Dear All,
I’m running RELAX for a coseismic model:
and I set the grid sizes as below (in meter):
SX=512
SX2=$((SX/2))
DX=2000
OMP_NUM_THREADS=16 relax <<EOF
#grid size
$SX $SX $SX2
#sampling (dx1,dx2,dx3,beta,nyquist)
$DX $DX $DX 0.4 2
The problem is:
When I set nyquist=0, the run finishes normally (very quickly, within about 1–5 minutes).
However, when I set nyquist=2, the computation never ends (likely due to memory usage or FFT grid size explosion). It runs for around 10 hours before I have to terminate it, as I believe the coseismic deformation and stress model should complete much sooner.
The coseismic fault patch size is 4 × 4 km, and the total number of patches is 1024.
Has anyone encountered a similar issue?
Is it important to keep nyquist=2, or is 0 sufficient?
Alternatively, should I keep nyquist=2 but reduce the grid size and increase SX to cover the same area?
Many thanks!
MN
Dear All,
I’m running RELAX for a coseismic model:
and I set the grid sizes as below (in meter):
SX=512
SX2=$((SX/2))
DX=2000
OMP_NUM_THREADS=16 relax <<EOF
#grid size
$SX $SX $SX2
#sampling (dx1,dx2,dx3,beta,nyquist)
$DX $DX $DX 0.4 2
The problem is:
When I set nyquist=0, the run finishes normally (very quickly, within about 1–5 minutes).
However, when I set nyquist=2, the computation never ends (likely due to memory usage or FFT grid size explosion). It runs for around 10 hours before I have to terminate it, as I believe the coseismic deformation and stress model should complete much sooner.
The coseismic fault patch size is 4 × 4 km, and the total number of patches is 1024.
Has anyone encountered a similar issue?
Is it important to keep nyquist=2, or is 0 sufficient?
Alternatively, should I keep nyquist=2 but reduce the grid size and increase SX to cover the same area?
Many thanks!
MN