[074d3d]: / mne / forward / _lead_dots.py

Download this file

611 lines (550 with data), 19.4 kB

  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
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
# Authors: The MNE-Python contributors.
# License: BSD-3-Clause
# Copyright the MNE-Python contributors.
# The computations in this code were primarily derived from Matti Hämäläinen's
# C code.
import os
import os.path as op
import numpy as np
from numpy.polynomial import legendre
from ..parallel import parallel_func
from ..utils import _get_extra_data_path, fill_doc, logger, verbose
##############################################################################
# FAST LEGENDRE (DERIVATIVE) POLYNOMIALS USING LOOKUP TABLE
def _next_legen_der(n, x, p0, p01, p0d, p0dd):
"""Compute the next Legendre polynomial and its derivatives."""
# only good for n > 1 !
old_p0 = p0
old_p0d = p0d
p0 = ((2 * n - 1) * x * old_p0 - (n - 1) * p01) / n
p0d = n * old_p0 + x * old_p0d
p0dd = (n + 1) * old_p0d + x * p0dd
return p0, p0d, p0dd
def _get_legen(x, n_coeff=100):
"""Get Legendre polynomials expanded about x."""
return legendre.legvander(x, n_coeff - 1)
def _get_legen_der(xx, n_coeff=100):
"""Get Legendre polynomial derivatives expanded about x."""
coeffs = np.empty((len(xx), n_coeff, 3))
for c, x in zip(coeffs, xx):
p0s, p0ds, p0dds = c[:, 0], c[:, 1], c[:, 2]
p0s[:2] = [1.0, x]
p0ds[:2] = [0.0, 1.0]
p0dds[:2] = [0.0, 0.0]
for n in range(2, n_coeff):
p0s[n], p0ds[n], p0dds[n] = _next_legen_der(
n, x, p0s[n - 1], p0s[n - 2], p0ds[n - 1], p0dds[n - 1]
)
return coeffs
@verbose
def _get_legen_table(
ch_type,
volume_integral=False,
n_coeff=100,
n_interp=20000,
force_calc=False,
verbose=None,
):
"""Return a (generated) LUT of Legendre (derivative) polynomial coeffs."""
if n_interp % 2 != 0:
raise RuntimeError("n_interp must be even")
fname = op.join(_get_extra_data_path(), "tables")
if not op.isdir(fname):
# Updated due to API change (GH 1167)
os.makedirs(fname)
if ch_type == "meg":
fname = op.join(fname, f"legder_{n_coeff}_{n_interp}.bin")
leg_fun = _get_legen_der
extra_str = " derivative"
lut_shape = (n_interp + 1, n_coeff, 3)
else: # 'eeg'
fname = op.join(fname, f"legval_{n_coeff}_{n_interp}.bin")
leg_fun = _get_legen
extra_str = ""
lut_shape = (n_interp + 1, n_coeff)
if not op.isfile(fname) or force_calc:
logger.info(f"Generating Legendre{extra_str} table...")
x_interp = np.linspace(-1, 1, n_interp + 1)
lut = leg_fun(x_interp, n_coeff).astype(np.float32)
if not force_calc:
with open(fname, "wb") as fid:
fid.write(lut.tobytes())
else:
logger.info(f"Reading Legendre{extra_str} table...")
with open(fname, "rb", buffering=0) as fid:
lut = np.fromfile(fid, np.float32)
lut.shape = lut_shape
# we need this for the integration step
n_fact = np.arange(1, n_coeff, dtype=float)
if ch_type == "meg":
n_facts = list() # multn, then mult, then multn * (n + 1)
if volume_integral:
n_facts.append(n_fact / ((2.0 * n_fact + 1.0) * (2.0 * n_fact + 3.0)))
else:
n_facts.append(n_fact / (2.0 * n_fact + 1.0))
n_facts.append(n_facts[0] / (n_fact + 1.0))
n_facts.append(n_facts[0] * (n_fact + 1.0))
# skip the first set of coefficients because they are not used
lut = lut[:, 1:, [0, 1, 1, 2]] # for multiplicative convenience later
# reshape this for convenience, too
n_facts = np.array(n_facts)[[2, 0, 1, 1], :].T
n_facts = np.ascontiguousarray(n_facts)
n_fact = n_facts
else: # 'eeg'
n_fact = (2.0 * n_fact + 1.0) * (2.0 * n_fact + 1.0) / n_fact
# skip the first set of coefficients because they are not used
lut = lut[:, 1:].copy()
return lut, n_fact
def _comp_sum_eeg(beta, ctheta, lut_fun, n_fact):
"""Lead field dot products using Legendre polynomial (P_n) series."""
