compare the information between two matrices R - r

I have two matrices, one is generated out of the other by deleting some rows. For example:
m = matrix(1:18, 6, 3)
m1 = m[c(-1, -3, -6),]
Suppose I do not know which rows in m were eliminated to create m1, how should I find it out by comparing the two matrices? The result I want looks like this:
1, 3, 6
The actual matrix I am dealing with is very big. I was wondering if there is any efficient way of conducting it.

Here are some approaches:
1) If we can assume that there are no duplicated rows in m -- this is the case in the example in the question -- then:
which(tail(!duplicated(rbind(m1, m)), nrow(m)))
## [1] 1 3 6
2) Transpose m and m1 giving tm and tm1 since it is more efficient to work on columns than rows.
Define match_indexes(i) which returns a vector r such that each row in m[r, ] matches m1[i, ].
Apply that to each i in 1:n1 and remove the result from 1:n.
n <- nrow(m); n1 <- nrow(m1)
tm <- t(m); tm1 <- t(m1)
match_indexes <- function(i) which(colSums(tm1[, i] == tm) == n1)
setdiff(1:n, unlist(lapply(1:n1, match_indexes)))
## [1] 1 3 6
3) Calculate an interaction vector for each matrix and then use setdiff and finally match to get the indexes:
i <- interaction(as.data.frame(m))
i1 <- interaction(as.data.frame(m1))
match(setdiff(i, i1), i)
## [1] 1 3 6
Added If there can be duplicates in m then (1) and (3) will only return the first of any multiply occurring row in m not in m1.
m <- matrix(1:18, 6, 3)
m1 <- m[c(2, 4, 5),]
m <- rbind(m, m[1:2, ])
# 1
which(tail(!duplicated(rbind(m1, m)), nrow(m)))
## 1 3 6
# 2
n <- nrow(m); n1 <- nrow(m1)
tm <- t(m); tm1 <- t(m1)
match_indexes <- function(i) which(colSums(tm1[, i] == tm) == n1)
setdiff(1:n, unlist(lapply(1:n1, match_indexes)))
## 1 3 6 7
# 3
i <- interaction(as.data.frame(m))
i1 <- interaction(as.data.frame(m1))
match(setdiff(i, i1), i)
## 1 3 6

A possible way is to represent each row as a string:
x1 <- apply(m, 1, paste0, collapse = ';')
x2 <- apply(m1, 1, paste0, collapse = ';')
which(!x1 %in% x2)
# [1] 1 3 6
Some benchmark with a large matrix using my solution and G. Grothendieck's solutions:
set.seed(123)
m <- matrix(rnorm(20000 * 5000), nrow = 20000)
m1 <- m[-sample.int(20000, 1000), ]
system.time({
which(tail(!duplicated(rbind(m1, m)), nrow(m)))
})
# user system elapsed
# 339.888 2.368 342.204
system.time({
x1 <- apply(m, 1, paste0, collapse = ';')
x2 <- apply(m1, 1, paste0, collapse = ';')
which(!x1 %in% x2)
})
# user system elapsed
# 395.428 0.568 395.955
system({
n <- nrow(m); n1 <- nrow(m1)
tm <- t(m); tm1 <- t(m1)
match_indexes <- function(i) which(colSums(tm1[, i] == tm) == n1)
setdiff(1:n, unlist(lapply(1:n1, match_indexes)))
})
# > 15 min, not finish
system({
i <- interaction(as.data.frame(m))
i1 <- interaction(as.data.frame(m1))
match(setdiff(i, i1), i)
})
# run out of memory. My 32G RAM machine crashed.

We can also use do.call
which(!do.call(paste, as.data.frame(m)) %in% do.call(paste, as.data.frame(m1)))
#[1] 1 3 6

