Mac issue

Question

Mac issue

emkober opened this issue 3 years ago · 5 comments

I've installed this on both a Mac laptop and desktop and it runs the swim problem just fine with mpirun using 4 processors. I then try to run my own problems with the command line

mpirun -n 4 python /Users/edward/pyDNMFk/main.py --p_r=4 --p_c=1 --process='pyDNMFk' --fpath='data/' --ftype='csv' --fname='HMX' --init='nnsvd' --itr=5000 --norm='kl' --method='mu' --results_path='results/34' --perturbations=100 --noise_var=0.015 --start_k=3 --end_k=4 --step_k=1 --sill_thr=.9 --sampling='uniform' --prune=true > log34.out &

And I get

/Users/edward/pyDNMFk/pyDNMFk/pyDNMF.py:238: RuntimeWarning: invalid value encountered in true_divide
col_err = np.sqrt(col_err_num / col_err_deno)

4 times (or for each number of processors I requested). Output files are full of data, but the selection plot is messed up. Anything obvious?

Answer 1 · 2021-09-29T17:20:24.000Z

Hi, can you try a larger k range. Maybe like this: mpirun -n 4 python /Users/edward/pyDNMFk/main.py --p_r=4 --p_c=1 --process='pyDNMFk' --fpath='data/' --ftype='csv' --fname='HMX' --init='nnsvd' --itr=5000 --norm='kl' --method='mu' --results_path='results/34' --perturbations=100 --noise_var=0.015 --start_k=3 --end_k=10 --step_k=1 --sill_thr=.9 --sampling='uniform' --prune=true > log34.out &

Answer 2 · 2021-09-29T18:00:30.000Z

I had started out with a larger range like that. Same error. Was trying to simplify it down for possible sources.

…

--------------------------------------------------- Edward M. Kober, Ph.D. Scientist 5 Physics and Chemistry of Materials Group T-1, MS B214 Los Alamos National Laboratory Los Alamos, NM 87545 ***@***.******@***.***> 505-667-5140 From: Maksim Ekin Eren ***@***.***> Reply-To: lanl/pyDNMFk ***@***.***> Date: Wednesday, September 29, 2021 at 11:20 AM To: lanl/pyDNMFk ***@***.***> Cc: "Kober, Ed" ***@***.***>, Author ***@***.***> Subject: [EXTERNAL] Re: [lanl/pyDNMFk] Mac issue (#5) Hi, can you try a larger k range. Maybe like this: mpirun -n 4 python /Users/edward/pyDNMFk/main.py --p_r=4 --p_c=1 --process='pyDNMFk' --fpath='data/' --ftype='csv' --fname='HMX' --init='nnsvd' --itr=5000 --norm='kl' --method='mu' --results_path='results/34' --perturbations=100 --noise_var=0.015 --start_k=3 --end_k=10 --step_k=1 --sill_thr=.9 --sampling='uniform' --prune=true > log34.out & — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#5 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACUQYWXVJNS6UY67OQPZPUTUENDGHANCNFSM5FAFK67A>. Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.

Answer 3 · 2021-09-29T19:39:01.000Z

small_HMX.csv
Here's a smallish (280x1000) data file as an example of what I'm trying to run if that helps.

Answer 4 · 2021-09-29T22:43:59.000Z

Well, going to a bigger k limit does generate rational looking selection plots. Though I still get those runtime warnings about invalid error in true_divide which is irritating, but not catastrophic? Will close for now.

Answer 5 · 2021-09-29T23:33:49.000Z

Hi @emkober, I am glad it is working now. I tried the sample data you shared with the following settings:

mpirun -n 4 python main.py --p_r=4 --p_c=1 --process='pyDNMFk' --fpath='data/' --ftype='csv' --fname='small_HMX' --init='nnsvd' --itr=2000 --norm='kl' --method='mu' --results_path='results/' --perturbations=50 --noise_var=0.015 --start_k=2 --end_k=22 --step_k=2 --sill_thr=.9 --sampling='uniform' --prune='true'

I did get same warning in the end, but the results seems okay with the high k-range. In this plot k-optimal seems to be either 6 or 14: