code in bmtool/connectors.py 
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
def edge_params(self, target_pop_idx=-1):
    """Create the arguments for BMTK add_edges() method"""
    if self.partition_source:
        params = {
            "source": self.targets[target_pop_idx],
            "target": self.source,
            "iterator": "one_to_all",
        }
    else:
        params = {
            "source": self.source,
            "target": self.targets[target_pop_idx],
            "iterator": "all_to_one",
        }
    params["connection_rule"] = self.make_connection
    return params

get_nodes_info(target_pop_idx=-1)

Get strings with source and target population information

Source code in bmtool/connectors.py 
2494
2495
2496
2497
2498
2499
def get_nodes_info(self, target_pop_idx=-1):
    """Get strings with source and target population information"""
    target = self.targets[target_pop_idx]
    source_str = self.source.network_name + ": " + self.source.filter_str
    target_str = target.network_name + ": " + target.filter_str
    return source_str, target_str

make_connection(source, targets, *args, **kwargs)

Assign one connection per iteration using all_to_one iterator

Source code in bmtool/connectors.py 
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
def make_connection(self, source, targets, *args, **kwargs):
    """Assign one connection per iteration using all_to_one iterator"""
    # Initialize in the first iteration
    if self.verbose:
        if self.iter_count == 0:
            # Very beginning
            self.target_count = 0
            src_str, trg_str = self.get_nodes_info()
            print(
                "\nStart building connection "
                + ("to " if self.partition_source else "from ")
                + src_str,
                flush=True,
            )
            self.timer = Timer()

        if self.iter_count == self.idx_range[self.target_count]:
            # Beginning of each target population
            src_str, trg_str = self.get_nodes_info(self.target_count)
            print(
                ("  %d. " % self.target_count)
                + ("from " if self.partition_source else "to ")
                + trg_str,
                flush=True,
            )
            self.target_count += 1
            self.timer_part = Timer()

    # Make connection
    nsyns = np.zeros(self.n_source, dtype=int)
    nsyns[self.iter_count] = self.n_syn
    self.iter_count += 1

    # Detect end of iteration
    if self.verbose:
        if self.iter_count == self.idx_range[self.target_count]:
            # End of each target population
            self.timer_part.report("    Time for this partition")
        if self.iter_count == self.n_source:
            # Very end
            self.timer.report("Done! \nTime for building connections")
    return nsyns

setup_nodes(source=None, target=None)

Must run this before building connections

Source code in bmtool/connectors.py 
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
def setup_nodes(self, source=None, target=None):
    """Must run this before building connections"""
    # Update node pools
    if self.partition_source:
        source, target = target, source
    if self.target_count == 0:
        if source is None or len(source) == 0:
            src_str, trg_str = self.get_nodes_info()
            raise ValueError(
                (f"{trg_str}" if self.partition_source else f"{src_str}")
                + " nodes do not exists"
            )
        self.source = source
        self.n_source = len(source)
    if target is None or len(target) == 0:
        raise ValueError(
            ("Source" if self.partition_source else "Target") + " nodes do not exists"
        )

    self.targets.append(target)
    self.idx_range.append(self.idx_range[-1] + len(target))
    self.target_count += 1

    if self.idx_range[-1] > self.n_source:
        if self.partition_source:
            raise ValueError(
                "Total target populations exceed the source population."
                if self.partition_source
                else "Total source populations exceed the target population."
            )

    if self.verbose and self.idx_range[-1] == self.n_source:
        print(
            "All "
            + ("source" if self.partition_source else "target")
            + " population partitions are filled.",
            flush=True,
        )

Synapse Helper Functions

bmtool.connectors.syn_const_delay(source=None, target=None, dist=100, min_delay=SYN_MIN_DELAY, velocity=SYN_VELOCITY, fluc_stdev=FLUC_STDEV, delay_bound=(DELAY_LOWBOUND, DELAY_UPBOUND), connector=None, rng=None, **kwargs)

Synapse delay approximately constant with some random fluctuation.

Parameters:

Name Type Description Default
source optional

Included for API consistency; not used in this implementation.

 None
target optional

Included for API consistency; not used in this implementation.

 None
dist float

Distance between source and target (micron).

 100
min_delay float

Minimum delay (ms).

 SYN_MIN_DELAY
velocity float

Synapse conduction velocity (micron/ms).

 SYN_VELOCITY
fluc_stdev float

Standard deviation of random Gaussian fluctuation (ms).

 FLUC_STDEV
delay_bound tuple of (float, float)

Lower and upper bounds for the delay (ms).

 (DELAY_LOWBOUND, DELAY_UPBOUND)
connector optional

Included for API consistency; not used in this implementation.

 None
rng Generator

Random number generator. If None, uses the default generator.