Related

R: Find set of columns which contain most 1s in matrix of 0 and 1

I have a matrix of 1s and 0s where the rows are individuals and the columns are events. A 1 indicates that an event happened to an individual and a 0 that it did not.
I want to find which set of (in the example) 5 columns/events that cover the most rows/individuals.
Test Data
#Make test data
set.seed(123)
d <- sapply(1:300, function(x) sample(c(0,1), 30, T, c(0.9,0.1)))
colnames(d) <- 1:300
rownames(d) <- 1:30
My attempt
My initial attempt was just based on combining the set of 5 columns with the highest colMeans:
#Get top 5 columns with highest row coverage
col_set <- head(sort(colMeans(d), decreasing = T), 5)
#Have a look the set
col_set
>
197 199 59 80 76
0.2666667 0.2666667 0.2333333 0.2333333 0.2000000
#Check row coverage of the column set
sum(apply(d[,colnames(d) %in% names(col_set)], 1, sum) > 0) / 30 #top 5
>
[1] 0.7
However this set does not cover the most rows. I tested this by pseudo-random sampling 10.000 different sets of 5 columns, and then finding the set with the highest coverage:
#Get 5 random columns using colMeans as prob in sample
##Random sample 10.000 times
set.seed(123)
result <- lapply(1:10000, function(x){
col_set2 <- sample(colMeans(d), 5, F, colMeans(d))
cover <- sum(apply(d[,colnames(d) %in% names(col_set2)], 1, sum) > 0) / 30 #random 5
list(set = col_set2, cover = cover)
})
##Have a look at the best set
result[which.max(sapply(result, function(x) x[["cover"]]))]
>
[[1]]
[[1]]$set
59 169 262 68 197
0.23333333 0.10000000 0.06666667 0.16666667 0.26666667
[[1]]$cover
[1] 0.7666667
The reason for supplying the colMeans to sample is that the columns with the highest coverages are the ones I am most interested in.
So, using pseudo-random sampling I can collect a set of columns with higher coverage than when just using the top 5 columns. However, since my actual data sets are larger than the example I am looking for a more efficient and rational way of finding the set of columns with the highest coverage.
EDIT
For the interested, I decided to microbenchmark the 3 solutions provided:
#Defining G. Grothendieck's coverage funciton outside his solutions
coverage <- function(ix) sum(rowSums(d[, ix]) > 0) / 30
#G. Grothendieck top solution
solution1 <- function(d){
cols <- tail(as.numeric(names(sort(colSums(d)))), 20)
co <- combn(cols, 5)
itop <- which.max(apply(co, 2, coverage))
co[, itop]
}
#G. Grothendieck "Older solution"
solution2 <- function(d){
require(lpSolve)
ones <- rep(1, 300)
res <- lp("max", colSums(d), t(ones), "<=", 5, all.bin = TRUE, num.bin.solns = 10)
m <- matrix(res$solution[1:3000] == 1, 300)
cols <- which(rowSums(m) > 0)
co <- combn(cols, 5)
itop <- which.max(apply(co, 2, coverage))
co[, itop]
}
#user2554330 solution
bestCols <- function(d, n = 5) {
result <- numeric(n)
for (i in seq_len(n)) {
result[i] <- which.max(colMeans(d))
d <- d[d[,result[i]] != 1,, drop = FALSE]
}
result
}
#Benchmarking...
microbenchmark::microbenchmark(solution1 = solution1(d),
solution2 = solution2(d),
solution3 = bestCols(d), times = 10)
>
Unit: microseconds
expr min lq mean median uq max neval
solution1 390811.850 497155.887 549314.385 578686.3475 607291.286 651093.16 10
solution2 55252.890 71492.781 84613.301 84811.7210 93916.544 117451.35 10
solution3 425.922 517.843 3087.758 589.3145 641.551 25742.11 10
