-
Notifications
You must be signed in to change notification settings - Fork 52
/
drives.py
549 lines (480 loc) · 21.1 KB
/
drives.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
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
361
362
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
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
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
"""External drives to network."""
# Authors: Mainak Jas <mjas@mgh.harvard.edu>
# Sam Neymotin <samnemo@gmail.com>
# Christopher Bailey <bailey.cj@gmail.com>
import numpy as np
from .params import (_extract_bias_specs_from_hnn_params,
_extract_drive_specs_from_hnn_params)
def _get_target_properties(weights_ampa, weights_nmda, synaptic_delays,
location, probability=1.0):
"""Retrieve drive properties associated with each target cell type
Note that target cell types of a drive are inferred from the synaptic
weight and delay parameters keys defined by the user.
"""
# allow passing weights as None, but make iterable here
if weights_ampa is None:
weights_ampa = dict()
if weights_nmda is None:
weights_nmda = dict()
weights_by_type = {cell_type: dict() for cell_type in
(set(weights_ampa.keys()) | set(weights_ampa.keys()))}
for cell_type in weights_ampa:
weights_by_type[cell_type].update({'ampa': weights_ampa[cell_type]})
for cell_type in weights_nmda:
weights_by_type[cell_type].update({'nmda': weights_nmda[cell_type]})
target_populations = set(weights_by_type)
if not target_populations:
raise ValueError('No target cell types have been given a synaptic '
'weight for this drive.')
# Distal drives should not target L5 basket cells according to the
# canonical Jones model
if location == 'distal' and 'L5_basket' in target_populations:
raise ValueError('When adding a distal drive, synaptic weight cannot '
'be defined for the L5_basket cell type as this '
'connection does not exist.')
if isinstance(synaptic_delays, float):
delays_by_type = {cell_type: synaptic_delays for cell_type in
target_populations}
else:
delays_by_type = synaptic_delays.copy()
if set(delays_by_type.keys()) != target_populations:
raise ValueError('synaptic_delays is either a common float or needs '
'to be specified as a dict for each of the cell '
'types defined in weights_ampa and weights_nmda '
f'({target_populations})')
if isinstance(probability, float):
probability_by_type = {cell_type: probability for cell_type in
target_populations}
else:
probability_by_type = probability.copy()
if set(probability_by_type.keys()) != target_populations:
raise ValueError('probability is either a common float or needs '
'to be specified as a dict for each of the cell '
'types defined in weights_ampa and weights_nmda '
f'({target_populations})')
return (target_populations, weights_by_type, delays_by_type,
probability_by_type)
def _check_drive_parameter_values(drive_type, **kwargs):
if 'tstop' in kwargs:
if kwargs['tstop'] is not None:
if kwargs['tstop'] < 0.:
raise ValueError(f'End time of {drive_type} drive cannot be '
'negative')
if 'tstart' in kwargs and kwargs['tstop'] < kwargs['tstart']:
raise ValueError(f'Duration of {drive_type} drive cannot be '
'negative')
if 'sigma' in kwargs:
if kwargs['sigma'] < 0.:
raise ValueError('Standard deviation cannot be negative')
if 'numspikes' in kwargs:
if not kwargs['numspikes'] > 0:
raise ValueError('Number of spikes must be greater than zero')
if 'tstart' in kwargs:
if kwargs['tstart'] < 0:
raise ValueError(f'Start time of {drive_type} drive cannot be '
'negative')
if ('numspikes' in kwargs and 'spike_isi' in kwargs and
'burst_rate' in kwargs):
n_spikes = kwargs['numspikes']
isi = kwargs['spike_isi']
burst_period = 1000. / kwargs['burst_rate']
burst_duration = (n_spikes - 1) * isi
if burst_duration > burst_period:
raise ValueError(f'Burst duration ({burst_duration}s) cannot'
f' be greater than burst period ({burst_period}s)'
'Consider increasing the spike ISI or burst rate')
def _check_poisson_rates(rate_constant, target_populations, all_cell_types):
if isinstance(rate_constant, dict):
constants_provided = set(rate_constant.keys())
if not target_populations.issubset(constants_provided):
raise ValueError(
f"Rate constants not provided for all target cell "
f"populations ({target_populations})")
if not constants_provided.issubset(all_cell_types):
offending_keys = constants_provided.difference(all_cell_types)
raise ValueError(
f"Rate constant provided for unknown target cell "
f"population: {offending_keys}")
for key, val in rate_constant.items():
if not val > 0.:
raise ValueError(
f"Rate constant must be positive ({key}, {val})")
else:
