Yixuan Qiu

This is related to what I call a sequential PCA: http://ieeexplore.ieee.org/abstract/document/8258403/. The idea is actually very simple: for a matrix `A`, if you have computed the largest `k` eigenvalues `Dk`...

@shivupa Pull requests are always appreciated. :smile:

Hi @Chris-Hickey-Arup, PardisoLLT seems to rely on external software, so I didn't include it in Spectra. One thing you can do is to write a wrapper class similar to [SparseCholesky](https://spectralib.org/doc/classspectra_1_1sparsecholesky),...

Great. And also note that the matrix in your example is not symmetric.

It is based on the Implicitly Restarted Arnoldi Method, first implemented by the [ARPACK](http://www.caam.rice.edu/software/ARPACK/) package. I think it is not straightforward to gain general improvement on the algorithm, but if...

It is possible to use different solvers to implement the `inv(A - sigma * I) * x` operation. Basically you just modify the `SparseSymShiftSolve` class to use other solvers such...

One more question before I try to write a solver for you: do you only need the smallest eigenvalues such that `sigma` can be assumed to be zero, or do...

Oh, I should have explained my question more clearly. If you only want the smallest `k` eigenvalues then it corresponds to the case of `sigma = 0`, and if you...

You can try the example attached below. Note that in this case you have to make sure that the matrix is positive definite. [cg-example.zip](https://github.com/yixuan/spectra/files/858422/cg-example.zip)

@bodhi1991 Do you have any success on this?