This looks like a relatively hard optimization problem, because of the ways columns interact. An approximate strategy would be to pick the column with the highest mean; then delete the rows with ones in that column, and repeat. You won't necessarily find the best solution this way, but you should get a fairly good one.
For example,
set.seed(123)
d <- sapply(1:300, function(x) sample(c(0,1), 30, T, c(0.9,0.1)))
colnames(d) <- 1:300
rownames(d) <- 1:30
bestCols <- function(d, n = 5) {
result <- numeric(n)
for (i in seq_len(n)) {
result[i] <- which.max(colMeans(d))
d <- d[d[,result[i]] != 1,, drop = FALSE]
}
cat("final dim is ", dim(d))
result
}
col_set <- bestCols(d)
sum(apply(d[,colnames(d) %in% col_set], 1, sum) > 0) / 30 #top 5
This gives 90% coverage.
The following provides a heuristic to find an approximate solution. Find the N=20 columns, say, with the most ones, cols, and then use brute force to find every subset of 5 columns out of those 20. The subset having the highest coverage is shown below and its coverage is 93.3%.
coverage <- function(ix) sum(rowSums(d[, ix]) > 0) / 30
N <- 20
cols <- tail(as.numeric(names(sort(colSums(d)))), N)
co <- combn(cols, 5)
itop <- which.max(apply(co, 2, coverage))
co[, itop]
## [1] 90 123 197 199 286
coverage(co[, itop])
## [1] 0.9333333
Repeating this for N=5, 10, 15 and 20 we get coverages of 83.3%, 86.7%, 90% and 93.3%. The higher the N the better the coverage but the lower the N the less the run time.
Older solution
We can approximate the problem with a knapsack problem that chooses the 5 columns with largest numbers of ones using integer linear programming.
We get the 10 best solutions to this approximate problem, get all columns which are in at least one of the 10 solutions. There are 14 such columns and we then use brute force to find which subset of 5 of the 14 columns has highest coverage.
library(lpSolve)
ones <- rep(1, 300)
res <- lp("max", colSums(d), t(ones), "<=", 5, all.bin = TRUE, num.bin.solns = 10)
coverage <- function(ix) sum(rowSums(d[, ix]) > 0) / 30
# each column of m is logical 300-vector defining possible soln
m <- matrix(res$solution[1:3000] == 1, 300)
# cols is the set of columns which are in any of the 10 solutions
cols <- which(rowSums(m) > 0)
length(cols)
## [1] 14
# use brute force to find the 5 best columns among cols
co <- combn(cols, 5)
itop <- which.max(apply(co, 2, coverage))
co[, itop]
## [1] 90 123 197 199 286
coverage(co[, itop])
## [1] 0.9333333
You can try to test if there is a better column and exchange this with the one currently in the selection.
n <- 5 #Number of columns / events
i <- rep(1, n)
for(k in 1:10) { #How many times itterate
tt <- i
for(j in seq_along(i)) {
x <- +(rowSums(d[,i[-j]]) > 0)
i[j] <- which.max(colSums(x == 0 & d == 1))
}
if(identical(tt, i)) break
}
sort(i)
#[1] 90 123 197 199 286
mean(rowSums(d[,i]) > 0)
#[1] 0.9333333
Taking into account, that the initial condition influences the result you can take random starts.
n <- 5 #Number of columns / events
x <- apply(d, 2, function(x) colSums(x == 0 & d == 1))
diag(x) <- -1
idx <- which(!apply(x==0, 1, any))
x <- apply(d, 2, function(x) colSums(x != d))
diag(x) <- -1
x[upper.tri(x)] <- -1
idx <- unname(c(idx, which(apply(x==0, 1, any))))
res <- sample(idx, n)
for(l in 1:100) {
i <- sample(idx, n)
for(k in 1:10) { #How many times itterate
tt <- i
for(j in seq_along(i)) {
x <- +(rowSums(d[,i[-j]]) > 0)
i[j] <- which.max(colSums(x == 0 & d == 1))
}
if(identical(tt, i)) break
}
if(sum(rowSums(d[,i]) > 0) > sum(rowSums(d[,res]) > 0)) res <- i
}
sort(res)
#[1] 90 123 197 199 286
mean(rowSums(d[,res]) > 0)
#[1] 0.9333333

