Stéphane MOTTELET

OK thanks, so let us the issue open ok ?

Thanks, then is my affirmation _when rhsQS =NULL the quadrature variable is implicitely defined as the derivative w.r.t. parameter p of the "regular" quadrature variable not depending on sensitivity (hence...

OK I have read 5.2.6.4 already. However, I do not want to start a dialectic discussion (not being a native english speaker), but it seemed (to me) the only possible...

Thanks !

Hi. Just a remark about OpenMP in Sundials and its potential interest. It is not true that OpenMP can speed up only nonlinear solver part. For very large (typically a...

a 10^6 eqs example (100x100x100 nodes). Thanks for the profiling idea I will do that this Monday !

Hello, with this profiling (tests with SUNDIALS 6.2.0 on a OLD Linux server) ```` CVode 99.95% 34.415714s 34.415714s 1 SUNNonlinSolSolve 84.63% 29.141854s 29.141854s 183 SUNLinSolSolve 73.17% 25.194771s 25.194771s 190 N_VLinearSum...

You can get the source there: [suncube_pt.cpp.zip](https://github.com/LLNL/sundials/files/9700113/suncube_pt.cpp.zip) I use threads in the RHS computation (with std::thread) and also via the NVector. A dedicated SUNProfiler is created and attached. I tested...

Thanks for pointing my mistakes. As you said the variability of results is due to the combination of non-associativity of IEEE754 addition together with the unpredictability of thread calling order....

Next version of Scilab (2024.0 in october) will give access to almost all SUNDIALS solvers (idas will be interfaced later) and preview builds are available here: https://www.utc.fr/~mottelet/scilab_for_macOS.html. I have started...