File: AutoGrid.R

package info (click to toggle)
r-bioc-delayedarray 0.24.0%2Bdfsg-1
  • links: PTS, VCS
  • area: main
  • in suites: bookworm
  • size: 1,480 kB
  • sloc: ansic: 727; makefile: 2
file content (511 lines) | stat: -rw-r--r-- 20,243 bytes parent folder | download
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
### =========================================================================
### Automatic grids
### -------------------------------------------------------------------------


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### defaultAutoGrid()
###

### Guaranteed to return an integer >= 1.
.normarg_block.length <- function(block.length, type)
{
    if (is.null(block.length))
        return(getAutoBlockLength(type))
    if (!isSingleNumber(block.length))
        stop(wmsg("'block.length' must be a single integer or NULL"))
    if (block.length < 1)
        stop(wmsg("'block.length' must be >= 1"))
    if (block.length > .Machine$integer.max)
        stop(wmsg("'block.length' is too big. Blocks of ",
                  "length > .Machine$integer.max are not supported yet. ",
                  "Please specify a smaller 'block.length'."))
    as.integer(block.length)
}

.normarg_chunk.grid <- function(chunk.grid, x, what="'x'")
{
    if (is.null(chunk.grid))
        return(chunkGrid(x))
    if (!is(chunk.grid, "ArrayGrid"))
        stop(wmsg("'chunk.grid' must be an ArrayGrid object or NULL"))
    if (!identical(refdim(chunk.grid), dim(x)))
        stop(wmsg("'chunk.grid' is incompatible with ", what))
    chunk.grid
}

.normarg_block.shape <- function(block.shape)
{
    if (is.null(block.shape))
        return(getAutoBlockShape())
    if (!(isSingleString(block.shape) &&
          block.shape %in% SUPPORTED_BLOCK_SHAPES))
    {
        in1string <- paste(paste0("\"", SUPPORTED_BLOCK_SHAPES, "\""),
                           collapse=", ")
        stop(wmsg("'block.shape' must be one of ", in1string, ", or NULL"))
    }
    block.shape
}

.block_grid <- function(x_dim, block_len, chunk_grid, block_shape)
{
    ## If 'x' is empty, we return a grid with a single (empty) block that
    ## has the dimensions of 'x'.
    if (any(x_dim == 0L))
        return(RegularArrayGrid(x_dim))
    if (is.null(chunk_grid)) {
        ans <- makeRegularArrayGridOfCappedLengthViewports(x_dim,
                                                           block_len,
                                                           block_shape)
        return(ans)
    }
    chunks_per_block <- max(block_len %/% maxlength(chunk_grid), 1L)
    ratio <- makeCappedVolumeBox(chunks_per_block, dim(chunk_grid), block_shape)
    downsample(chunk_grid, ratio)
}

### Return a grid that is "optimal" for block processing of array-like
### object 'x'.
### The grid is returned as an ArrayGrid object on reference array 'x'.
### The grid elements define the blocks that will be used for processing 'x'
### by block. The grid is "optimal" in the sense that:
###  - It's "compatible" with the chunk grid (i.e. with 'chunkGrid(x)' or
###    with the chunk grid supplied via the 'chunk.grid' argument), that is,
###    the chunks are contained in the blocks. In other words, chunks never
###    cross block boundaries.
###  - Its "resolution" is such that the blocks have a length that is as
###    close as possibe to (but does not exceed) 'block.length'.
###    An exception is when some chunks are already >= 'block.length',
###    in which case the returned grid is the same as the chunk grid.
### Note that the returned grid is regular (i.e. RegularArrayGrid object)
### unless the chunk grid is not regular (i.e. is an ArbitraryArrayGrid
### object).
defaultAutoGrid <-
    function(x, block.length=NULL, chunk.grid=NULL, block.shape=NULL)
{
    x_dim <- dim(x)
    if (is.null(x_dim))
        stop(wmsg("'x' must be an array-like object"))
    block_len <- .normarg_block.length(block.length, type(x))
    chunk_grid <- .normarg_chunk.grid(chunk.grid, x)
    block_shape <- .normarg_block.shape(block.shape)
    .block_grid(x_dim, block_len, chunk_grid, block_shape)
}

blockGrid <- function(...)
{
    .Defunct("defaultAutoGrid")
    defaultAutoGrid(...)
}


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### rowAutoGrid() and colAutoGrid()
###