if not rate_constant > 0.:
raise ValueError(
f"Rate constant must be positive, got {rate_constant}")
def _add_drives_from_params(net):
drive_specs = _extract_drive_specs_from_hnn_params(
net._params, list(net.cell_types.keys()))
bias_specs = _extract_bias_specs_from_hnn_params(
net._params, list(net.cell_types.keys()))
for drive_name in sorted(drive_specs.keys()): # order matters
specs = drive_specs[drive_name]
if specs['type'] == 'evoked':
net.add_evoked_drive(
drive_name, mu=specs['dynamics']['mu'],
sigma=specs['dynamics']['sigma'],
numspikes=specs['dynamics']['numspikes'],
n_drive_cells=specs['dynamics']['n_drive_cells'],
cell_specific=specs['cell_specific'],
weights_ampa=specs['weights_ampa'],
weights_nmda=specs['weights_nmda'],
location=specs['location'], event_seed=specs['event_seed'],
synaptic_delays=specs['synaptic_delays'],
space_constant=specs['space_constant'])
elif specs['type'] == 'poisson':
net.add_poisson_drive(
drive_name, tstart=specs['dynamics']['tstart'],
tstop=specs['dynamics']['tstop'],
rate_constant=specs['dynamics']['rate_constant'],
weights_ampa=specs['weights_ampa'],
weights_nmda=specs['weights_nmda'],
location=specs['location'], event_seed=specs['event_seed'],
synaptic_delays=specs['synaptic_delays'],
space_constant=specs['space_constant'])
elif specs['type'] == 'gaussian':
net.add_evoked_drive( # 'gaussian' is just evoked
drive_name, mu=specs['dynamics']['mu'],
sigma=specs['dynamics']['sigma'],
numspikes=specs['dynamics']['numspikes'],
weights_ampa=specs['weights_ampa'],
weights_nmda=specs['weights_nmda'],
location=specs['location'], event_seed=specs['event_seed'],
synaptic_delays=specs['synaptic_delays'],
space_constant=specs['space_constant'])
elif specs['type'] == 'bursty':
net.add_bursty_drive(
drive_name,
tstart=specs['dynamics']['tstart'],
tstart_std=specs['dynamics']['tstart_std'],
tstop=specs['dynamics']['tstop'],
burst_rate=specs['dynamics']['burst_rate'],
burst_std=specs['dynamics']['burst_std'],
numspikes=specs['dynamics']['numspikes'],
spike_isi=specs['dynamics']['spike_isi'],
n_drive_cells=specs['dynamics']['n_drive_cells'],
cell_specific=specs['cell_specific'],
weights_ampa=specs['weights_ampa'],
weights_nmda=specs['weights_nmda'],
location=specs['location'],
space_constant=specs['space_constant'],
synaptic_delays=specs['synaptic_delays'],
event_seed=specs['event_seed'])
# add tonic biases if present in params
for cellname in bias_specs['tonic']:
net.add_tonic_bias(
cell_type=cellname,
amplitude=bias_specs['tonic'][cellname]['amplitude'],
t0=bias_specs['tonic'][cellname]['t0'],
tstop=bias_specs['tonic'][cellname]['tstop'])
# in HNN-GUI, seed is determined by "absolute GID" instead of the
# gid offset with respect to the first cell of a population.
for drive_name, drive in net.external_drives.items():
drive['event_seed'] += net.gid_ranges[drive_name][0]
def _get_prng(seed, gid, sync_evinput=False):
"""Random generator for this instance.
Parameters
----------
seed : int
The seed for random state generator.
gid : int
The cell ID
sync_evinput : bool
If True, all cells get the same prng
Returns
-------
prng : instance of RandomState
The seed for events assuming a given start time.
prng2 : instance of RandomState
The seed for generating randomized start times.
Used in _create_bursty_input
"""
# XXX: some param files use seed < 0 but numpy
# does not allow this.
if seed >= 0:
# only used for randomisation of t0 of bursty drives
prng2 = np.random.RandomState(seed)
else:
prng2 = None
if not sync_evinput:
seed = seed + gid
prng = np.random.RandomState(seed)
return prng, prng2
def _drive_cell_event_times(drive_type, dynamics, tstop, target_type='any',
trial_idx=0, drive_cell_gid=0, event_seed=0):
"""Generate event times for one artificial drive cell based on dynamics.
Parameters
----------
drive_type : str
The drive type, which is one of
'poisson' : Poisson-distributed dynamics from t0 to T
'gaussian' : Gaussian-distributed dynamics from t0 to T
'evoked' : Spikes occur at specified time (mu) with dispersion (sigma)
'bursty' : Spikes occur in bursts (events_per_cycle spikes at
cycle_events_isi intervals) in a rhythmic (f_input) fashion
dynamics : dict
Parameters of the event time dynamics to simulate
tstop : float
The simulation stop time (ms).
target_type : str
Type of cell (e.g. 'L2_basket') this drive cell will target. If
'any' (default), the drive cell is non-specific.
trial_idx : int
The index number of the current trial of a simulation (default=1).
drive_cell_gid : int
Optional gid of current artificial cell (used for seeding)
event_seed : int
Optional initial seed for random number generator.
