# Possible bug in KrylovKit.eigsolve

**URL:** <https://discourse.julialang.org/t/possible-bug-in-krylovkit-eigsolve/55962>\
**Category:** Performance\
**Created:** [February 24, 2021, 9:34pm UTC](https://discourse.julialang.org/t/possible-bug-in-krylovkit-eigsolve/55962 "2021-02-24T21:34:01Z")\
**Posts on this page:** 1\
**Showing post:** 2

<div class="post-metadata">

**Author:** ![fgerick](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/fgerick/32/13228_2.png) [@fgerick](https://discourse.julialang.org/u/fgerick)\
**Post date:** [February 24, 2021, 9:47pm UTC](https://discourse.julialang.org/t/possible-bug-in-krylovkit-eigsolve/55962/2 "2021-02-24T21:47:58Z")

</div>

Okay to be fair, the first post you made was not very delicately formulated also. I agree that sometimes discussions here tend towards nitpicking the way questions are formulated, but you can’t go ahead and say things like

> [@Bug detected in Arpack](https://discourse.julialang.org/t/bug-detected-in-arpack/55334/1):
>
> A language meant to do high performance computing should be able to find 4 eigenvalues of a 16X16 symmetric matrix. I could do it myself on pen and paper.

without people getting upset.

I’m sure for the matrices you are considering there is a good algebraic reason, why trying to find the eigenvalues of smallest real part is causing issues. But I’m not qualified to judge that. `KrylovKit` seems to give a reason already for the Matrix in this thread here.

If you’re looking at such small matrices, you can always use the standard library package `LinearAlgebra` and use

```julia
julia> using LinearAlgebra

julia> sort(real(eigvals(K)))[1:4]
4-element Vector{Float64}:
 -1.1102230246251565e-16
 -1.1102230246251565e-16
  3.9999999999999982
  3.9999999999999982

```

---

_[View the full topic](https://discourse.julialang.org/t/possible-bug-in-krylovkit-eigsolve/55962)._