.get_auto_nrow <- function(x_dim, block.length, x_type)
{
    x_nrow <- x_dim[[1L]]
    x_ncol <- x_dim[[2L]]
    block_len <- .normarg_block.length(block.length, x_type)
    nrow <- block_len %/% x_ncol

    if (is.na(nrow)) # NA if x_ncol=0.
        return(x_nrow)
    if (nrow < 1L)
        nrow <- 1L
    min(x_nrow, nrow)
}

### Return a RegularArrayGrid object describing a grid on matrix-like
### object 'x' where the grid elements are blocks made of full rows of 'x'.
rowAutoGrid <- function(x, nrow=NULL, block.length=NULL)
{
    x_dim <- dim(x)
    if (length(x_dim) != 2L)
        stop(wmsg("'x' must have exactly 2 dimensions"))
    x_nrow <- x_dim[[1L]]
    x_ncol <- x_dim[[2L]]
    if (is.null(nrow)) {
        nrow <- .get_auto_nrow(x_dim, block.length, type(x))
        spacings <- c(nrow, x_ncol)
    } else {
        if (!is.null(block.length))
            warning("'block.length' is ignored when 'nrow' is not NULL")
        if (!isSingleNumber(nrow))
            stop(wmsg("'nrow' must be a single integer or NULL"))
        nrow <- as.integer(nrow)
        if (nrow < 1L || nrow > x_nrow)
            stop(wmsg("'nrow' must be >= 1 and <= nrow(x)"))
        spacings <- c(nrow, x_ncol)
        if (prod(spacings) > .Machine$integer.max)
            stop(wmsg("'nrow' is too big. Blocks of length > ",
                      ".Machine$integer.max are not supported yet. ",
                      "Please specify a smaller 'nrow'."))
    }
    RegularArrayGrid(x_dim, spacings)
}

rowGrid <- function(...)
{
    .Defunct("rowAutoGrid")
    rowAutoGrid(...)
}

### Return a RegularArrayGrid object describing a grid on matrix-like
### object 'x' where the grid elements are blocks made of full columns of 'x'.
colAutoGrid <- function(x, ncol=NULL, block.length=NULL)
{
    x_dim <- dim(x)
    if (length(x_dim) != 2L)
        stop(wmsg("'x' must have exactly 2 dimensions"))
    x_nrow <- x_dim[[1L]]
    x_ncol <- x_dim[[2L]]
    if (is.null(ncol)) {
        ncol <- .get_auto_nrow(rev(x_dim), block.length, type(x))
        spacings <- c(x_nrow, ncol)
    } else {
        if (!is.null(block.length))
            warning("'block.length' is ignored when 'ncol' is not NULL")
        if (!isSingleNumber(ncol))
            stop(wmsg("'ncol' must be a single integer or NULL"))
        ncol <- as.integer(ncol)
        if (ncol < 1L || ncol > x_ncol)
            stop(wmsg("'ncol' must be >= 1 and <= ncol(x)"))
        spacings <- c(x_nrow, ncol)
        if (prod(spacings) > .Machine$integer.max)
            stop(wmsg("'ncol' is too big. Blocks of length > ",
                      ".Machine$integer.max are not supported yet. ",
                      "Please specify a smaller 'ncol'."))
    }
    RegularArrayGrid(x_dim, spacings)
}

colGrid <- function(...)
{
    .Defunct("colAutoGrid")
    colAutoGrid(...)
}


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### set/getAutoGridMaker()
###

### We set the automatic grid maker to defaultAutoGrid() by default.
setAutoGridMaker <- function(GRIDMAKER="defaultAutoGrid")
{
    match.fun(GRIDMAKER)  # sanity check
    set_user_option("auto.grid.maker", GRIDMAKER)
}

getAutoGridMaker <- function() get_user_option("auto.grid.maker")


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### normarg_grid()
###

normarg_grid <- function(grid, x)
{
    if (is.null(grid)) {
        etc <- c("Please use setAutoGridMaker() ",
                 "to set a valid automatic grid maker.")
        GRIDMAKER <- match.fun(getAutoGridMaker())
        grid <- try(GRIDMAKER(x), silent=TRUE)
        if (is(grid, "try-error"))
            stop(wmsg("The current automatic grid maker returned an ",
                      "error when called on 'x'. ", etc))
        if (!is(grid, "ArrayGrid"))
            stop(wmsg("The current automatic grid maker didn't return an ",
                      "ArrayGrid object. ", etc))
        if (!identical(refdim(grid), dim(x)))
            stop(wmsg("The current automatic grid maker returned a grid ",
                      "that is incompatible with 'x'. ", etc))
    } else {
        if (!is(grid, "ArrayGrid"))
            stop(wmsg("'grid' must be NULL or an ArrayGrid object"))
        if (!identical(refdim(grid), dim(x)))
            stop(wmsg("'grid' is incompatible with 'x'"))
    }
    grid
}


