I'm solving least squares problems using Eigen.
here https://eigen.tuxfamily.org/dox/group__DenseDecompositionBenchmark.html is a good comparison of different solvers to solve for A*x=y in a least squares sense.
I have a lot of very similar problems where the solutions differ only slightly. Can I accelerate some (or all?) of these solvers by initializing them somehow with the close result I already have?
If so, how?