[Libmesh-users] LinearImplicitSystem and LinearSolver Methods

[Barna B. <bar...@gm...>]

Hello,

I have a problem, where I need to solve a linear system in a transient simulation. However, the system I solve has no transient character, i.e. in each time step I just need to solve a new system.
Anyway, at some point of my simulation, it seems like the result of my linear system becomes NaNs. I then wanted to check why this could be the case by using methods such as
LinearImplicitSystem::n_linear_iterations() and LinearImplicitSystem::final_linear_residual() after having called the LinearImplicitSystem::solve() method.
However, the two functions return 1 and 0 consistently. It seems to me that something does not get set properly, or perhaps I am missing something. Also calling
LinearSolver::get_converged_reason() gives me a segfault. I would have liked to get some info as to why the solve “fails”/why I get back NaNs.

Thank you for your help.

Barna
 <https://www.youtube.com/channel/UCQiVLRF9ub-T8sW-RKIULMA>