Faster computation of double for loop?

I have a piece of working code that is taking too many hours (days?) to compute.
I have a sparse matrix of 1s and 0s, I need to subtract each row from any other row, in all possible combinations, multiply the resulting vector by another vector, and finally average the values in it so to get a single scalar which I need to insert in a matrix. What I have is:
m <- matrix(
c(0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0), nrow=4,ncol=4,
byrow = TRUE)
b <- c(1,2,3,4)
for (j in 1:dim(m)[1]){
for (i in 1:dim(m)[1]){
a <- m[j,] - m[i,]
a[i] <- 0L
a[a < 0] <- 0L
c <- a*b
d[i,j] <- mean(c[c > 0])
}
}
The desired output is matrix with the same dimensions of m, where each entry is the result of these operations.
This loop works, but are there any ideas on how to make this more efficient? Thank you
My stupid solution is to use apply or sapply function, instead of for loop to do the iterations:
sapply(1:dim(m)[1], function(k) {z <- t(apply(m, 1, function(x) m[k,]-x)); diag(z) <- 0; z[z<0] <- 0; apply(t(apply(z, 1, function(x) x*b)),1,function(x) mean(x[x>0]))})
I tried to compare your solution and this in terms of running time in my computer, yours takes
t1 <- Sys.time()
d1 <- m
for (j in 1:dim(m)[1]){
for (i in 1:dim(m)[1]){
a <- m[j,] - m[i,]
a[i] <- 0L
a[a < 0] <- 0L
c <- a*b
d1[i,j] <- mean(c[c > 0])
}
}
Sys.time()-t1
Yours needs Time difference of 0.02799988 secs. For mine, it is reduced a bit but not too much, i.e., Time difference of 0.01899815 secs, when you run
t2 <- Sys.time()
d2 <- sapply(1:dim(m)[1], function(k) {z <- t(apply(m, 1, function(x) m[k,]-x)); diag(z) <- 0; z[z<0] <- 0; apply(t(apply(z, 1, function(x) x*b)),1,function(x) mean(x[x>0]))})
Sys.time()-t2
You can try it on your own computer with larger matrix, good luck!
1) create test sparse matrix:
nc <- nr <- 100
p <- 0.001
require(Matrix)
M <- Matrix(0L, nr, nc, sparse = T) # 0 matrix
n1 <- ceiling(p * (prod(dim(M)))) # 1 count
M[1:n1] <- 1L # fill only first column, to approximate max non 0 row count
# (each row has at maximum 1 positive element)
sum(M)/(prod(dim(M)))
b <- 1:ncol(M)
sum(rowSums(M))
So, if the proportion given is correct then we have at most 10 rows that contain non 0 elements
Based on this fact and your supplied calculations:
# a <- m[j, ] - m[i, ]
# a[i] <- 0L
# a[a < 0] <- 0L
# c <- a*b
# mean(c[c > 0])
we can see that the result will be meaningful only form[, j] rows which have at least 1 non 0 element
==> we can skip calculations for all m[, j] which contain only 0s, so:
minem <- function() { # write as function
t1 <- proc.time() # timing
require(data.table)
i <- CJ(1:nr, 1:nr) # generate all combinations
k <- rowSums(M) > 0L # get index where at least 1 element is greater that 0
i <- i[data.table(V1 = 1:nr, k), on = 'V1'] # merge
cat('at moust', i[, sum(k)/.N*100], '% of rows needs to be calculated \n')
i[k == T, rowN := 1:.N] # add row nr for 0 subset
i2 <- i[k == T] # subset only those indexes who need calculation
a <- M[i2[[1]],] - M[i2[[2]],] # operate on all combinations at once
a <- drop0(a) # clean up 0
ids <- as.matrix(i2[, .(rowN, V2)]) # ids for 0 subset
a[ids] <- 0L # your line: a[i] <- 0L
a <- drop0(a) # clean up 0
a[a < 0] <- 0L # the same as your line
a <- drop0(a) # clean up 0
c <- t(t(a)*b) # multiply each row with vector
c <- drop0(c) # clean up 0
c[c < 0L] <- 0L # for mean calculation
c <- drop0(c) # clean up 0
r <- rowSums(c)/rowSums(c > 0L) # row means
i[k == T, result := r] # assign results to data.table
i[is.na(result), result := NaN] # set rest to NaN
d2 <- matrix(i$result, nr, nr, byrow = F) # create resulting matrix
t2 <- proc.time() # timing
cat(t2[3] - t1[3], 'sec \n')
d2
}
d2 <- minem()
# at most 10 % of rows needs to be calculated
# 0.05 sec
Test on smaller example if results matches
d <- matrix(NA, nrow(M), ncol(M))
for (j in 1:dim(M)[1]) {
for (i in 1:dim(M)[1]) {
a <- M[j, ] - M[i, ]
a[i] <- 0L
a[a < 0] <- 0L
c <- a*b
d[i, j] <- mean(c[c > 0])
}
}
all.equal(d, d2)
Can we get results for your real data size?:
# generate data:
nc <- nr <- 6663L
b <- 1:nr
p <- 0.0001074096 # proportion of 1s
M <- Matrix(0L, nr, nc, sparse = T) # 0 matrix
n1 <- ceiling(p * (prod(dim(M)))) # 1 count
M[1:n1] <- 1L
object.size(as.matrix(M))/object.size(M)
# storing this data in usual matrix uses 4000+ times more memory
# calculation:
d2 <- minem()
# at most 71.57437 % of rows needs to be calculated
# 28.33 sec
So you need to convert your matrix to sparse one with
M <- Matrix(m, sparse = T)