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### defaultSinkAutoGrid()
###
### To create a grid that is "optimal" for use in sinkApply().
###

defaultSinkAutoGrid <- function(sink, block.length=NULL, chunk.grid=NULL)
{
    sink_dim <- dim(sink)
    if (is.null(sink_dim))
        stop(wmsg("'sink_dim' must be an array-like object"))
    block_len <- .normarg_block.length(block.length, type(sink))
    chunk_grid <- .normarg_chunk.grid(chunk.grid, sink, "'sink'")
    .block_grid(sink_dim, block_len, chunk_grid, "first-dim-grows-first")
}


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### normarg_sink_grid()
###
### For use in sinkApply().
###
### Like normarg_grid() except that it calls 'defaultSinkAutoGrid(sink)'
### instead of:
###     GRIDMAKER <- match.fun(getAutoGridMaker())
###     grid <- GRIDMAKER(sink)
### to create the grid when it's not supplied.
###

normarg_sink_grid <- function(grid, sink)
{
    if (is.null(grid)) {
        grid <- defaultSinkAutoGrid(sink)
    } else {
        if (!is(grid, "ArrayGrid"))
            stop(wmsg("'grid' must be NULL or an ArrayGrid object"))
        if (!identical(refdim(grid), dim(sink)))
            stop(wmsg("'grid' is incompatible with 'sink'"))
    }
    grid
}


### - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
### defaultMultAutoGrids()
###

### Performs the same checks as .check_chunkdim() (see chunkGrid.R).
.normarg_z_chunkdim <- function(z_chunkdim, z_dim,
                                what="z", nrow_what="x", ncol_what="y")
{
    if (!is.numeric(z_chunkdim))
        stop(wmsg("'", what, ".chunkdim' must be NULL or an integer vector"))
    if (!is.integer(z_chunkdim))
        z_chunkdim <- as.integer(z_chunkdim)
    if (length(z_chunkdim) != 2L)
        stop(wmsg("'", what, ".chunkdim' must be of length 2"))
    if (S4Vectors:::anyMissingOrOutside(z_chunkdim, 0L))
        stop(wmsg("'", what, ".chunkdim' should not contain negative ",
                  "or NA values"))
    if (z_chunkdim[[1L]] > z_dim[[1L]])
        stop(wmsg("'", what, ".chunkdim[[1L]]' must be ",
                  "<= 'nrow(", nrow_what, ")'"))
    if (z_chunkdim[[1L]] == 0L && z_dim[[1L]] != 0L)
        stop(wmsg("'", what, ".chunkdim[[1L]]' should not be 0 ",
                  "unless 'nrow(", nrow_what, ")' is also 0"))
    if (z_chunkdim[[2L]] > z_dim[[2L]])
        stop(wmsg("'", what, ".chunkdim[[2L]]' must be ",
                  "<= 'ncol(", ncol_what, ")'"))
    if (z_chunkdim[[2L]] == 0L && z_dim[[2L]] != 0L)
        stop(wmsg("'", what, ".chunkdim[[2L]]' should not be 0 ",
                  "unless 'ncol(", ncol_what, ")' is also 0"))
    if (prod(z_chunkdim) > .Machine$integer.max)
        stop(wmsg("The chunk dimensions specified in 'z.chunkdim' are too ",
                  "big. Their product must be <= .Machine$integer.max"))
    z_chunkdim
}