# Compute the sum occurring in the evaluation.
# The result is
# sums[:] (2n+1)^2/n beta^n P_n
n_chunk = 50000000 // (8 * max(n_fact.shape) * 2)
lims = np.concatenate([np.arange(0, beta.size, n_chunk), [beta.size]])
s0 = np.empty(beta.shape)
for start, stop in zip(lims[:-1], lims[1:]):
coeffs = lut_fun(ctheta[start:stop])
betans = np.tile(beta[start:stop][:, np.newaxis], (1, n_fact.shape[0]))
np.cumprod(betans, axis=1, out=betans) # run inplace
coeffs *= betans
s0[start:stop] = np.dot(coeffs, n_fact) # == weighted sum across cols
return s0
def _comp_sums_meg(beta, ctheta, lut_fun, n_fact, volume_integral):
"""Lead field dot products using Legendre polynomial (P_n) series.
Parameters
----------
beta : array, shape (n_points * n_points, 1)
Coefficients of the integration.
ctheta : array, shape (n_points * n_points, 1)
Cosine of the angle between the sensor integration points.
lut_fun : callable
Look-up table for evaluating Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
volume_integral : bool
If True, compute volume integral.
Returns
-------
sums : array, shape (4, n_points * n_points)
The results.
"""
# Compute the sums occurring in the evaluation.
# Two point magnetometers on the xz plane are assumed.
# The four sums are:
# * sums[:, 0] n(n+1)/(2n+1) beta^(n+1) P_n
# * sums[:, 1] n/(2n+1) beta^(n+1) P_n'
# * sums[:, 2] n/((2n+1)(n+1)) beta^(n+1) P_n'
# * sums[:, 3] n/((2n+1)(n+1)) beta^(n+1) P_n''
# This is equivalent, but slower:
# sums = np.sum(bbeta[:, :, np.newaxis].T * n_fact * coeffs, axis=1)
# sums = np.rollaxis(sums, 2)
# or
# sums = np.einsum('ji,jk,ijk->ki', bbeta, n_fact, lut_fun(ctheta)))
sums = np.empty((n_fact.shape[1], len(beta)))
# beta can be e.g. 3 million elements, which ends up using lots of memory
# so we split up the computations into ~50 MB blocks
n_chunk = 50000000 // (8 * max(n_fact.shape) * 2)
lims = np.concatenate([np.arange(0, beta.size, n_chunk), [beta.size]])
for start, stop in zip(lims[:-1], lims[1:]):
bbeta = np.tile(beta[start:stop][np.newaxis], (n_fact.shape[0], 1))
bbeta[0] *= beta[start:stop]
np.cumprod(bbeta, axis=0, out=bbeta) # run inplace
np.einsum(
"ji,jk,ijk->ki",
bbeta,
n_fact,
lut_fun(ctheta[start:stop]),
out=sums[:, start:stop],
)
return sums
###############################################################################
# SPHERE DOTS
_meg_const = 4e-14 * np.pi # This is \mu_0^2/4\pi
_eeg_const = 1.0 / (4.0 * np.pi)
def _fast_sphere_dot_r0(
r,
rr1_orig,
rr2s,
lr1,
lr2s,
cosmags1,
cosmags2s,
w1,
w2s,
volume_integral,
lut,
n_fact,
ch_type,
):
"""Lead field dot product computation for M/EEG in the sphere model.
