Closed Marius1311 closed 4 years ago
This adresses a memory issue we had with very large sample sizes. Essentially, we should never create a matrix of size n_samples x n_samples. Previously, we did this in commands such as np.diag(1. / np.sqrt(eta), which raises the memory error.
n_samples x n_samples
np.diag(1. / np.sqrt(eta)
@giovp, this should hopefully fix your memory error!
okay, should be done!
thanks!
This adresses a memory issue we had with very large sample sizes. Essentially, we should never create a matrix of size
n_samples x n_samples
. Previously, we did this in commands such asnp.diag(1. / np.sqrt(eta)
, which raises the memory error.