Hi,
I have been running simulations with the kspaceFirstOrder3D solver and with the CPU and GPU versions (kspaceFirstOrder3DC and kspaceFirstOrder3DG) and it works well.
But when I add attenuation to the medium (using medium.alpha_coeff), kspaceFirstOrder3D returns the expected result while the CPU and GPU versions return very weird fields that are nearly zero everywhere except around one location.
Do you have any ideas on what the problem could be ?
Thank you,
Célestine Angla
Could you post a small example that reproduces the problem? Note, the C++ codes don’t support the alpha_mode inputs.
Thank you for your quick answer.
I didn't know that the C++ code didn't support the alpha_mode. As I set it usually to "no_dispersion", I guess this is the reason why the results are so weird (probably numerically unstable).
But as the GPU version works for configurations where alpha_coeff = 0.001 for the whole computation zone, does it means that dispersion is only taken into account when alpha_coeff is set to another value for the medium of interest ? How does alpha_coeff affect the dispersion computation ?
I don't know how to provide you my source field and computation zone through this forum, but I do simulations through a homogeneous medium with rho=1600kg/m3 and c=2400m/s at a frequency around 1MHz.
If the C++ code does not support alpha_mode inputs then I guess it is impossible to model a medium with absorption and without dispersion ?
I have another question : in the paper "Benchmark problems for transcranial ultrasound simulation:
Intercomparison of compressional wave models" by Aubry et al., I read that you used a 3D version of k-Wave optimized for high performance computing clusters, is it based on the same C++ code as kspaceFirstOrder3DC and kspaceFirstOrder3DG ?
In this paper, did you modeled only absorption without dispersion ? If so, how did you do it ?
Thank you.
Please, does anyone know how absorption was modeled in the benchmark configurations of the paper "Benchmark problems for transcranial ultrasound simulation" ?
Is it possible to model absorption without dispersion with the 3D version of k-Wave optimized for high performance computing clusters ?
Thank you
If your source has a single frequency, the frequency dependent power law can be set to anything, as long as the value for alpha_coeff is adjusted accordingly to give the correct absorption in dB/cm. To avoid dispersion, you can set alpha_power to 2. This is the approach I used for the intercomparison simulations.
The code used is the same one as publicly available, except for the very large 3D simulations which were done using the MPI code.
Makes sense, thank you very much for your answer !
Is there an example of how to generate the kwave results for the benchmark data, rather than download it?
Hi djps,
I'll tidy up all the k-Wave intercomparison code when I get a chance and move to a public GitHub repo.
Brad.
That would be great, thanks!
perhaps you could just say how the perfectly matched layers were constructed in the benchmarks?
You must log in to post.