Giter Club home page Giter Club logo

longyanu / fwi-using-temporal-fourth-fd-modelling Goto Github PK

View Code? Open in Web Editor NEW

This project forked from tumugenshu/fwi-using-temporal-fourth-fd-modelling

2.0 1.0 0.0 43.78 MB

These codes are used to complete elastic full waveform inversion using temporal fourth-order FD modelling. In order to accelete the algorith, GPU-based versions are privided here. What's more, the GPU shared memory is used in block level and the efficiency improvement is about 10%~15% on GTX 750ti, where efficiency improvement is dependent on the length of FD orders.

Makefile 0.66% Cuda 48.14% C++ 26.47% C 1.82% Shell 0.35% MATLAB 22.57%

fwi-using-temporal-fourth-fd-modelling's Introduction

FWI-using-temporal-fourth-FD-modelling

These codes are used to complete elastic full waveform inversion using temporal fourth-order FD modelling. In order to accelete the algorith, GPU-based versions are privided here. What's more, the GPU shared memory is used in block level and the efficiency improvement is about 10%~15% on GTX 750ti, where efficiency improvement is dependent on the length of FD orders.

The codes contains two parts: one is the GPU-based elastic FWI using temporal fourth-order Finite-different modelling method, the other is that we use GPU-shared-memory to optimize the first one further. The tested GPU device is GTX 750ti. The FWI section are contains forward section, which can be used to output the forward seismograms.

How to run the code

Firstly, install the MPICH and Cuda, and set the path correctly on your ~/.bashrc file and source it. Then modify the mpicc and nvcc paths in the Makefile in the package. Then compile and generate executable file, run it by sh run.sh.

main files in the package

Seven files are:

  1. *.cpp,main control code for elastic FWI by MPI. It includes the functions of domain decomposition, multi-scale and encoding number parameters, and the MPI data exchange, model parameter update, and etc.
  2. *.cu, the CUDA code and designed for simultaneous source propagation. The main four GPU kernel functions are forward, reconstructed, backward sections.
  3. headmulti.h, used for declaring all the functions to be called in the elastic_fdtd_3d_FWI_ModelCMultiscale.cpp.
  4. Makefile, used for compiling the program with the CUDA’s nvcc, in addition, some shared library functions are listed in the file. One thing to be mentioned is that we should make sure the paths of cuda and mpi are set correctly in this file.
  5. hostgpu, a name list of nodes of GPUs.
  6. run.sh, used to run the elastic FWI on the nodes listed in hostsgpu by mpiexec or mpirun.
  7. nohup.out, a log file used to write the running status. Created automatically by running run.sh. Two folders are:
  8. input/, used to save initial P- and S-wave velocity and density which are binary and the stored along the depth direction.
  9. output/, is used to store the inversion information, including the inversions of different frequency bands, objective function misfit, and temporal results, such as the synthetic data, residual data andgradients.

FWI inversion data

Overthrust model inversion comparsion between FWI with the temporal 4th-order and with the temporal 2th-order FD. "Tradition" folder contains results of FWI with the temporal 2th-order, and "High_order" folder is FWI with the temporal 4th-order FD.

fwi-using-temporal-fourth-fd-modelling's People

Contributors

jedfang avatar

Stargazers

Pankaj K Mishra avatar Wenquan avatar

Watchers

James Cloos avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.