This repository contains codes for solving inverse obstacle scattering problems to recover the shape of the obstacle for sound soft, sound hard, and penetrable objects.
- Fortran compiler, such as gfortran packaged with GCC (> 6.3.0)
- GNU make
- For linux, run
make install - For Mac OSX, run
cp make.inc.macosx make.incfollowed bymake matlab - For Windows, follow install instructions to get MingW 6.3.0 and command line tools here, then run
cp make.inc.mingw.windows make.incfollowed bymake matlab
This should generate the appropriate mex files in the matlab/
directory.
After the generation of the mex file, run startup.m in MATLAB, this
will set correctly the PATH variables in MATLAB.
To verify successful installation of the program, run tests/test_all.m in
MATLAB. The output in matlab should be::
Successfully completed 3 out of 3 tests in the ios2d testing suite
This package can be used in two modes:
- Data generation mode: To generate sensor measurements corresponding to the solution of acoustic scattering
from a known obstacle. The generation of such a
data set for a star shaped domain is demonstrated in
examples/data_generation_scripts/starn_tensor_data.m - Inverse problem mode: To generate the shape of the obstacle, given sensor measurements.
An example solution for the inverse problem, given a precomputed
dataset, is demonstrated in
examples/inverse_solver_scripts/star3_inverse_test.m