### Define grids for "block matrix multiplication".
###
### The problem
### -----------
### 'x' and 'y' being matrix-like objects that can be multiplied (i.e.
### ncol(x) == nrow(y)), and 'z' being their product (i.e. z = x %*% y),
### defaultMultAutoGrids() tries to solve the problem of finding 3 grids
### (on 'x', 'y', and 'z', respectively) that can be used for "block matrix
### multiplication" of 'x' by 'y'.
### The grid triplet ('x_grid', 'y_grid', 'z_grid') must satisfy:
###   - get_spacings_along(x_grid, 1L)' and 'get_spacings_along(z_grid, 1L)'
###   - get_spacings_along(y_grid, 2L)' and 'get_spacings_along(z_grid, 2L)'
###   - get_spacings_along(x_grid, 2L)' and 'get_spacings_along(y_grid, 1L)'
###
### Many grid triplets satisfy these constraints, from the triplet made of
### the coarsest grids (i.e. each grid is made of a single block covering
### the full matrix e.g. 'x_grid <- RegularArrayGrid(dim(x))') to the triplet
### made of the finest grids (i.e. each grid is made of blocks that contain a
### single matrix element e.g. 'x_grid <- RegularArrayGrid(dim(x), c(1, 1))').
###
### Between these 2 extremes, which would both lead to bad performance in
### general, is a wide range solutions that cover the entire spectrum of
### coarseness. Here is a notable solution with intermediate coarseness:
###   - x_grid <- rowAutoGrid(x, nrow=1)  # blocks on 'x' are single rows
###   - y_grid <- colAutoGrid(y, ncol=1)  # blocks on 'y' are single columns
###   - z_grid <- RegularArrayGrid(dim(z), c(1, 1))
###
### We could also let rowAutoGrid() and colAutoGrid() automatically choose
### the maximum number of rows and columns to put in a block based on the
### "automatic block size" (see ?getAutoBlockSize) by not specifying the
### 'nrow' or 'ncol' arguments:
###  - x_grid <- rowAutoGrid(x)
###  - y_grid <- colAutoGrid(y)
###  - z_grid <- RegularArrayGrid(dim(z), c(a, c))
###    where 'a' is 'nrow(x_grid[[1L]])' and 'c' is 'ncol(y_grid[[1L]])'
###
### Grids that will perform "well"
### ------------------------------
### Amongst all the possible grid triplets, we want to pick up one that will
### perform "well". More precisely:
###
###   (1) In order to cap memory usage during multiplication of any 2 blocks
###       (e.g. 'x_grid[[i, k]]' and 'y_grid[[k, j]]'), we decide (somewhat
###       arbitrarily) that their cumulated length must be as close as possibe
###       to (but must not exceed) half of 'block.length'. The length of the
###       block corresponding to the product ('z_grid[[i, j]]') must also be
###       as close as possibe (but must not exceed) half of 'block.length'.
###
###   (2) We want to minimize the number of times a given block from 'x'
###       and/or 'y' will be loaded in memory. Note that the total number
###       of block multiplications will be 'K * length(z_grid)' where 'K'
###       is 'ncol(x_grid)' (or 'nrow(y_grid)', which is the same). Each
###       block in 'x' will participate to 'ncol(z_grid)' multiplications
###       and each block in 'y' will participate to 'nrow(z_grid)'
###       multiplications. So assuming that blocks are loaded each time
###       they need to participate to a multiplication and not kept in
###       memory, the data in 'x' will be loaded 'ncol(z_grid)' times and
###       the data in 'y' will be loaded 'nrow(z_grid)' times. This means
###       that we want 'z_grid' to be as coarse as possible. Also,
###       depending on which of 'x' or 'y' is bigger, we should try to
###       reduce 'ncol(z_grid)' (i.e. increase coarseness along 'z' 2nd
###       dimension) which would be to the detriment of 'nrow(z_grid)', or
###       vice-versa.
###
###   (3) Ideally we'd want the 3 grids to be compatible with the underlying
###       chunk geometry of the matrix-like object that they are on.
###       Unfortunately this is in general impossible to achieve for the
###       3 grids at the same time (the 3 coarsest grids would achieve this
###       but to the detriment of memory usage) so we'll have to compromise.
###       One thing we don't want to compromise  with though is compatibility
###       of 'z_grid' with the underlying chunk geometry of 'z'. The reason
###       this is that in practice 'z' will typically be a RealizationSink
###       object where the output blocks will be written with write_block()
###       and some realization backends like RleRealizationSink or
###       TENxRealizationSink require write_block() to respect the chunk
###       geometry.
###
### Simplifying the problem
### -----------------------
### In order to simplify the problem we restrict the search space to
### **regular** grids (RegularArrayGrid objects), so no arbitrary grids
### (ArbitraryArrayGrid) will be considered. This reduces the opportunity to
### find the best possible grids but that's ok for now.
### Thanks to this simplification, our problem is reduced to finding 3
### non-negative integer numbers a, b, c from which the the 3 grids will
### be defined as follow:
###  - x_grid <- RegularArrayGrid(dim(x), c(a, b))
###  - y_grid <- RegularArrayGrid(dim(x), c(b, c))
###  - z_grid <- RegularArrayGrid(dim(z), c(a, c))
###