Extract the combinations of cells without repeating the index

I am trying to calculate the combinations of elements of a matrix but each element should appear only once.
The (real) matrix is symmetric, and can have more then 5 elements (up to ~2000):
o <- matrix(runif(25), ncol = 5, nrow = 5)
dimnames(o) <- list(LETTERS[1:5], LETTERS[1:5])
# A B C D E
# A 0.4400317 0.1715681 0.7319108946 0.3994685 0.4466997
# B 0.5190471 0.1666164 0.3430245044 0.3837903 0.9322599
# C 0.3249180 0.6122229 0.6312876740 0.8017402 0.0141673
# D 0.1641411 0.1581701 0.0001703419 0.7379847 0.8347536
# E 0.4853255 0.5865909 0.6096330935 0.8749807 0.7230507
I desire to calculate the product of all the combinations of pairs (If possible it should appear all elements:AB, CD, EF if the matrix is of 6 elements), where for each pair one letter is the column, the other one is the row. Here are some combinations:
AB, CD, E
AC, BD, E
AD, BC, E
AE, BC, D
AE, BD, C
Where the value of the single element is just 1.
Combinations not desired:
AB, BC: Element B appears twice
AB, AC: Element A appears twice
Things I tried:
I thought about removing the unwanted part of the matrix:
out <- which(upper.tri(o), arr.ind = TRUE)
out <- cbind.data.frame(out, value = o[upper.tri(o)])
out[, 1] <- colnames(o)[out[, 1]]
out[, 2] <- colnames(o)[out[, 2]]
# row col value
# 1 A B 0.1715681
# 2 A C 0.7319109
# 3 B C 0.3430245
# 4 A D 0.3994685
# 5 B D 0.3837903
# 6 C D 0.8017402
# 7 A E 0.4466997
# 8 B E 0.9322599
# 9 C E 0.0141673
# 10 D E 0.8347536
My attempt involves the following process:
Make a copy of the matrix (out)
Store first value of the first row.
Remove all the pairs that involve any of the pair.
Select the next pair of the resulting matrix
Repeat until all rows are removed of the matrix
Repeat 2:5 starting from a different row
However, this method has one big problem, it doesn't guarantee that all the combinations are stored, and it could store several times the same combination.
My expected output is a vector, where each element is the product of the values in the cell selected by the combination:
AB, CD: 0.137553
How can I extract all those combinations efficiently?
This might work. I tested this on N elements = 5 and 6.
Note that this is not optimised, and hopefully can provide a framework for you to work from. With a much larger array, I can see steps involving apply and combn being a bottleneck.
The idea here is to generate a collection of unique sets first before calculating the product of the sets from another data.frame that stores values of sets.
Unique sets are identified by counting the number of unique elements in all combination pairs. For example, if N elements = 6, we expect length(unlist(combination)) == 6. The same is true if N elements = 7 (there will only be 3 pairs plus a remainder element). In cases where N elements is odd, we can ignore the remaining, unpaired element since it is constrained by the other elements.
library(dplyr)
library(reshape2)
## some functions
unique_by_n <- function(inlist, N){
## select unique combinations by count
## if unique, expect n = 6 if n elements = 6)
if(N %% 2) N <- N - 1 ## for odd numbers
return(length(unique(unlist(inlist))) == N)
}
get_combs <- function(x,xall){
## format and catches remainder if matrix of odd elements
xu <- unlist(x)
remainder <- setdiff(xall,xu) ## catch remainder if any
xset <- unlist(lapply(x, paste0, collapse=''))
finalset <- c(xset, remainder)
return(finalset)
}
## make dataset
set.seed(0) ## set reproducible example
#o <- matrix(runif(25), ncol = 5, nrow = 5) ## uncomment to test 5
#dimnames(o) <- list(LETTERS[1:5], LETTERS[1:5])
o <- matrix(runif(36), ncol = 6, nrow = 6)
dimnames(o) <- list(LETTERS[1:6], LETTERS[1:6])
o[lower.tri(o)] <- t(o)[lower.tri(o)] ## make matrix symmetric
n_elements = nrow(o)
#### get matrix
dat <- melt(o, varnames = c('Rw', 'Cl'), as.is = TRUE)
dat$Set <- apply(dat, 1, function(x) paste0(sort(unique(x[1:2])), collapse = ''))
## get unique sets (since your matrix is symmetric)
dat <- subset(dat, !duplicated(Set))
#### get sets
elements <- rownames(o)
allpairs <- expand.grid(Rw = elements, Cl = elements) %>%
filter(Rw != Cl) ## get all pairs
uniqpairsgrid <- unique(t(apply(allpairs,1,sort)))
uniqpairs <- split(uniqpairsgrid, seq(nrow(uniqpairsgrid))) ## get unique pairs
allpaircombs <- combn(uniqpairs,floor(n_elements/2)) ## get combinations of pairs
uniqcombs <- allpaircombs[,apply(allpaircombs, 2, unique_by_n, N = n_elements)] ## remove pairs with repeats
finalcombs <- apply(uniqcombs, 2, get_combs, xall=elements)
#### calculate results
res <- apply(finalcombs, 2, function(x) prod(subset(dat, Set %in% x)$value)) ## calculate product
names(res) <- apply(finalcombs, 2, paste0, collapse=',') ## add names
resdf <- data.frame(Sets = names(res), Products = res, stringsAsFactors = FALSE, row.names = NULL)
print(resdf)
#> Sets Products
#> 1 AB,CD,EF 0.130063454
#> 2 AB,CE,DF 0.171200062
#> 3 AB,CF,DE 0.007212619
#> 4 AC,BD,EF 0.012494787
#> 5 AC,BE,DF 0.023285088
#> 6 AC,BF,DE 0.001139712
#> 7 AD,BC,EF 0.126900247
#> 8 AD,BE,CF 0.158919605
#> 9 AD,BF,CE 0.184631344
#> 10 AE,BC,DF 0.042572488
#> 11 AE,BD,CF 0.028608495
#> 12 AE,BF,CD 0.047056905
#> 13 AF,BC,DE 0.003131029
#> 14 AF,BD,CE 0.049941770
#> 15 AF,BE,CD 0.070707311
Created on 2018-07-23 by the [reprex package](http://reprex.tidyverse.org) (v0.2.0.9000).
Maybe the following does what you want.
Note that I was more interested in being right than in performance.
Also, I have set the RNG seed, to have reproducible results.
set.seed(9840) # Make reproducible results
o <- matrix(runif(25), ncol = 5, nrow = 5)
dimnames(o) <- list(LETTERS[1:5], LETTERS[1:5])
cmb <- combn(LETTERS[1:5], 2)
n <- ncol(cmb)
res <- NULL
nms <- NULL
for(i in seq_len(n)){
for(j in seq_len(n)[-seq_len(i)]){
x <- unique(c(cmb[, i], cmb[, j]))
if(length(x) == 4){
res <- c(res, o[cmb[1, i], cmb[2, i]] * o[cmb[1, j], cmb[2, j]])
nms <- c(nms, paste0(cmb[1, i], cmb[2, i], '*', cmb[1, j], cmb[2, j]))
}
}
}
names(res) <- nms
res

