You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Below is a "ridiculous" example: compute the square of a 1000x1000 identity matrix with the top right hand entry set to a large integer. The computation immediately grabs a lot of memory (about 40Gbyte on my computer), runs for a while, then is killed (presumably by systemd-oomd). The computation should be faster, and should not require so much RAM space.
Here is the OSCAR input to reproduce the problem -- it completes with sz = 500; or with exp = 5000
This is much simplified from the original problematic computation -- a product of two upper triangular matrices with relatively few quite large entries.
The text was updated successfully, but these errors were encountered:
This is a FLINT issue. fmpz_mat_mul chooses an algorithm based on the dimension and maximum height, but does not currently account for the distribution of heights.
Below is a "ridiculous" example: compute the square of a 1000x1000 identity matrix with the top right hand entry set to a large integer. The computation immediately grabs a lot of memory (about 40Gbyte on my computer), runs for a while, then is killed (presumably by
systemd-oomd
). The computation should be faster, and should not require so much RAM space.Here is the OSCAR input to reproduce the problem -- it completes with
sz = 500;
or withexp = 5000
This is much simplified from the original problematic computation -- a product of two upper triangular matrices with relatively few quite large entries.
The text was updated successfully, but these errors were encountered: