Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

The problem of mesh generation in large-scale computation #795

Open
pickhun opened this issue Aug 18, 2022 · 1 comment
Open

The problem of mesh generation in large-scale computation #795

pickhun opened this issue Aug 18, 2022 · 1 comment

Comments

@pickhun
Copy link

pickhun commented Aug 18, 2022

Hello Specfem developers,
I got some errors when I was doing the global calculation with parameters like these:

NCHUNKS                         = 6
NEX_XI                          = 1280
NEX_ETA                         = 1280 
NPROC_XI                        = 20
NPROC_ETA                       = 20
.....
 Error Jacobian rank:         112 has invalid Jacobian   -1.7341319145770486E-011
   Jacobian:   -1.7341319145770486E-011 xxi/eta/gamma  -2.0014604925344681E-005  -5.8006096436846732E-004   5.6412993062471273E-004 yxi/eta/gamma   5.7396394731065725E-004   2.6131361999248703E-005  -3.3654015271666893E-005 zxi -0.36238342785651995       -2.0800740777460192E-004
   location x/y/z:   0.31748488389093243       0.15004926106803654       0.92655532817117825     
   location r    :    6312.80957     km - lat/lon:    69.2436829       25.2962894    
 3D Jacobian undefined in recalc_jacobian_gll3D.f90
 Error detected, aborting MPI... proc          112
 Error Jacobian rank:         288 has invalid Jacobian   -1.4970956583369370E-012
   Jacobian:   -1.4970956583369370E-012 xxi/eta/gamma  -2.6258967192993588E-005  -5.7149931204916737E-004  -5.7021198320161837E-004 yxi/eta/gamma   5.7061434079935802E-004   2.8668296352396538E-005   2.9312942425585575E-005 zxi -0.39413432425764749       -1.9543512180586475E-004
   location x/y/z:  -0.32292386399244666      -0.13895433528715900       0.92755535522665955     
   location r    :    6319.65625     km - lat/lon:    69.2428818      -156.717758    
 3D Jacobian undefined in recalc_jacobian_gll3D.f90
 Error detected, aborting MPI... proc          288
 Error Jacobian rank:         212 has invalid Jacobian   -3.8457273246922743E-011
   Jacobian:   -3.8457273246922743E-011 xxi/eta/gamma   1.9551839928438897E-005  -1.1894699812674894E-003  -8.8128743636582685E-006 yxi/eta/gamma   1.1363529112820014E-003  -2.9460177165235857E-005  -5.7281792260666098E-004 zxi   3.4006083257111386        2.3927104302751134E-004
   location x/y/z:   -3.5828415197796372E-002  0.18358277632332748       0.97274758730172051     
   location r    :    6310.90625     km - lat/lon:    79.1156387       101.043159    
 3D Jacobian undefined in recalc_jacobian_gll3D.f90
 Error detected, aborting MPI... proc          212
.....

But when I set TOPOGRAPHY as .false., the mesh generation is working properly.
I have tried using highly smooth ETOPO1 data (SIZE_FILTER_ONE_SIDE=50 at run_create_topo_bathy_file.py), but it also doesn't work.
What should I do to avoid this problem?

@ammcpherson
Copy link
Contributor

Hi,

I am not a developer, but I ran into the same problem a couple months ago where I was trying to use a very smoothed version of ETOPO1 and was getting invalid Jacobians. I eventually concluded that there is an anomalous value in the ETOPO1 file somewhere, that, despite smoothing, is creating this numerical instability. If you have the capability to visualize the topography file, you might be able to verify if that is true for you, and manually reassign those values.

Best of luck, and I hope this helps.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants