info on reference solutions

[jandy123]

I've noticed that all siconos data sets contain a reference solution. Is there any information how this solution is obtained? How accurate is it?

Also, the LMGC90 data sets do not contain such reference solutions, which, provided they are accurate, are very useful to have. Is there a way to include such solutions also in these data sets?

Thanks for the great work!

[vacary]

Apologies for the late reply
The reference solution is generally calculated using a first-order method (NSGS) with a fairly coarse tolerance (1e-4). As most problems have multiple solutions for contact forces, it is quite difficult to reuse them to test an algorithm or make comparisons.

One would need to be able to use an algorithm capable of providing highly accurate solutions to these problems in order to use these reference solutions. I was actually wondering whether I should remove these problem solutions.
For your information, I am looking for new problems. Your contribution would be welcome.

[jandy123]

Thank you for your reply! I will get back when we can contribute with new problems.

[vacary]

👍