Poor Stability of Solve

[michaellindon]

I am trying to do a linear solve with a poorly conditioned, but not terribly conditioned, matrix Sigma (symmetric, positive definite)

The eigenvalues of Sigma are

(:S (la/svd Sigma))
#vectorz/vector [0.00944222396770215,0.007705139713474439,0.006118795643968743,0.003412358080515295,0.0031039370984308505,0.0024409586692006042,0.0023054287904142974,0.0019385205264753872,0.001546173882624919,2.480022617777184E-4]

When I do my linear solve (la/solve Sigma Bols), the result is nil. I am using the vectorz implementaiton of core.matrix.

I wouldn't expect to get Nil for such a matrix. Any ideas what is going on here?

[mikera]

Hi I'm not an expert on this part of the code base, looks like this is an issue with vectorz-clj though rather than core.matrix? If so can you file an issue here (https://github.com/mikera/vectorz-clj/issues)?

Do you have a minimal example of a matrix Sigma that this fails on?