Solver stops after Gebhard factors

[kevinarden]

I monitored the system resources during the run, and it peaked at 10 GB (32 GB system) used before it threw the error


RadiationFactors: View factors filling (%) 8.7350E+01
TabulateEmissivity: We used real emissivity for all elements!
DirectSolver: Using direct method: umfpack
Error occurred in umf4num: -1.0000000000000000
STOP 1

so its not RAM

The umfpack manual does say -1 is out of memory, but more specifically

The ANSI C malloc or realloc
routine failed

so maybe more of a memory allocation error not an out of memory error.

[raback]

Hi,

One can also use iterative solvers for computing Gebhart factors. Standard "Linear System ..." keywords apply. Just use "radiation: " as a name space. E.g.

Code: Select all

radiation: Linear System Solver = iterative

The computation is heavy since linear system is solved for each row. Direct solvers usually are preferable as they only do one factoring and the use multiple r.h.s. vectors.

-Peter

[kevinarden]

This works

radiation: Linear System Solver = direct
radiation: Linear System Direct Method = Mumps

[Lhoeltke]

The iterative solver works for me too if I use the keywords Peter posted.
I currently cannot test the MUMPS solver, as I currently use Windows, but will try it next time on a Linux system.

Thank you for resolving these issues for me, I will let you know here if it should arise again with more complex meshes.