Returns
-------
event_times : list
The event times at which spikes occur.
"""
prng, prng2 = _get_prng(seed=event_seed + trial_idx,
gid=drive_cell_gid)
# check drive name validity, allowing substring matches
valid_drives = ['evoked', 'poisson', 'gaussian', 'bursty']
# NB check if drive_type has a valid substring, not vice versa
matches = [f for f in valid_drives if f in drive_type]
if len(matches) == 0:
raise ValueError('Invalid external drive: %s' % drive_type)
elif len(matches) > 1:
raise ValueError('Ambiguous external drive: %s' % drive_type)
event_times = list()
if drive_type == 'poisson':
if target_type == 'any':
rate_constant = dynamics['rate_constant']
elif target_type in dynamics['rate_constant']:
rate_constant = dynamics['rate_constant'][target_type]
# XXX required for legacy mode since drive cells are created in network
# for which rate constant may not be defined
if target_type == 'any' or target_type in dynamics['rate_constant']:
event_times = _create_extpois(
t0=dynamics['tstart'],
T=dynamics['tstop'],
lamtha=rate_constant,
prng=prng)
elif drive_type == 'evoked' or drive_type == 'gaussian':
event_times = _create_gauss(
mu=dynamics['mu'],
sigma=dynamics['sigma'],
numspikes=dynamics['numspikes'],
prng=prng)
elif drive_type == 'bursty':
event_times = _create_bursty_input(
t0=dynamics['tstart'],
t0_stdev=dynamics['tstart_std'],
tstop=dynamics['tstop'],
f_input=dynamics['burst_rate'],
events_jitter_std=dynamics['burst_std'],
events_per_cycle=dynamics['numspikes'],
cycle_events_isi=dynamics['spike_isi'],
prng=prng,
prng2=prng2)
# brute force remove non-zero times. Might result in fewer vals
# than desired
# values MUST be sorted for VecStim()!
if len(event_times) > 0:
event_times = event_times[np.logical_and(event_times > 0,
event_times <= tstop)]
event_times.sort()
event_times = event_times.tolist()
return event_times
def drive_event_times(drive_type, target_cell_type, params, gid, trial_idx=0):
"""External spike input times.
An external input drive to the network, i.e., one that is independent of
the spiking output of cells in the network.
Parameters
----------
drive_type : str
The drive type, which is one of
'extpois' : Poisson-distributed input to proximal dendrites
'extgauss' : Gaussian-distributed input to proximal dendrites
'evprox' : Proximal input at specified time (or Gaussian spread)
'evdist' : Distal input at specified time (or Gaussian spread)
'common' : As opposed to other drive types, these have timing that is
identical (synchronous) for all real cells in the network. Proximal
and distal dendrites have separate parameter sets, and need not be
synchronous. Note that not all cells classes (types) are required to
receive 'common' input---separate conductivity values can be assigned
to basket vs. pyramidal cells and AMPA vs. NMDA synapses
target_cell_type : str | None
The target cell type of the drive, e.g., 'L2_basket', 'L5_pyramidal',
etc., or None for 'common' inputs
params : dict
Parameters of the external input drive, arranged into a dictionary.
gid : int
The cell ID.
trial_idx : int
The index number of the current trial of a simulation.
Returns
-------
event_times : list
The event times at which spikes occur.
"""
prng, prng2 = _get_prng(
seed=params['prng_seedcore'] + trial_idx,
gid=gid,
sync_evinput=params.get('sync_evinput', False))
# check drive name validity, allowing substring matches ('evprox1' etc)
valid_drives = ['extpois', 'extgauss', 'common', 'evprox', 'evdist']
# NB check if drive_type has a valid substring, not vice versa
matches = [f for f in valid_drives if f in drive_type]
if len(matches) == 0:
raise ValueError('Invalid external drive: %s' % drive_type)
elif len(matches) > 1:
raise ValueError('Ambiguous external drive: %s' % drive_type)
