# SVD: Better default to gesvd! instead of gesdd!?

**URL:** https://discourse.julialang.org/t/svd-better-default-to-gesvd-instead-of-gesdd/20603
**Category:** Internals & Design
**Tags:** linearalgebra
**Created:** [February 9, 2019, 11:17am UTC](https://discourse.julialang.org/t/svd-better-default-to-gesvd-instead-of-gesdd/20603 "2019-02-09T11:17:53Z")
**Posts on this page:** 1
**Showing post:** 20

<div class="post-metadata">

### Author: ![Ralph\_Smith](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ralph_smith/32/10344_2.png) [@Ralph\_Smith](https://discourse.julialang.org/u/Ralph_Smith)
#### Post date: [February 10, 2019, 8:51pm UTC](https://discourse.julialang.org/t/svd-better-default-to-gesvd-instead-of-gesdd/20603/20 "2019-02-10T20:51:55Z")

</div>

On looking deeper, I see that LAPACK also has a Jacobi method for the SVD (as mentioned in [Calling Lapack's Jacobi SVD - dgesvj](https://discourse.julialang.org/t/calling-lapacks-jacobi-svd-dgesvj/20602)) This approach is not mentioned in the Users’ Guide, but (often?) gives singular values with _elementwise relative_ accuracy close to unit rounding error. It works for all of the s.v. for the companion matrices described above until N\>64 where things are so bad that `NaN`s are generated.

---

_[View the full topic](https://discourse.julialang.org/t/svd-better-default-to-gesvd-instead-of-gesdd/20603)._