Computing number of bits that are set to 1 for matching rows in terms of hamming distance between two data frames

I have two data frames of same number of columns (but not rows) df1 and df2. For each row in df2, I was able to find the best (and second best) matching rows from df1 in terms of hamming distance, in my previous post. In that post, we have been using the following example data:
set.seed(0)
df1 <- as.data.frame(matrix(sample(1:10), ncol = 2)) ## 5 rows 2 cols
df2 <- as.data.frame(matrix(sample(1:6), ncol = 2)) ## 3 rows 2 cols
I now need to compute the number of bits equal to 1 for:
each row in df2
the best matching rows in df1
the second matching rows in df1
The number of bits equal to 1 of an integer a maybe computed as
sum(as.integer(intToBits(a)))
And I have applied this to #ZheyuanLi's original function, so I have got item 1>. However I'm unable to apply the same logic to get item 2> and 3>, by simple modification of #ZheyuanLi's function.
Below are the functions from #ZheyuanLi's with modification:
hmd <- function(x,y) {
rawx <- intToBits(x)
rawy <- intToBits(y)
nx <- length(rawx)
ny <- length(rawy)
if (nx == ny) {
## quick return
return (sum(as.logical(xor(rawx,rawy))))
} else if (nx < ny) {
## pivoting
tmp <- rawx; rawx <- rawy; rawy <- tmp
tmp <- nx; nx <- ny; ny <- tmp
}
if (nx %% ny) stop("unconformable length!") else {
nc <- nx / ny ## number of cycles
return(unname(tapply(as.logical(xor(rawx,rawy)), rep(1:nc, each=ny), sum)))
}
}
foo <- function(df1, df2, p = 2) {
## check p
if (p > nrow(df2)) p <- nrow(df2)
## transpose for CPU cache friendly code
xt <- t(as.matrix(df1))
yt <- t(as.matrix(df2))
## after transpose, we compute hamming distance column by column
## a for loop is decent; no performance gain from apply family
n <- ncol(yt)
id <- integer(n * p)
d <- numeric(n * p)
sb <- integer(n)
k <- 1:p
for (i in 1:n) {
set.bits <- sum(as.integer(intToBits(yt[,i])))
distance <- hmd(xt, yt[,i])
minp <- order(distance)[1:p]
id[k] <- minp
d[k] <- distance[minp]
sb[i] <- set.bits
k <- k + p
}
## recode "id", "d" and "sb" into data frame and return
id <- as.data.frame(matrix(id, ncol = p, byrow = TRUE))
colnames(id) <- paste0("min.", 1:p)
d <- as.data.frame(matrix(d, ncol = p, byrow = TRUE))
colnames(d) <- paste0("mindist.", 1:p)
sb <- as.data.frame(matrix(sb, ncol = 1)) ## no need for byrow as you have only 1 column
colnames(sb) <- "set.bits.1"
list(id = id, d = d, sb = sb)
}
Running these gives:
> foo(df1, df2)
$id
min1 min2 ## row id for best/second best match in df1
1 1 4
2 2 3
3 5 2
$d
mindist.1 mindist.2 ## minimum 2 hamming distance
1 2 2
2 1 3
3 1 3
$sb
set.bits.1 ## number of bits equal to 1 for each row of df2
1 3
2 2
3 4
OK, after reading through while re-editing your question (many times!), I think I know what you want. Essentially we need change nothing to hmd(). Your required items 1>, 2>, 3> can all be computed after the for loop in foo().
