Hi, I'm fairly new to Eigen and ran into a discrepancy between Matlab's 'ldl' and Eigen's LDLT module. It seems the lower triangular Matrix returned by the Eigen LDLT module (matrixL()) has a row and column swap when compared to Matlab's 'ldl' subroutine. There's also a swap in the diagonal elements between the two. My input matrix is symmetric, A = A^T. Is there another decomposition I should be using to get results that are inline with 'ldl' ?

Our LDLt does column pivoting by default that is more stable. there is patch hanging somewhere adding a NoPivotingLDLT class...