Parameters
----------
r : float
The integration radius. It is used to calculate beta as:
beta = (r * r) / (lr1 * lr2).
rr1 : array, shape (n_points x 3)
Normalized position vectors of integrations points in first sensor.
rr2s : list
Normalized position vector of integration points in second sensor.
lr1 : array, shape (n_points x 1)
Magnitude of position vector of integration points in first sensor.
lr2s : list
Magnitude of position vector of integration points in second sensor.
cosmags1 : array, shape (n_points x 1)
Direction of integration points in first sensor.
cosmags2s : list
Direction of integration points in second sensor.
w1 : array, shape (n_points x 1) | None
Weights of integration points in the first sensor.
w2s : list
Weights of integration points in the second sensor.
volume_integral : bool
If True, compute volume integral.
lut : callable
Look-up table for evaluating Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
ch_type : str
The channel type. It can be 'meg' or 'eeg'.
Returns
-------
result : float
The integration sum.
"""
if w1 is None: # operating on surface, treat independently
out_shape = (len(rr2s), len(rr1_orig))
sum_axis = 1 # operate along second axis only at the end
else:
out_shape = (len(rr2s),)
sum_axis = None # operate on flattened array at the end
out = np.empty(out_shape)
rr2 = np.concatenate(rr2s)
lr2 = np.concatenate(lr2s)
cosmags2 = np.concatenate(cosmags2s)
# outer product, sum over coords
ct = np.einsum("ik,jk->ij", rr1_orig, rr2)
np.clip(ct, -1, 1, ct)
# expand axes
rr1 = rr1_orig[:, np.newaxis, :] # (n_rr1, n_rr2, n_coord) e.g. 4x4x3
rr2 = rr2[np.newaxis, :, :]
lr1lr2 = lr1[:, np.newaxis] * lr2[np.newaxis, :]
beta = (r * r) / lr1lr2
if ch_type == "meg":
sums = _comp_sums_meg(
beta.flatten(), ct.flatten(), lut, n_fact, volume_integral
)
sums.shape = (4,) + beta.shape
# Accumulate the result, a little bit streamlined version
# cosmags1 = cosmags1[:, np.newaxis, :]
# cosmags2 = cosmags2[np.newaxis, :, :]
# n1c1 = np.sum(cosmags1 * rr1, axis=2)
# n1c2 = np.sum(cosmags1 * rr2, axis=2)
# n2c1 = np.sum(cosmags2 * rr1, axis=2)
# n2c2 = np.sum(cosmags2 * rr2, axis=2)
# n1n2 = np.sum(cosmags1 * cosmags2, axis=2)
n1c1 = np.einsum("ik,ijk->ij", cosmags1, rr1)
n1c2 = np.einsum("ik,ijk->ij", cosmags1, rr2)
n2c1 = np.einsum("jk,ijk->ij", cosmags2, rr1)
n2c2 = np.einsum("jk,ijk->ij", cosmags2, rr2)
n1n2 = np.einsum("ik,jk->ij", cosmags1, cosmags2)
part1 = ct * n1c1 * n2c2
part2 = n1c1 * n2c1 + n1c2 * n2c2
result = (
n1c1 * n2c2 * sums[0]
+ (2.0 * part1 - part2) * sums[1]
+ (n1n2 + part1 - part2) * sums[2]
+ (n1c2 - ct * n1c1) * (n2c1 - ct * n2c2) * sums[3]
)
# Give it a finishing touch!
result *= _meg_const / lr1lr2
if volume_integral:
result *= r
else: # 'eeg'
result = _comp_sum_eeg(beta.flatten(), ct.flatten(), lut, n_fact)
result.shape = beta.shape
# Give it a finishing touch!
result *= _eeg_const
result /= lr1lr2
# now we add them all up with weights
offset = 0
result *= np.concatenate(w2s)
if w1 is not None:
result *= w1[:, np.newaxis]
for ii, w2 in enumerate(w2s):
out[ii] = np.sum(result[:, offset : offset + len(w2)], axis=sum_axis)
offset += len(w2)
return out
@fill_doc
def _do_self_dots(intrad, volume, coils, r0, ch_type, lut, n_fact, n_jobs):
"""Perform the lead field dot product integrations.
