Hi,

I am trying to run NMF on my single cell data. I am working in a slurm cluster. I am using the NMF version 0.25 because the 0.28 was doing a maximun of 2 CPUs for paralellization

As i test I tried this simple command that worked

it <- nmf(nmf_input_dense, rank = 4, method = "brunet", nrun = 2, .options = "vP2")

NMF algorithm: 'brunet'

Multiple runs: 2

Mode: parallel (2/256 core(s))

Runs: |==================================================| 100%

System time:

user system elapsed

798.002 692.236 1505.560

I have tried many different parameters and repetitively I am getting the error

Error in which.min(sapply(res.runs, "[[", "residuals")) :
'list' object cannot be coerced to type 'double'
In addition: Warning message:
In mclapply(argsList, FUN, mc.preschedule = preschedule, mc.set.seed = set.seed, :
scheduled cores 1, 2, 3 did not deliver results, all values of the jobs will be affected
Timing stopped at: 205.3 236.3 671.6

These are the commands i tested that generated the error

nmf(nmf_input_dense, rank = 2, method = "brunet", nrun = 10, .options = "vP5")

nmf(nmf_input_dense, rank = 2, method = "brunet", nrun = 10, .options = "vP3")

t <- nmf(nmf_input_dense, rank = 2, method = "brunet", nrun = 10, .options = "vP2")

t <- nmf(nmf_input_dense, rank = 2, method = "brunet", nrun = 7, .options = "vP5")

it <- nmf(nmf_input_dense, rank = 2, method = "brunet", nrun = 7, .options = "vP3")

I also tried intead of .options = "vP" to use .options = "vp", which led to the same error. if i do vp1 it takes too long. Any idea of how i can solve this?