pjemris not working in windows

[Hu Cheng]

Dear Tony and Jemris team,

I installed jemris on windows 10. To speed up simulation, I installed MPICH2 (mpich2-1.4.1p1-win-x86-64) on windows successfully according to the instruction. I also changed the path in JEMRIS_sim.m so that localhost (parallel) was enabled. However, the simulation could not complete. So I tested "pjemris" on the cmd prompt and it complained that "the program can't start because mpich2mpi.dll is missing from your computer". But this file is actually in C:\Windows\system32. I copied this file to C:\Program files (x86)\jemris\, that error was gone but pjemris still failed without specific cause.  

Do you have any idea to make it work on Windows? I have installed Jemris on ununtu too. The parallel computing is working but I have more CPU cores on my Windows machine. 

Thanks,

Hu

[tony.s...@gmail.com]

Dear Hu,

which JEMRIS version are you using? Since version 2.9 windows is no longer supported. Some of the new features are not compatible. However, it seems that WSL (Windows Subsystem for Linux) is a good solution to compile and run JEMRIS on windows. I have not tried myself, but I was told that it works well. 

In case you used the JEMRIS 2.8.3 binaries: they were working for me on Windows 10, but I am not a Windows expert and cannot help much with that. Anyway, I'd recommend to try WSL, since the new JEMRIS version has several important new features.

Best wishes, 

Tony

[Hu Cheng]

Dear Tony,

Thank you for your reply. I use 2.83 on Windows. I will try it on WSL. 

Now I have another problem and hope you can help. I'm interested in the noise behavior. So I tried to run simulation with added random noise. The behavior of jemris on Windows is as expected. However, something unual happends with jemris 2.9.0 on my linux machine (ubuntu 20.4). First, parallel computing returns noise-free signal even with noise added. Only sequential computing returns noisy data. Second, noise amplitude is different for Mx and My. The ratio of Noise_Mx/Noise_My is around 1.7. The ratio is 1 from jemris 2.8.3 on Windows.

I tried two different installations on Linux (from the deb file and building the source codes downloaded from github), the results are the same. 

Maybe I did something wrong on linux? I assume the noise amplitude is equal for Mx and My. Can I can just add noise on the simulated signal?

Thanks!

Hu 

[tony.s...@gmail.com]

Dear Hu,

Thanks for reporting, these were two bugs in version 2.9.0!  Both are now fixed in the new release 2.9.1 (parallel jemris now correctly adds noise, and noise has again the same scaling on both axes)
Best wishes,

Tony