colRowLogSumExps

matrixStats: Benchmark report

---

colLogSumExps() and rowLogSumExps() benchmarks

This report benchmark the performance of colLogSumExps() and rowLogSumExps() against alternative methods.

Alternative methods

• apply() + matrixStats::logSumExp()
• apply() + logSumExp0()

where

> logSumExp0 <- function(lx, ...) {
+     iMax <- which.max(lx)
+     log1p(sum(exp(lx[-iMax] - lx[iMax]))) + lx[iMax]
+ }

Data

> rmatrix <- function(nrow, ncol, mode = c("logical", "double", "integer", "index"), range = c(-100, 
+     +100), naProb = 0) {
+     mode <- match.arg(mode)
+     n <- nrow * ncol
+     if (mode == "logical") {
+         X <- sample(c(FALSE, TRUE), size = n, replace = TRUE)
+     }     else if (mode == "index") {
+         X <- seq_len(n)
+         mode <- "integer"
+     }     else {
+         X <- runif(n, min = range[1], max = range[2])
+     }
+     storage.mode(X) <- mode
+     if (naProb > 0) 
+         X[sample(n, size = naProb * n)] <- NA
+     dim(X) <- c(nrow, ncol)
+     X
+ }
> rmatrices <- function(scale = 10, seed = 1, ...) {
+     set.seed(seed)
+     data <- list()
+     data[[1]] <- rmatrix(nrow = scale * 1, ncol = scale * 1, ...)
+     data[[2]] <- rmatrix(nrow = scale * 10, ncol = scale * 10, ...)
+     data[[3]] <- rmatrix(nrow = scale * 100, ncol = scale * 1, ...)
+     data[[4]] <- t(data[[3]])
+     data[[5]] <- rmatrix(nrow = scale * 10, ncol = scale * 100, ...)
+     data[[6]] <- t(data[[5]])
+     names(data) <- sapply(data, FUN = function(x) paste(dim(x), collapse = "x"))
+     data
+ }
> data <- rmatrices(mode = "double")

Results

10x10 matrix

> X <- data[["10x10"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   654179 35.0    1168576  62.5  1168576  62.5
Vcells 12247870 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652373 34.9    1168576  62.5  1168576  62.5
Vcells 12242543 93.5   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	0.0115	0.0175	0.0231	0.0216	0.0262	0.1139
apply+logSumExp	0.0858	0.1066	0.1348	0.1409	0.1526	0.2922
apply+logSumExp0	0.1143	0.1347	0.1832	0.1823	0.2102	0.5759

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	7.433	6.088	5.837	6.536	5.831	2.564
apply+logSumExp0	9.900	7.692	7.934	8.455	8.029	5.054
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	0.0108	0.0119	0.0156	0.0146	0.0158	0.1455
apply+logSumExp	0.0828	0.0874	0.0891	0.0889	0.0905	0.0985
apply+logSumExp0	0.1082	0.1163	0.1222	0.1191	0.1209	0.2749

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.0000
apply+logSumExp	7.678	7.322	5.717	6.079	5.731	0.6772
apply+logSumExp0	10.036	9.741	7.846	8.144	7.658	1.8889
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x10 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 10x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowLogSumExps	10.78	11.94	15.58	14.63	15.78	145.5
1	colLogSumExps	11.55	17.52	23.09	21.56	26.18	113.9

	expr	min	lq	mean	median	uq	max
2	rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.0000
1	colLogSumExps	1.071	1.468	1.482	1.474	1.659	0.7831
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 10x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x100 matrix

> X <- data[["100x100"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652504 34.9    1168576  62.5  1168576  62.5
Vcells 12243794 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652498 34.9    1168576  62.5  1168576  62.5
Vcells 12253837 93.5   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	0.4985	0.7018	0.7336	0.7291	0.7589	1.921
apply+logSumExp	1.1048	1.7958	1.9953	1.9003	1.9706	13.446
apply+logSumExp0	1.5514	2.4949	2.6120	2.6127	2.6864	4.294

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.00	1.000	1.000	1.000
apply+logSumExp	2.216	2.559	2.72	2.606	2.596	7.000
apply+logSumExp0	3.112	3.555	3.56	3.583	3.540	2.235
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	0.5278	0.7882	0.8966	0.8354	0.8804	3.462
apply+logSumExp	1.1887	1.8312	2.2091	1.9340	2.1284	13.836
apply+logSumExp0	1.6145	2.4868	2.6728	2.6475	2.7463	9.375

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	2.252	2.323	2.464	2.315	2.418	3.997
apply+logSumExp0	3.059	3.155	2.981	3.169	3.119	2.708
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x100 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 100x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	498.5	701.8	733.6	729.1	758.9	1921
rowLogSumExps	527.8	788.2	896.6	835.4	880.4	3462

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.00	1.000
rowLogSumExps	1.059	1.123	1.222	1.146	1.16	1.802
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 100x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x10 matrix