# Return values not checked: False if all weights for given drive type
# are zero. Designed to be silent so that zeroing input weights
# effectively disables each.
target_syn_weights_zero = False
if target_cell_type in params:
target_syn_weights_zero = (params[target_cell_type][0] <= 0.0 and
params[target_cell_type][1] <= 0.0)
all_syn_weights_zero = True
for key in params.keys():
if key.startswith(('L2Pyr', 'L5Pyr', 'L2Bask', 'L5Bask')):
if params[key][0] > 0.0:
all_syn_weights_zero = False
event_times = list()
if drive_type == 'extpois' and not target_syn_weights_zero:
event_times = _create_extpois(
t0=params['t_interval'][0],
T=params['t_interval'][1],
# ind 3 is frequency (lamtha))
lamtha=params[target_cell_type][3],
prng=prng)
elif drive_type.startswith(('evprox', 'evdist')) and \
target_cell_type in params:
event_times = _create_gauss(
mu=params['t0'],
# ind 3 is sigma_t (stdev))
sigma=params[target_cell_type][3],
numspikes=int(params['numspikes']),
prng=prng)
elif drive_type == 'extgauss' and not target_syn_weights_zero:
event_times = _create_gauss(
mu=params[target_cell_type][3],
sigma=params[target_cell_type][4],
numspikes=50,
prng=prng)
elif drive_type == 'common' and not all_syn_weights_zero:
event_times = _create_bursty_input(
t0=params['t0'],
t0_stdev=params['t0_stdev'],
tstop=params['tstop'],
f_input=params['f_input'],
events_jitter_std=params['stdev'],
events_per_cycle=params['events_per_cycle'],
cycle_events_isi=10,
prng=prng,
prng2=prng2)
# brute force remove non-zero times. Might result in fewer vals
# than desired
# values MUST be sorted for VecStim()!
if len(event_times) > 0:
event_times = event_times[event_times > 0]
event_times.sort()
event_times = event_times.tolist()
return event_times
def _create_extpois(*, t0, T, lamtha, prng):
"""Create poisson inputs.
Parameters
----------
t0 : float
The start time (in ms).
T : float
The end time (in ms).
lamtha : float
The rate parameter for spike train (in Hz)
prng : instance of RandomState
The random state.
Returns
-------
event_times : array
The event times.
"""
# see: http://www.cns.nyu.edu/~david/handouts/poisson.pdf
if t0 < 0:
raise ValueError('The start time for Poisson inputs must be'
f'greater than 0. Got {t0}')
if T < t0:
raise ValueError('The end time for Poisson inputs must be'
f'greater than start time. Got ({t0}, {T})')
if lamtha <= 0.:
raise ValueError(f'Rate must be > 0. Got {lamtha}')
event_times = list()
t_gen = t0
while t_gen < T:
t_gen += prng.exponential(1. / lamtha) * 1000.
if t_gen < T:
event_times.append(t_gen)
return np.array(event_times)
def _create_gauss(*, mu, sigma, numspikes, prng):
"""Create gaussian inputs (used by extgauss and evoked).
Parameters
----------
mu : float
The mean time of spikes.
sigma : float
The standard deviation.
numspikes : float
The number of spikes.
prng : instance of RandomState
The random state.
Returns
-------
event_times : array
The event times.
"""
return prng.normal(mu, sigma, numspikes)
def _create_bursty_input(*, t0, t0_stdev, tstop, f_input,
events_jitter_std, events_per_cycle=2,
cycle_events_isi=10, prng, prng2):
"""Creates the bursty ongoing external inputs.
Used for, e.g., for rhythmic inputs in alpha/beta generation.
Parameters
----------
t0 : float
The start times. If -1, then randomize the start time
of inputs uniformly between 25 ms and 125 ms.
t0_stdev : float
If greater than 0 and t0 != -1, randomize start time
of inputs from a normal distribution with t0_stdev as standard
deviation.
tstop : float
The stop time.
f_input : float
The frequency of input bursts.
events_jitter_std : float
The standard deviation (in ms) of each burst event.
events_per_cycle : int
The events per cycle. This is the spikes/burst parameter in the GUI.
Default: 2 (doublet)
cycle_events_isi : float
Time between spike events within a cycle (ISI). Default: 10 ms
prng : instance of RandomState
The random state.
prng2 : instance of RandomState
The random state used for jitter to start time (see t0_stdev).
Returns
-------
event_times : array
The event times.
"""
if t0_stdev > 0.0:
t0 = prng2.normal(t0, t0_stdev)
burst_period = 1000. / f_input
burst_duration = (events_per_cycle - 1) * cycle_events_isi
if burst_duration > burst_period:
raise ValueError(f'Burst duration ({burst_duration}s) cannot'
f' be greater than burst period ({burst_period}s)'
'Consider increasing the spike ISI or burst rate')
# array of mean stimulus times, starts at t0
isi_array = np.arange(t0, tstop, burst_period)
# array of single stimulus times -- no doublets
t_array = prng.normal(isi_array, events_jitter_std)
if events_per_cycle > 1:
cycle = (np.arange(events_per_cycle) - (events_per_cycle - 1) / 2)
t_array = np.ravel([t_array + cycle_events_isi * cyc for cyc in cycle])
return t_array