Parameters
----------
intrad : float
The integration radius. It is used to calculate beta as:
beta = (intrad * intrad) / (r1 * r2).
volume : bool
If True, perform volume integral.
coils : list of dict
The coils.
r0 : array, shape (3 x 1)
The origin of the sphere.
ch_type : str
The channel type. It can be 'meg' or 'eeg'.
lut : callable
Look-up table for evaluating Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
%(n_jobs)s
Returns
-------
products : array, shape (n_coils, n_coils)
The integration products.
"""
if ch_type == "eeg":
intrad = intrad * 0.7
# convert to normalized distances from expansion center
rmags = [coil["rmag"] - r0[np.newaxis, :] for coil in coils]
rlens = [np.sqrt(np.sum(r * r, axis=1)) for r in rmags]
rmags = [r / rl[:, np.newaxis] for r, rl in zip(rmags, rlens)]
cosmags = [coil["cosmag"] for coil in coils]
ws = [coil["w"] for coil in coils]
parallel, p_fun, n_jobs = parallel_func(_do_self_dots_subset, n_jobs)
prods = parallel(
p_fun(intrad, rmags, rlens, cosmags, ws, volume, lut, n_fact, ch_type, idx)
for idx in np.array_split(np.arange(len(rmags)), n_jobs)
)
products = np.sum(prods, axis=0)
return products
def _do_self_dots_subset(
intrad, rmags, rlens, cosmags, ws, volume, lut, n_fact, ch_type, idx
):
"""Parallelize."""
# all possible combinations of two magnetometers
products = np.zeros((len(rmags), len(rmags)))
for ci1 in idx:
ci2 = ci1 + 1
res = _fast_sphere_dot_r0(
intrad,
rmags[ci1],
rmags[:ci2],
rlens[ci1],
rlens[:ci2],
cosmags[ci1],
cosmags[:ci2],
ws[ci1],
ws[:ci2],
volume,
lut,
n_fact,
ch_type,
)
products[ci1, :ci2] = res
products[:ci2, ci1] = res
return products
def _do_cross_dots(intrad, volume, coils1, coils2, r0, ch_type, lut, n_fact):
"""Compute lead field dot product integrations between two coil sets.
The code is a direct translation of MNE-C code found in
`mne_map_data/lead_dots.c`.
Parameters
----------
intrad : float
The integration radius. It is used to calculate beta as:
beta = (intrad * intrad) / (r1 * r2).
volume : bool
If True, compute volume integral.
coils1 : list of dict
The original coils.
coils2 : list of dict
The coils to which data is being mapped.
r0 : array, shape (3 x 1).
The origin of the sphere.
ch_type : str
The channel type. It can be 'meg' or 'eeg'
lut : callable
Look-up table for evaluating Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
Returns
-------
products : array, shape (n_coils, n_coils)
The integration products.
"""
if ch_type == "eeg":
intrad = intrad * 0.7
rmags1 = [coil["rmag"] - r0[np.newaxis, :] for coil in coils1]
rmags2 = [coil["rmag"] - r0[np.newaxis, :] for coil in coils2]
rlens1 = [np.sqrt(np.sum(r * r, axis=1)) for r in rmags1]
rlens2 = [np.sqrt(np.sum(r * r, axis=1)) for r in rmags2]
rmags1 = [r / rl[:, np.newaxis] for r, rl in zip(rmags1, rlens1)]
rmags2 = [r / rl[:, np.newaxis] for r, rl in zip(rmags2, rlens2)]
ws1 = [coil["w"] for coil in coils1]
ws2 = [coil["w"] for coil in coils2]
cosmags1 = [coil["cosmag"] for coil in coils1]
cosmags2 = [coil["cosmag"] for coil in coils2]
products = np.zeros((len(rmags1), len(rmags2)))
for ci1 in range(len(coils1)):
res = _fast_sphere_dot_r0(
intrad,
rmags1[ci1],
rmags2,
rlens1[ci1],
rlens2,
cosmags1[ci1],
cosmags2,
ws1[ci1],
ws2,
volume,
lut,
n_fact,
ch_type,
)
products[ci1, :] = res
return products
@fill_doc
def _do_surface_dots(
intrad, volume, coils, surf, sel, r0, ch_type, lut, n_fact, n_jobs
):
"""Compute the map construction products.
