INFO:cooler.balance:variance is nan

[aakashsur]

cooler.txt

Looks like balance (called from command line) starts working but then blows up the weights, and runs into an overflow error causing the variance to go to NaN. Surprisingly this doesn't throw an error, and continues until the iteration limit is reached. Perhaps there should be an error thrown? Also any tips on getting this to work? I'm experimenting with downsampled Hi-C reads so there are few reads in this set.

[aakashsur]

I was just thinking maybe it would be nice to have a convergence policy which stores the unbalanced data if there's a problem with balancing?