|
[Sponsors] |
June 13, 2014, 11:13 |
matrix storage in LU-SSOR
|
#1 |
Senior Member
Joachim
Join Date: Mar 2012
Location: Paris, France
Posts: 145
Rep Power: 15 |
Hey!
In the LU-SSOR implicit scheme, when solving the system (E+D+F)dU = -R, we execute several sweeps (E+D)dU(k+1/2) = -R(n) - FdU(k) (F+D)dU(k+1) = -R(n) - EdU(k+1/2) If I understand correctly, the matrices E, D and F are typically computed at time step n. My question: is it better to compute E, D and F at the beginning of the loop and store them (requires lots of memory, each matrix is actually a 5x5 matrix per cell), or recompute them at each iteration (waste of time...). I thought that SSOR was supposed to be a low-memory scheme. How is this situation typically handled? Thank you very much for your help! Joachim |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Matrix storage and diagonal coeffients calculation | santiagomarquezd | OpenFOAM Programming & Development | 16 | January 14, 2022 05:48 |
more equation in block matrix system | yhaomin2007 | OpenFOAM | 1 | September 6, 2012 09:33 |
Force can not converge | colopolo | CFX | 13 | October 4, 2011 23:03 |
OpenFOAM version 1.6 details | lakeat | OpenFOAM Running, Solving & CFD | 42 | August 26, 2009 22:47 |
Diagonal storage for c-grid airfoil possible? | zonexo | Main CFD Forum | 1 | May 10, 2006 14:43 |