gplotsel2d memory error?

[andrewgait]

Hello, I'm trying to use this connection set algebra in our work with neuromorphic computing using PyNN on SpiNNaker. I've installed CSA using pip install and then followed the README.md file; everything seems to work fine until I get to the gplotsel2d(...) command, at which point I get a MemoryError:

Traceback (most recent call last):
  File "/localhome/mbbssag3/spinnaker/git/MySpiNNakerProblems/user_problems/test_csa_commands.py", line 97, in <module>
    gplotsel2d(g,c,434)
  File "/usr/local/lib/python2.7/dist-packages/csa/plot.py", line 55, in gplotsel2d
    N = len (cset)
  File "/usr/local/lib/python2.7/dist-packages/csa/connset.py", line 233, in __len__
    for c in self:
  File "/usr/local/lib/python2.7/dist-packages/csa/connset.py", line 726, in simpleIterator
    state):
  File "/usr/local/lib/python2.7/dist-packages/csa/_misc.py", line 87, in iterator
    for j in range (low1, high1):
MemoryError

Do you have any idea what might be causing this to happen?

[andrewgait]

I've done a bit more investigating of this error myself: it seems to be a problem with not specifying a source and a target in gplotsel2d(...); if in the README.md file, I use e.g. gplotsel2d(g, c, 434, range(400,500)) then I get a plot that looks reasonable. So it perhaps seems as though elementary.N isn't producing the expected connection set at this point... ?

[mdjurfeldt]

Hi Andrew,

Interesting---the error I get is:

AssertionError: infinite ISetBoundedMask:s currently not supported

Thank you for the bug report. What has happened here is that an optimization in one place has exposed a limitation of the implementation in another place.

I will fix this and update you. However, I can do that first in a few days.

Best regards,
Mikael

[sanchalitorpe-source]

I submitted a PR that fixes the MemoryError in gplotsel2d by limiting the default source and target to the first 100 neurons. Users can still pass custom ranges if they want. See PR #33