Memory utilization for large problems

Applications
1

Hi everyone!

I’m currently struggling with large acoustic problems.
The number of nodes/elements yields to an overload of RAM.
Is there any way to write matrix entries, e.g. from boundary operators, block-wise in temporary files to compute them bit by bit later on instead of accessing the whole matrix at once?
This especially occurs when multiplying a boundary operator and a grid function.

Best regards,

Fabian

2

Dear Fabian,

this is not possible. We are currently developing new code that will be optimised for parallel clusters. But this will still take a couple of months to finish.