Hi,
I am using TVB to generate NMM data, which is then used to train a DL model (DeepSIF). The code used is from github (
https://github.com/bfinl/DeepSIF), called generate_tvb_data.py.
When I run the code, it takes about 2 hours to run 200 sec of simulation.
When using run in parallel (given in the code), it still takes 3 hours to generate data for 4 regions (each 200 sec). I find this a long time, so wanted to check if maybe something is wrong.
I ran
The code in Github contains the connectivity file for 76 regions, but I used the file for 998 regions (called connectivity_998.zip), which can be found here (
https://github.com/bfinl/DeepSIF/tree/main/anatomy). I uncommented the following lines (60-64) in the code.
# data[:, 7] = data[:, 994]
# data[:, 325] = data[:, 997]
# data[:, 921] = data[:, 996]
# data[:, 949] = data[:, 995]
# data = data[:, :994]
Is it normal that it takes so long to generate NMM data? And would anyone be able to take a look and tell if there is a way to speed up the generation of data? Thanks in advance.
Best regards,
Michiel Aten