Prompted by Feligernon Bit's recent post on Rbloggers, I tried to check out BiocParallel.
Whilst trying the example in section 4.2.1 ("Ad-hoc clusters of multiple machines, sockets" of the "Introduction to BiocParallel" (May 6, 2015, December 15, 2015), I am encountering an error: "Error in 0.4 * nx/tasks : non-numeric argument to binary operator".
The only change that I make to the example is to use one "localhost" instead of the three "rhinoN" entries. I get the same failure with one "localhost" & second host "foobar". I can "ssh localhost" no problem.
The code, error, & systemInfo are below. I believe that I am using the most up-to-date software, etc.
Seems to me that the issue is probably that "tasks" is zero, causing a divide by zero resulting in a non-numeric Infinity.
Can some kind soul point out what I am doing wrong?
A big Thank You to the BioConductor & BiocParallel developers.
Thank you,
Lee
> library("BiocParallel", lib.loc="~/R/i686-pc-linux-gnu-library/3.2")
> hosts <- c("localhost")
> param <- SnowParam(workers = hosts, type = "SOCK")
> FUN <- function(i) system("hostname", intern=TRUE)
> bplapply(1:4, FUN, BPPARAM = param)
Error in 0.4 * nx/tasks : non-numeric argument to binary operator
> sessionInfo()
R version 3.2.3 (2015-12-10)
Platform: i686-pc-linux-gnu (32-bit)
Running under: Ubuntu 15.10
locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] BiocParallel_1.4.3
loaded via a namespace (and not attached):
[1] parallel_3.2.3 futile.logger_1.4.1 tools_3.2.3 lambda.r_1.1.7 futile.options_1.0.0
As a workaround specify workers as a numeric value;
SnowParam(workers=1); a fix is in development. Thanks for the report.Thank you for the timely workaround suggestion. If I understand correctly SnowParam(workers=1) will put all workers on my local node? I have been using the workers=numeric (I used 3) & making believe that I was running on remotes to help me learn BiocParallel. When the fullness of time brings the next version of BiocParallel, I will try again with remote nodes.
Lee
Yep, all the workers on a local node; obviously not particularly satisfactory, but I expect a better solution withing the week.
Thank you. I was testing my understanding. I have plenty enough right now to help me understand BiocParallel & understand entirely that good fixes take time. I am more than happy to find out that I did not miss something obvious in attempting to duplicate the example. If the examples where not so good & extensive, I never would have gone down that path ;-)
Lee