Parameters
----------
intrad : float
The integration radius. It is used to calculate beta as:
beta = (intrad * intrad) / (r1 * r2)
volume : bool
If True, compute a volume integral.
coils : list of dict
The coils.
surf : dict
The surface on which the field is interpolated.
sel : array
Indices of the surface vertices to select.
r0 : array, shape (3 x 1)
The origin of the sphere.
ch_type : str
The channel type. It can be 'meg' or 'eeg'.
lut : callable
Look-up table for Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
%(n_jobs)s
Returns
-------
products : array, shape (n_coils, n_coils)
The integration products.
"""
# convert to normalized distances from expansion center
rmags = [coil["rmag"] - r0[np.newaxis, :] for coil in coils]
rlens = [np.sqrt(np.sum(r * r, axis=1)) for r in rmags]
rmags = [r / rl[:, np.newaxis] for r, rl in zip(rmags, rlens)]
cosmags = [coil["cosmag"] for coil in coils]
ws = [coil["w"] for coil in coils]
rref = None
refl = None
# virt_ref = False
if ch_type == "eeg":
intrad = intrad * 0.7
# The virtual ref code is untested and unused, so it is
# commented out for now
# if virt_ref:
# rref = virt_ref[np.newaxis, :] - r0[np.newaxis, :]
# refl = np.sqrt(np.sum(rref * rref, axis=1))
# rref /= refl[:, np.newaxis]
rsurf = surf["rr"][sel] - r0[np.newaxis, :]
lsurf = np.sqrt(np.sum(rsurf * rsurf, axis=1))
rsurf /= lsurf[:, np.newaxis]
this_nn = surf["nn"][sel]
# loop over the coils
parallel, p_fun, n_jobs = parallel_func(_do_surface_dots_subset, n_jobs)
prods = parallel(
p_fun(
intrad,
rsurf,
rmags,
rref,
refl,
lsurf,
rlens,
this_nn,
cosmags,
ws,
volume,
lut,
n_fact,
ch_type,
idx,
)
for idx in np.array_split(np.arange(len(rmags)), n_jobs)
)
products = np.sum(prods, axis=0)
return products
def _do_surface_dots_subset(
intrad,
rsurf,
rmags,
rref,
refl,
lsurf,
rlens,
this_nn,
cosmags,
ws,
volume,
lut,
n_fact,
ch_type,
idx,
):
"""Parallelize.
Parameters
----------
refl : array | None
If ch_type is 'eeg', the magnitude of position vector of the
virtual reference (never used).
lsurf : array
Magnitude of position vector of the surface points.
rlens : list of arrays of length n_coils
Magnitude of position vector.
this_nn : array, shape (n_vertices, 3)
Surface normals.
cosmags : list of array.
Direction of the integration points in the coils.
ws : list of array
Integration weights of the coils.
volume : bool
If True, compute volume integral.
lut : callable
Look-up table for evaluating Legendre polynomials.
n_fact : array
Coefficients in the integration sum.
ch_type : str
'meg' or 'eeg'
idx : array, shape (n_coils x 1)
Index of coil.
Returns
-------
products : array, shape (n_coils, n_coils)
The integration products.
"""
products = _fast_sphere_dot_r0(
intrad,
rsurf,
rmags,
lsurf,
rlens,
this_nn,
cosmags,
None,
ws,
volume,
lut,
n_fact,
ch_type,
).T
if rref is not None:
raise NotImplementedError # we don't ever use this, isn't tested
# vres = _fast_sphere_dot_r0(
# intrad, rref, rmags, refl, rlens, this_nn, cosmags, None, ws,
# volume, lut, n_fact, ch_type)
# products -= vres
return products