search

SPECFEM / specfem3d

SPECFEM3D_Cartesian simulates acoustic (fluid), elastic (solid), coupled acoustic/elastic, poroelastic or seismic wave propagation in any type of conforming mesh of hexahedra (structured or not).
GNU General Public License v3.0
398 stars 225 forks source link

replace hardwired material parameter thresholds in model_update.f90 with a new clip_sem.f90 utility #424

Closed rmodrak closed 9 years ago

rmodrak commented 9 years ago

Currently, model_update.f90 contains hardwired vp,vs, and rho thresholds based on a particular geologic region (Southern California). A more useful and flexible approach is to provide a clip_sem utility that obtains threshold vales from the command line. The utility could be modeled closely on Daniel’s existing smooth_sem program.

If there are no objections, I will go ahead and create such a utility in a new src/postprocess directory.

casarotti commented 9 years ago

Hi rmodrak, I think that the hardwired value are not considered in the script... I suppose we can remove directly.... Emanuele

On Thu Feb 19 2015 at 05:01:41 rmodrak notifications@github.com wrote:

Currently, model_update.f90 contains hardwired vp,vs, and rho thresholds based on a particular geologic region (Southern California). A more useful and flexible approach is to provide a clip_sem utility that obtains threshold vales from the command line. The utility could be modeled closely on Daniel’s existing smooth_sem program.

If there are no objections, I will go ahead and create such a utility in a new src/postprocess directory.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/424.

rmodrak commented 9 years ago

Hi Emmanuele,

It looks like the hardwired thresholds in lines of 57-64 of model_update are applied to vp,vs,rho arrays. Then, in lines 344-358, the thresholded vp,vs,rho arrays are used to compute the updated model. It seems like if the hardwired thresholds were removed, it would produce different slightly results from the program. Is this right? Please let me know if I am missing anything.

Thanks, Ryan

On Thu, Feb 19, 2015 at 4:17 AM, emanuele casarotti < notifications@github.com> wrote:

Hi rmodrak, I think that the hardwired value are not considered in the script... I suppose we can remove directly.... Emanuele

On Thu Feb 19 2015 at 05:01:41 rmodrak notifications@github.com wrote:

Currently, model_update.f90 contains hardwired vp,vs, and rho thresholds based on a particular geologic region (Southern California). A more useful and flexible approach is to provide a clip_sem utility that obtains threshold vales from the command line. The utility could be modeled closely on Daniel’s existing smooth_sem program.

If there are no objections, I will go ahead and create such a utility in a new src/postprocess directory.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/424.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/424#issuecomment-75019435 .

komatits commented 9 years ago

Hi all,

Let us remove all hardwired values and let us clarify / simplify this.

Thanks, Dimitri.

On 02/19/2015 02:59 PM, rmodrak wrote:

Hi Emmanuele,

It looks like the hardwired thresholds in lines of 57-64 of model_update are applied to vp,vs,rho arrays. Then, in lines 344-358, the thresholded vp,vs,rho arrays are used to compute the updated model. It seems like if the hardwired thresholds were removed, it would produce different slightly results from the program. Is this right? Please let me know if I am missing anything.

Thanks, Ryan

On Thu, Feb 19, 2015 at 4:17 AM, emanuele casarotti < notifications@github.com> wrote:

Hi rmodrak, I think that the hardwired value are not considered in the script... I suppose we can remove directly.... Emanuele

On Thu Feb 19 2015 at 05:01:41 rmodrak notifications@github.com wrote:

Currently, model_update.f90 contains hardwired vp,vs, and rho thresholds based on a particular geologic region (Southern California). A more useful and flexible approach is to provide a clip_sem utility that obtains threshold vales from the command line. The utility could be modeled closely on Daniel’s existing smooth_sem program.

If there are no objections, I will go ahead and create such a utility in a new src/postprocess directory.

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/424.

— Reply to this email directly or view it on GitHub

https://github.com/geodynamics/specfem3d/issues/424#issuecomment-75019435 .

— Reply to this email directly or view it on GitHub https://github.com/geodynamics/specfem3d/issues/424#issuecomment-75055673.

Dimitri Komatitsch CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics, UPR 7051, Marseille, France http://komatitsch.free.fr

rmodrak commented 9 years ago

I've submitted a pull request for a new utility, xclip_sem.

rmodrak commented 9 years ago

Hi Dimitri, To create a place to respond to your comment "Let us remove all hardwired values", I am opening a new issue about refactoring nonlinear optimization routines. (The issue we are on currently seems a bit narrow to hold the kind of discussion that would be required.)

komatits commented 9 years ago

Done by Ryan @rmodrak .