To get item 1>, which you called sb, we can use a tapply(). However, your computation of sb along the for loop is fine, so I will not change it. In the following, I will demonstrate the basic procedure to get item 2> and item 3>.
The id vector inside foo() stores all matching rows in df1:
id <- c(1, 4, 2, 3, 5, 2)
so we can simply extract those rows of df1 (actually, columns of xt), to compute the number of bits equal to 1. As you can see, there are lots of duplicity in id, so we can only computes on unique(id):
id0 <- sort(unique(id))
## [1] 1 2 3 4 5
We now extract those subset columns of xt:
sub_xt <- xt[, id0]
## [,1] [,2] [,3] [,4] [,5]
## V1 9 3 10 5 6
## V2 2 4 8 7 1
To compute the number of bits equal to 1 for each column of sub_xt, we again use tapply() and vectorized approach.
rawbits <- as.integer(intToBits(as.numeric(sub_xt))) ## convert sub_xt to binary
sbxt0 <- unname(tapply(X = rawbits,
INDEX = rep(1:length(id0), each = length(rawbits) / length(id0)),
FUN = sum))
## [1] 3 3 3 5 3
Now we need to map sbxt0 to sbxt:
sbxt <- sbxt0[match(id, id0)]
## [1] 3 5 3 3 3 3
Then we can convert sbxt to a data frame sb1:
sb1 <- as.data.frame(matrix(sbxt, ncol = p, byrow = TRUE))
colnames(sb1) <- paste(paste0("min.", 1:p), "set.bits.1", sep = ".")
## min.1.set.bits.1 min.2.set.bits.1
## 1 3 5
## 2 3 3
## 3 3 3
Finally we can assemble these things up:
foo <- function(df1, df2, p = 2) {
## check p
if (p > nrow(df2)) p <- nrow(df2)
## transpose for CPU cache friendly code
xt <- t(as.matrix(df1))
yt <- t(as.matrix(df2))
## after transpose, we compute hamming distance column by column
## a for loop is decent; no performance gain from apply family
n <- ncol(yt)
id <- integer(n * p)
d <- numeric(n * p)
sb2 <- integer(n)
k <- 1:p
for (i in 1:n) {
set.bits <- sum(as.integer(intToBits(yt[,i])))
distance <- hmd(xt, yt[,i])
minp <- order(distance)[1:p]
id[k] <- minp
d[k] <- distance[minp]
sb2[i] <- set.bits
k <- k + p
}
## compute "sb1"
id0 <- sort(unique(id))
sub_xt <- xt[, id0]
rawbits <- as.integer(intToBits(as.numeric(sub_xt))) ## convert sub_xt to binary
sbxt0 <- unname(tapply(X = rawbits,
INDEX = rep(1:length(id0), each = length(rawbits) / length(id0)),
FUN = sum))
sbxt <- sbxt0[match(id, id0)]
sb1 <- as.data.frame(matrix(sbxt, ncol = p, byrow = TRUE))
colnames(sb1) <- paste(paste0("min.", 1:p), "set.bits.1", sep = ".")
## recode "id", "d" and "sb2" into data frame and return
id <- as.data.frame(matrix(id, ncol = p, byrow = TRUE))
colnames(id) <- paste0("min.", 1:p)
d <- as.data.frame(matrix(d, ncol = p, byrow = TRUE))
colnames(d) <- paste0("mindist.", 1:p)
sb2 <- as.data.frame(matrix(sb2, ncol = 1)) ## no need for byrow as you have only 1 column
colnames(sb2) <- "set.bits.1"
list(id = id, d = d, sb1 = sb1, sb2 = sb2)
}
Now, running foo(df1, df2) gives:
> foo(df1,df2)
$id
min.1 min.2
1 1 4
2 2 3
3 5 2
$d
mindist.1 mindist.2
1 2 2
2 1 3
3 1 3
$sb1
min.1.set.bits.1 min.2.set.bits.1
1 3 5
2 3 3
3 3 3
$sb2
set.bits.1
1 3
2 2
3 4
Note that I have renamed the sb you used to sb2.