.find_abc <- function(A, B, C, half_block_len,
                      x_chunkdim, y_chunkdim, z_chunkdim)
{
    is_a_multiple_of <- function(m, n) { m == 0L || m %% n == 0L }
    a0 <- z_chunkdim[[1L]]
    if (!is_a_multiple_of(a0, x_chunkdim[[1L]]))
        warning(wmsg("The chunk geometry for 'z' is incompatible with ",
                     "the chunk geometry for 'x' (the number of rows of ",
                     "a chunk in 'z' should be a multiple of the number ",
                     "of rows of a chunk in 'x'). ",
                     "This will likely degrade performance of the block ",
                     "matrix multiplication."), immediate.=TRUE)
    c0 <- z_chunkdim[[2L]]
    if (!is_a_multiple_of(c0, y_chunkdim[[2L]]))
        warning(wmsg("The chunk geometry for 'z' is incompatible with ",
                     "the chunk geometry for 'y' (the number of columns of ",
                     "a chunk in 'z' should be a multiple of the number ",
                     "of columns of a chunk in 'y'). ",
                     "This will likely degrade performance of the block ",
                     "matrix multiplication."), immediate.=TRUE)
    if (C > A && A != 0L)
        a0 <- a0 * as.integer(C / A)
    if (A > C && C != 0L)
        c0 <- c0 * as.integer(A / C)

    ## Find 'a' and 'c'.
    z_dim <- c(A, C)
    z_tile_dim <- pmin(c(a0, c0), z_dim)
    z_tile_grid <- RegularArrayGrid(z_dim, z_tile_dim)
    z_grid <- .block_grid(z_dim, max(as.integer(half_block_len), 1L),
                          z_tile_grid, "scale")
    z_spacings <- dim(z_grid[[1L]])
    a <- z_spacings[[1L]]
    c <- z_spacings[[2L]]

    ## Find 'b'.
    ## TODO: Revisit this.
    b <- max(as.integer(half_block_len / (a + c)), 1L)
    b0 <- max(x_chunkdim[[2L]], y_chunkdim[[1L]])
    r <- b %% b0
    if (r != 0L)
        b <- b + (b0 - r)  # increase 'b' to the next multiple of 'b0'
    if (b > B)
        b <- B

    c(a=a, b=b, c=c)
}

### Return list(x_grid, y_grid, z_grid).
defaultMultAutoGrids <-
    function(x, y, block.length=NULL,
             x.chunkdim=NULL, y.chunkdim=NULL, z.chunkdim=NULL)
{
    x_dim <- dim(x)
    y_dim <- dim(y)
    if (length(x_dim) != 2L || length(y_dim) != 2L)
        stop(wmsg("'x' and 'y' must have exactly 2 dimensions"))
    x_ncol <- x_dim[[2L]]
    y_nrow <- y_dim[[1L]]
    if (x_ncol != y_nrow)
        stop(wmsg("'ncol(x)' must be equal to 'nrow(y)'"))
    A <- x_dim[[1L]]
    B <- x_ncol  # same as 'y_nrow'
    C <- y_dim[[2L]]
    z_dim <- c(A, C)

    ## Assuming that 'x', 'y', and 'z' are all of type "double". This could
    ## be refined.
    half_block_len <- .normarg_block.length(block.length, "double") / 2

    if (is.null(x.chunkdim)) {
        x_chunkdim <- chunkdim(x)
        if (is.null(x_chunkdim))
            x_chunkdim <- pmin(1L, x_dim)
    } else {
        x_chunkdim <- .normarg_z_chunkdim(x.chunkdim, x_dim,
                                          what="x", ncol_what="x")
    }
    if (is.null(y.chunkdim)) {
        y_chunkdim <- chunkdim(y)
        if (is.null(y_chunkdim))
            y_chunkdim <- pmin(1L, y_dim)
    } else {
        y_chunkdim <- .normarg_z_chunkdim(y.chunkdim, y_dim,
                                          what="y", nrow_what="y")
    }
    if (is.null(z.chunkdim)) {
        z_chunkdim <- pmin(1L, z_dim)
    } else {
        z_chunkdim <- .normarg_z_chunkdim(z.chunkdim, z_dim)
    }

    abc <- .find_abc(A, B, C, half_block_len,
                     x_chunkdim, y_chunkdim, z_chunkdim)

    x_grid <- RegularArrayGrid(x_dim, abc[-3L])
    y_grid <- RegularArrayGrid(y_dim, abc[-1L])
    z_grid <- RegularArrayGrid(z_dim, abc[-2L])
    list(x_grid, y_grid, z_grid)
}

multGrids <- function(...)
{
    .Defunct("defaultMultAutoGrids")
    defaultMultAutoGrids(...)
}