> X <- data[["1000x10"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652544 34.9    1168576  62.5  1168576  62.5
Vcells 12244028 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652538 34.9    1168576  62.5  1168576  62.5
Vcells 12254071 93.5   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x10 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	0.4897	0.696	0.6985	0.7129	0.726	0.8088
apply+logSumExp	0.7403	1.167	1.2029	1.2134	1.252	2.3798
apply+logSumExp0	0.9616	1.454	1.4749	1.4996	1.535	1.7550

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	1.512	1.677	1.722	1.702	1.725	2.942
apply+logSumExp0	1.964	2.089	2.111	2.103	2.115	2.170
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x10 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	0.5193	0.5380	0.6781	0.733	0.7728	0.8196
apply+logSumExp	0.7291	0.8126	1.0324	1.130	1.2203	1.8924
apply+logSumExp0	0.9366	1.0334	1.2983	1.442	1.5029	1.7704

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	1.404	1.511	1.522	1.541	1.579	2.309
apply+logSumExp0	1.804	1.921	1.915	1.967	1.945	2.160
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x10 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 1000x10 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	489.7	696	698.5	712.9	726.0	808.8
rowLogSumExps	519.3	538	678.1	733.0	772.8	819.6

expr	min	lq	mean	median	uq	max
colLogSumExps	1.00	1.000	1.0000	1.000	1.000	1.000
rowLogSumExps	1.06	0.773	0.9708	1.028	1.064	1.013
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 1000x10 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

10x1000 matrix

> X <- data[["10x1000"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652574 34.9    1168576  62.5  1168576  62.5
Vcells 12244684 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652568 34.9    1168576  62.5  1168576  62.5
Vcells 12254727 93.5   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	0.5478	0.5782	0.6905	0.7152	0.7816	1.268
apply+logSumExp	4.5359	4.9241	7.3256	7.3143	8.1495	37.867
apply+logSumExp0	7.5624	8.4844	11.3999	11.6254	13.1914	20.483

expr	min	lq	mean	median	uq	max
colLogSumExps	1.00	1.000	1.00	1.00	1.00	1.00
apply+logSumExp	8.28	8.516	10.61	10.23	10.43	29.85
apply+logSumExp0	13.81	14.674	16.51	16.25	16.88	16.15
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	0.5682	0.6077	0.7344	0.7154	0.8517	1.487
apply+logSumExp	4.4404	4.8479	6.6653	6.3040	8.0638	16.329
apply+logSumExp0	7.5632	8.2886	11.5799	12.2136	13.3396	19.998

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.00
apply+logSumExp	7.815	7.978	9.076	8.811	9.468	10.98
apply+logSumExp0	13.311	13.640	15.767	17.072	15.662	13.45
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 10x1000 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 10x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	547.8	578.2	690.5	715.2	781.6	1268
rowLogSumExps	568.2	607.7	734.4	715.4	851.7	1487

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1	1.00	1.000
rowLogSumExps	1.037	1.051	1.064	1	1.09	1.172
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 10x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

100x1000 matrix

> X <- data[["100x1000"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652618 34.9    1168576  62.5  1168576  62.5
Vcells 12245096 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652612 34.9    1168576  62.5  1168576  62.5
Vcells 12345139 94.2   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x1000 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	4.987	5.125	6.197	6.467	7.034	8.882
apply+logSumExp	10.689	12.077	16.137	16.907	17.795	35.634
apply+logSumExp0	15.399	18.637	22.930	23.486	25.178	38.215

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	2.143	2.356	2.604	2.614	2.530	4.012
apply+logSumExp0	3.088	3.636	3.700	3.631	3.579	4.303
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x1000 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	5.711	5.756	6.783	5.974	7.872	12.23
apply+logSumExp	11.405	13.741	16.343	16.003	18.308	26.92
apply+logSumExp0	16.547	18.789	22.424	21.983	25.082	38.72

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	1.997	2.387	2.409	2.679	2.326	2.201
apply+logSumExp0	2.897	3.264	3.306	3.680	3.186	3.166
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 100x1000 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 100x1000 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

	expr	min	lq	mean	median	uq	max
2	rowLogSumExps	5.711	5.756	6.783	5.974	7.872	12.229
1	colLogSumExps	4.987	5.125	6.197	6.467	7.034	8.882

	expr	min	lq	mean	median	uq	max
2	rowLogSumExps	1.0000	1.0000	1.0000	1.000	1.0000	1.0000
1	colLogSumExps	0.8733	0.8904	0.9136	1.083	0.8936	0.7263
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 100x1000 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

