The effects of integrator accuracy

Provide easy-to-use, extensible software for modeling, simulating, controlling, and analyzing the neuromusculoskeletal system.
POST REPLY
User avatar
Yuichiro Omura
Posts: 44
Joined: Thu Apr 04, 2019 10:03 pm

The effects of integrator accuracy

Post by Yuichiro Omura » Thu Mar 10, 2022 8:23 pm

Hi.
I perform forward dynamics simulations of human standing using C++ API.
I thought of changing the integrator accuracy to accelerate simulations, but I wonder how the integrator accuracy affects the forward simulation results.
If I decrease the accuracy too much, will I get strange simulation results?
Also, is there recommended integrator accuracy? (or does it depend largely on the problem?)
Now, I perform simulations with the accuracy 1.0e-4.

Best regards,
Omura

Tags:

User avatar
Carmichael Ong
Posts: 401
Joined: Fri Feb 24, 2012 11:50 am

Re: The effects of integrator accuracy

Post by Carmichael Ong » Sat Mar 12, 2022 5:18 pm

Yes, if the accuracy is set too low, your results will start to differ and become an issue. Valid settings will depend on the problem, with a general rule that tighter accuracies (i.e., smaller accuracy number) may be needed for higher impact motions (e.g., running vs walking).

One way to try to figure this out for your problem is to do a sensitivity analysis, by varying your accuracy and seeing how different the results are. As you tighten the accuracy, the results should become similar run to run, and then you can find the balance between speed and accuracy,

User avatar
Yuichiro Omura
Posts: 44
Joined: Thu Apr 04, 2019 10:03 pm

Re: The effects of integrator accuracy

Post by Yuichiro Omura » Wed Mar 16, 2022 12:21 am

Thank you for your reply.
I understand. I'll try to do a sensitivity analysis.
That helped me a lot, thank you!

POST REPLY