 None
Source code in bmtool/connectors.py 
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
def syn_const_delay(
    source=None,
    target=None,
    dist=100,
    min_delay=SYN_MIN_DELAY,
    velocity=SYN_VELOCITY,
    fluc_stdev=FLUC_STDEV,
    delay_bound=(DELAY_LOWBOUND, DELAY_UPBOUND),
    connector=None,
    rng=None, **kwargs
):
    """Synapse delay approximately constant with some random fluctuation.
    Parameters
    ----------
    source, target : optional
        Included for API consistency; not used in this implementation.
    dist : float, optional
        Distance between source and target (micron).
    min_delay : float, optional
        Minimum delay (ms).
    velocity : float, optional
        Synapse conduction velocity (micron/ms).
    fluc_stdev : float, optional
        Standard deviation of random Gaussian fluctuation (ms).
    delay_bound : tuple of (float, float), optional
        Lower and upper bounds for the delay (ms).
    connector : optional
        Included for API consistency; not used in this implementation.
    rng : numpy.random.Generator, optional
        Random number generator. If None, uses the default generator.
    """
    rng = _get_default_rng() if rng is None else rng
    del_fluc = fluc_stdev * rng.normal()
    delay = dist / velocity + min_delay + del_fluc
    delay = min(max(delay, delay_bound[0]), delay_bound[1])
    return delay

bmtool.connectors.syn_dist_delay_feng(source, target, min_delay=SYN_MIN_DELAY, velocity=SYN_VELOCITY, fluc_stdev=FLUC_STDEV, delay_bound=(DELAY_LOWBOUND, DELAY_UPBOUND), connector=None, rng=None, **kwargs)

Synpase delay linearly dependent on distance. min_delay: minimum delay (ms) velocity: synapse conduction velocity (micron/ms) fluc_stdev: standard deviation of random Gaussian fluctuation (ms) delay_bound: (lower, upper) bounds of delay (ms) connector: connector object from which to read distance rng : numpy.random.Generator, optional Random number generator. If None, uses the default generator.

Source code in bmtool/connectors.py 
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
def syn_dist_delay_feng(
    source,
    target,
    min_delay=SYN_MIN_DELAY,
    velocity=SYN_VELOCITY,
    fluc_stdev=FLUC_STDEV,
    delay_bound=(DELAY_LOWBOUND, DELAY_UPBOUND),
    connector=None,
    rng=None,
    **kwargs
):
    """Synpase delay linearly dependent on distance.
    min_delay: minimum delay (ms)
    velocity: synapse conduction velocity (micron/ms)
    fluc_stdev: standard deviation of random Gaussian fluctuation (ms)
    delay_bound: (lower, upper) bounds of delay (ms)
    connector: connector object from which to read distance
    rng : numpy.random.Generator, optional
        Random number generator. If None, uses the default generator.
    """
    rng = _get_default_rng() if rng is None else rng
    if connector is None:
        dist = euclid_dist(target["positions"], source["positions"])
    else:
        dist = connector.get_conn_prop(source.node_id, target.node_id)
    del_fluc = fluc_stdev * rng.normal()
    delay = dist / velocity + min_delay + del_fluc
    delay = min(max(delay, delay_bound[0]), delay_bound[1])
    return delay

bmtool.connectors.syn_section_PN(source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs)

Synapse location follows a Bernoulli distribution, with probability p to obtain the former in sec_id and sec_x

Source code in bmtool/connectors.py 
2581
2582
2583
2584
2585
2586
def syn_section_PN(source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs):
    """Synapse location follows a Bernoulli distribution, with probability p
    to obtain the former in sec_id and sec_x"""
    rng = _get_default_rng() if rng is None else rng
    syn_loc = int(not decision(p, rng=rng))
    return sec_id[syn_loc], sec_x[syn_loc]

bmtool.connectors.syn_const_delay_feng_section_PN(source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs)

Assign both synapse delay and location with constant distance assumed

Source code in bmtool/connectors.py 
2589
2590
2591
2592
2593
2594
2595
def syn_const_delay_feng_section_PN(
    source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs
):
    """Assign both synapse delay and location with constant distance assumed"""
    delay = syn_const_delay(source, target, rng=rng, **kwargs)
    s_id, s_x = syn_section_PN(source, target, p=p, sec_id=sec_id, sec_x=sec_x, rng=rng)
    return delay, s_id, s_x

bmtool.connectors.syn_dist_delay_feng_section_PN(source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs)

Assign both synapse delay and location

Source code in bmtool/connectors.py 
2598
2599
2600
2601
2602
2603
2604
def syn_dist_delay_feng_section_PN(
    source, target, p=0.9, sec_id=(1, 2), sec_x=(0.4, 0.6), rng=None, **kwargs
):
    """Assign both synapse delay and location"""
    delay = syn_dist_delay_feng(source, target, rng=rng, **kwargs)
    s_id, s_x = syn_section_PN(source, target, p=p, sec_id=sec_id, sec_x=sec_x, rng=rng)
    return delay, s_id, s_x

bmtool.connectors.syn_uniform_delay_section(source, target, low=DELAY_LOWBOUND, high=DELAY_UPBOUND, rng=None, **kwargs)

Source code in bmtool/connectors.py 
2607
2608
2609
def syn_uniform_delay_section(source, target, low=DELAY_LOWBOUND, high=DELAY_UPBOUND, rng=None, **kwargs):
    rng = _get_default_rng() if rng is None else rng
    return rng.uniform(low, high)