R Sum every k columns in matrix

I have a matrix temp1 (dimensions Nx16) (generally, NxM)
I would like to sum every k columns in each row to one value.
Here is what I got to so far:
cbind(rowSums(temp1[,c(1:4)]), rowSums(temp1[,c(5:8)]), rowSums(temp1[,c(9:12)]), rowSums(temp1[,c(13:16)]))
There must be a more elegant (and generalized) method to do it.
I have noticed similar question here:
sum specific columns among rows
couldn't make it work with Ananda's solution;
Got following error:
sapply(split.default(temp1, 0:(length(temp1)-1) %/% 4), rowSums)
Error in FUN(X[[1L]], ...) :
'x' must be an array of at least two dimensions
Please advise.
You can use by:
do.call(cbind, by(t(temp1), (seq(ncol(temp1)) - 1) %/% 4, FUN = colSums))
If the dimensions are equal for the sub matrices, you could change the dimensions to an array and then do the rowSums
m1 <- as.matrix(temp1)
n <- 4
dim(m1) <- c(nrow(m1), ncol(m1)/n, n)
res <- matrix(rowSums(apply(m1, 2, I)), ncol=n)
identical(res[,1],rowSums(temp1[,1:4]))
#[1] TRUE
Or if the dimensions are unequal
t(sapply(seq(1,ncol(temp2), by=4), function(i) {
indx <- i:(i+3)
rowSums(temp2[indx[indx <= ncol(temp2)]])}))
data
set.seed(24)
temp1 <- as.data.frame(matrix(sample(1:20, 16*4, replace=TRUE), ncol=16))
set.seed(35)
temp2 <- as.data.frame(matrix(sample(1:20, 17*4, replace=TRUE), ncol=17))
Another possibility:
x1<-sapply(1:(ncol(temp1)/4),function(x){rowSums(temp1[,1:4+(x-1)*4])})
## check
x0<-cbind(rowSums(temp1[,c(1:4)]), rowSums(temp1[,c(5:8)]), rowSums(temp1[,c(9:12)]), rowSums(temp1[,c(13:16)]))
identical(x1,x0)
# TRUE
Here's another approach. Convert the matrix to an array and then use apply with sum.
n <- 4
apply(array(temp1, dim=c(dim(temp1)/c(1,n), n)), MARGIN=c(1,3), FUN=sum)
Using #akrun's data
set.seed(24)
temp1 <- matrix(sample(1:20, 16*4, replace=TRUE), ncol=16)
a function which sums matrix columns with each group of size n columns
set.seed(1618)
mat <- matrix(rnorm(24 * 16), 24, 16)
f <- function(mat, n = 4) {
if (ncol(mat) %% n != 0)
stop()
cols <- split(colSums(mat), rep(1:(ncol(mat) / n), each = n))
## or use this to have n mean the number of groups you want
# cols <- split(colSums(mat), rep(1:n, each = ncol(mat) / n))
sapply(cols, sum)
}
f(mat, 4)
# 1 2 3 4
# -17.287137 -1.732936 -5.762159 -4.371258
c(sum(mat[,1:4]), sum(mat[,5:8]), sum(mat[,9:12]), sum(mat[,13:16]))
# [1] -17.287137 -1.732936 -5.762159 -4.371258
More examples:
## first 8 and last 8 cols
f(mat, 8)
# 1 2
# -19.02007 -10.13342
## each group is 16 cols, ie, the entire matrix
f(mat, 16)
# 1
# -29.15349
sum(mat)
# [1] -29.15349

Resources