-
Seems completely unexpected:
```
julia> A
4×4 Matrix{Float64}:
0.839128 0.0446467 0.631044 0.142581
0.517163 0.541462 0.988977 0.550456
0.515481 0.448386 0.0857754 0.808353
0.469357 …
-
As mentioned in #31738, it would be good to have specialized eigensolvers for upper-Hessenberg matrices (i.e. our `UpperHessenberg`) type.
In particular, Hessenberg factorization is the first step …
-
There are a handful of eigenproblem decompositions that I would like to use but are not provided by PyTorch (e.g. Hessenberg and Schur). This is one area where PyTorch noticeably lags behind other app…
-
Our current `ipu_eigh` uses the Jacobi algorithm. It is believed for other hw accelerators that the QR algorithm becomes faster than jacobi for larger `d>=512`. Since we are targeting `d>=512` we are …
-
I heard of the following bug report through William.
[Bug report]
Simple 8X8 matrix determinant computation makes sage hang:
```
def genVar(i):
return "x%i"%i
def matrix_from_hash(h):
R=…
-
Maybe there is a good reason for the current ordering, but otherwise I'd like to request that the schur function return its outputs in a different order.
The hessenberg function returns its outputs…
-
- [x] arg (phase angle in radians)
- [ ] epsilon
- [ ] erfinv
- [ ] erfcinv
- [x] factorial
- [x] trunc
- [x] pow2
- [x] root
- [x] sign
- [x] nonzeros
- [x] norm
- [ ] device[get, push, po…
-
Sorry for spamming :laughing:
But copy of a CUDA plan does not work:
```
p = NFFT.plan_nfft(coords, (size(x,1), size(x,2)))
MethodError: no method matching copy(::CuNFFT.CuNFFTPlan{Float32, …
-
**Description**
* DGEHRD reduces a general matrix A to upper Hessenberg form
* DHSEQR computes eigenvalues of a matrix of Hessenberg form and the Schur form
but in dget38.f, comments are only "…
-
A number of functions in this package and in [RobustAndOptimalControl.jl](https://github.com/JuliaControl/RobustAndOptimalControl.jl) fail when trying to compute gradients through them using ForwardDi…