1000x100 matrix

> X <- data[["1000x100"]]
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652649 34.9    1168576  62.5  1168576  62.5
Vcells 12245594 93.5   35610798 271.7 68120027 519.8
> colStats <- microbenchmark(colLogSumExps = colLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 2L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 2L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")
> X <- t(X)
> gc()
           used (Mb) gc trigger  (Mb) max used  (Mb)
Ncells   652643 34.9    1168576  62.5  1168576  62.5
Vcells 12345637 94.2   35610798 271.7 68120027 519.8
> rowStats <- microbenchmark(rowLogSumExps = rowLogSumExps(X, na.rm = FALSE), `apply+logSumExp` = apply(X, 
+     MARGIN = 1L, FUN = logSumExp, na.rm = FALSE), `apply+logSumExp0` = apply(X, MARGIN = 1L, FUN = logSumExp0, 
+     na.rm = FALSE), unit = "ms")

Table: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x100 data. The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	4.894	4.920	5.847	5.186	6.928	8.537
apply+logSumExp	7.117	7.917	9.369	8.845	11.085	15.001
apply+logSumExp0	9.180	10.060	11.784	11.339	13.444	21.254

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.00	1.000
apply+logSumExp	1.454	1.609	1.602	1.706	1.60	1.757
apply+logSumExp0	1.876	2.045	2.015	2.186	1.94	2.490
Table: Benchmarking of rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x100 data (transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
rowLogSumExps	5.601	5.655	6.79	5.975	8.103	8.957
apply+logSumExp	7.598	9.131	10.13	9.409	11.600	16.795
apply+logSumExp0	9.843	11.255	12.51	12.061	13.926	18.681

expr	min	lq	mean	median	uq	max
rowLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
apply+logSumExp	1.357	1.615	1.491	1.575	1.431	1.875
apply+logSumExp0	1.757	1.990	1.843	2.019	1.719	2.086
Figure: Benchmarking of colLogSumExps(), apply+logSumExp() and apply+logSumExp0() on 1000x100 data as well as rowLogSumExps(), apply+logSumExp() and apply+logSumExp0() on the same data transposed. Outliers are displayed as crosses. Times are in milliseconds.

Table: Benchmarking of colLogSumExps() and rowLogSumExps() on 1000x100 data (original and transposed). The top panel shows times in milliseconds and the bottom panel shows relative times.

expr	min	lq	mean	median	uq	max
colLogSumExps	4.894	4.920	5.847	5.186	6.928	8.537
rowLogSumExps	5.601	5.655	6.790	5.975	8.103	8.957

expr	min	lq	mean	median	uq	max
colLogSumExps	1.000	1.000	1.000	1.000	1.000	1.000
rowLogSumExps	1.145	1.149	1.161	1.152	1.169	1.049
Figure: Benchmarking of colLogSumExps() and rowLogSumExps() on 1000x100 data (original and transposed). Outliers are displayed as crosses. Times are in milliseconds.

Appendix

Session information

R Under development (unstable) (2015-02-27 r67909)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] markdown_0.7.7          microbenchmark_1.4-2    matrixStats_0.14.0-9000
[4] ggplot2_1.0.0           knitr_1.9.3             R.devices_2.13.0       
[7] R.utils_2.0.0           R.oo_1.19.0             R.methodsS3_1.7.0      

loaded via a namespace (and not attached):
 [1] Rcpp_0.11.4         splines_3.2.0       MASS_7.3-39        
 [4] munsell_0.4.2       lattice_0.20-30     colorspace_1.2-4   
 [7] R.cache_0.11.1-9000 multcomp_1.3-9      stringr_0.6.2      
[10] plyr_1.8.1          tools_3.2.0         grid_3.2.0         
[13] gtable_0.1.2        TH.data_1.0-6       survival_2.38-1    
[16] digest_0.6.8        R.rsp_0.20.0        reshape2_1.4.1     
[19] formatR_1.0.3       base64enc_0.1-3     mime_0.2.1         
[22] evaluate_0.5.7      labeling_0.3        sandwich_2.3-2     
[25] scales_0.2.4        mvtnorm_1.0-2       zoo_1.7-12         
[28] Cairo_1.5-6         proto_0.3-10       

Total processing time was 36.84 secs.

Reproducibility

To reproduce this report, do:

html <- matrixStats:::benchmark('colLogSumExps')

---

Copyright Henrik Bengtsson. Last updated on 2015-03-02 17:03:36 (-0800 UTC). Powered by RSP.

<script> var link = document.createElement('link'); link.rel = 'icon'; link.href = "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAACAAAAAgCAMAAABEpIrGAAAADFBMVEX9/v0AAP/9/v3//wBEQjoBAAAABHRSTlP//wD//gy7CwAAAGJJREFUOI3N0rESwCAIA9Ag///PXdoiBk0HhmbNO49DMETQCexNCSyFgdlGoO5DYOr9ThLgPosA7osIQP0sHuDOog8UI/ALa988wzdwXJRctf4s+d36YPTJ6aMd8ux3+QO4ABTtB85yDAh9AAAAAElFTkSuQmCC" document.getElementsByTagName('head')[0].appendChild(link); </script>