r - using foreach and %dopar% to output the desired results -

so in simplistic world, lets take data:

set.seed(123)  require(doparallel) cl<-makecluster(2) registerdoparallel(cl)  m <- 10  result = foreach(i=1:m,.combine=cbind) %dopar% {    rows <- (1:10)   <- rnorm(10, 5, 1)   <- round(a, 0)    b <- rnorm(10, 6, 1)   b <- round(b, 0)    df <- data.frame(rows,a,b)    output_1 <- length(df$a[df$a == df$b])   # save number of accounts == b   output_2 <- length(df$a[df$a != df$b])   # save number of accounts  , b not equal    result <- rbind(output_1,output_2)  } 

using can check our output

result[1,] result[2,] 

but want include rows (a string of row identities, not count) correspond a!=b in output of result

the problem not single value , varies per alteration. how can 1 achieve this?

update

if add

output_3 <- setdiff(df$rows, df$rows[df$a == df$b]) 

and adjust

result <- rbind(output_1,output_2, output_3) 

we run more simulations required

here apparently desire:

set.seed(123)  require(doparallel) cl<-makecluster(2) registerdoparallel(cl)  m <- 10  result = foreach(i=seq_len(m)) %dopar% {    rows <- (1:10)   <- rnorm(10, 5, 1)   <- round(a, 0)    b <- rnorm(10, 6, 1)   b <- round(b, 0)    df <- data.frame(rows,a,b)    output_1 <- length(df$a[df$a == df$b])   # save number of accounts == b   output_2 <- length(df$a[df$a != df$b])   # save number of accounts  , b not equal    list(rbind(output_1, output_2), which(df$a != df$b))  } stopcluster(cl) 

this returns list of lists. note need use package dorng if want pass random seed workers.

this vectorized approach use:

set.seed(42) <- matrix(round(rnorm(m * 10, 5, 1), 0), ncol = m) b <- matrix(round(rnorm(m * 10, 6, 1), 0), ncol = m)  which(a != b, arr.ind = true) colsums(a != b) colsums(a == b)