dfm
Memory leak with corner.corner()
I'm using the corner package, version 2.0.1 to produce some corner plots after importing matrices stored in text files. My laptop has 16 GB RAM + 16 GB SWAP. The cluster at my institute has ~128 GB of memory. Matrices to be loaded may vary from 1.2e6 x 14 elements up to 13e6 x 26 elements and they are loaded and stored through np.loadtxt() without any problem. I load matrices and produce the corner plots within a for cycle so that at the following cycle step the previous matrix is overwritten by the new one. To reduce memory consumption I also delete both the stored matrix and the figure variable produced by corner.corner() with the following commands
- del
- gc.collect() The number of panels in the corner plots may vary from 105 to 351.
Monitoring the memory status through htop the memory consumption starts increasing when corner.corner(matrix) is running and for the biggest matrices even at the cluster shell I got 'Killed' when memory is exhausted.
The matrices I'm dealing with are large, but not insanely large. Could you have a check to this memory issue?
Best regards, Andrea
