Yixuan Qiu

@LTLA I've checked the code and can confirm that the program stalls at the matrix factorization part, which calls the `Eigen::SparseLU` solver in Eigen. In general the sparse matrix factorization...

@LTLA Yes, all these methods are valid in mathematics, but the stability and accuracy may depend on the distribution of all eigenvalues. I agree that it deserves to write a...

Unfortunately at this stage Spectra has no support for complex matrices yet. :slightly_frowning_face: You still need to use packages such as ARPACK.

I see. If `A` is positive definite and `B` is symmetric, then you can use the [SymGEigsSolver](https://spectralib.org/doc/classspectra_1_1symgeigssolver_3_01scalar_00_01selectionrule_00_01optype_00_01boptype_00_01geigs__cholesky_01_4) solver in Spectra.

What kind of error? Compilation? Runtime? Numerical/logical?

This is because you passed a dense matrix to a sparse operator. Try the code below. ```cpp #include #include #include #include #include #include using namespace Spectra; int main() { //...

This means the algorithm does not converge, and you may need to adjust the `ncv` parameter. A larger number leads to a better convergence, but takes more computational time ([documentation](https://spectralib.org/doc/classspectra_1_1symgeigssolver_3_01scalar_00_01selectionrule_00_01optype_00_01boptype_00_01geigs__cholesky_01_4))....

@jschueller Thanks. I'll take a look. @JensWehner Sure. We can discuss the details offline via email or other means. I may not be available to reply promptly as I have...

Hi all, here are some updates. I've created a [1.y.z](https://github.com/yixuan/spectra/tree/1.y.z) branch as a preview of the planned changes in the next major version. A brief changelog can be found [here](https://github.com/yixuan/spectra/commit/d6b33b8961bd90186b96c15f31625b4cb44d0ba6),...

@jschueller Sure. It has been added: https://github.com/yixuan/spectra/commit/a97bf2034dc7e3b84862ae6a3aadf577e54724dc.