use QR for linear subsolver in parametric RLM#1859
Conversation
| jacobian_tangent_basis = DefaultOrthogonalBasis(), | ||
| initial_residual_values, | ||
| initial_jacobian_f, | ||
| linear_subsolver!, |
There was a problem hiding this comment.
@mateuszbaran, can you please help me with how to use a custom linear_subsolver! in LevenbergMarquardt!. I couldn't quite figure out where to pass it in. QR is more numerically stable for SLAM problems so would like to try it out. Also note, ldiv! has an issue with sparse arrays but is fixed in julia 1.12.
"MethodError: no method matching ldiv!(::SparseArrays.CHOLMOD.Factor{Float64, Int64}, ::Vector{Float64})"
There was a problem hiding this comment.
I might also be wrong here with using LevenbergMarquardt! directly?
There was a problem hiding this comment.
Ah, I forgot to put the subsolver as a keyword argument in the LevenbergMarquardt! function. I will add it tomorrow.
There was a problem hiding this comment.
Manopt 0.5.14 (just tagged) has the keyword argument. Let me know if you encounter any further issues.
3 participants
No description provided.