Warning: "Preconditioned conjugate gradients method did not converge" in notebook 90-full-analysis about diva-workshops HOT 3 CLOSED

Hello Javiera,
the warning comes from the conjugategradient( ) function, this function is used to solve linear system such as A x = b. The message indicates there was a problem in the convergence of the iterative method.

It is possible to use another type of methods to solve the system: https://gher-uliege.github.io/DIVAnd.jl/stable/#DIVAnd.DIVAndrun

inversion: direct solver (:chol for Cholesky factorization), an interative solver (:pcg for preconditioned conjugate gradient [1]) can be used or :cg_amg_sa for a multigrid method with preconditioned conjugate gradient. The two last methods are iterative methods who a controlled by the number of iterations maxit and the tolerance tol.

or to increase the number of iterations (parameter maxit).

The reason why it did not convergence it not clear to me. Can you show a plot with the data distribution?

from diva-workshops.

Hello,
thank you, Charles. I will try to solve it with the suggestions you've made.
I attached figures of data distribution and bathymetry. Furthermore, I think that the resolution of latr and lonr may be too fine (since it's a small domain) and it's causing this issue.

dx = 0.0125 dy = dx lonr = -74.:dx:-72. latr = -52.5:dy:-51.

Cheers,
Javiera

from diva-workshops.