FiniteDifferences.jl

FiniteDifferences.jl copied to clipboard

Closes #206 I am tempted to generalize this from `Bool` to `Integer`, matching ChainRules's convention. If we do then we can at some point in the future workout a way...

oxinabox

Cross derivatives of a multivariate function?

3

Hello, I hope all is well. I have a question about the title. I am interested in computing a $M\times N \times N$ three-way array of derivatives: $\frac{\partial^{2}f_{i}}{\partial x_{j}\partial x_{k}}$...

daisukeadachi

ref https://github.com/JuliaDiff/ChainRulesTestUtils.jl/issues/258 https://github.com/JuliaDiff/FiniteDifferences.jl/blob/5c2979e7b84cb000ead2e256c2c5bdba005b0d22/src/to_vec.jl#L36-L57

vpuri3

the formulea for calculating derivatives of different error orders

5

Sorry to misuse issues to ask questions, but I could't find any other way to communicate with you developers. I am using this package to calculate the first order derivative...

thudjx

reproducer: ```julia using FiniteDifferences,LinearAlgebra;SparseArrays,Random grad(central_fdm(5,1),norm,B) I_n = spdiagm(ones(n)) Random.seed!(0) B = sprand(n, n, 0.5) * 0.2 + I_n FiniteDifferences.grad(central_fdm(5,1),norm,B) ``` it gives: ``` ERROR: TypeError: in new, expected Int64, got...

longemen3000

make `to_vec(::Integer)` an empty vector

4

Closes #188 breaks a few rules in ChainRules, need to take a look at that before merging

mzgubic

Fix CI Cache bug. add x86 jobs

4

There was a CI Cache bug affecting the previous ci files (found in a different project). there was also an exclude for macOS x86 jobs, but there were no x86...

fchorney

Decrease allocations

2

Turns out type-stability was causing a bunch of allocations. Type inference was failing for some reason. Also tracking history surpisingly, causes allocations. We don't need to track history when doing...

oxinabox

Error if given nonfinite primal?

2

We can't really do finite differencing around `Inf` or `NaN` can we?

oxinabox

Should `to_vec(::Integer)` return an empty vector

8

Integers can't be perturbed, so their `to_vec` should be empty. Originally suggested in https://github.com/JuliaDiff/FiniteDifferences.jl/pull/187 ```julia function to_vec(x::Integer) Integer_from_vec(v) = x return Bool[], Integer_from_vec end ``` Would also fix (a part...

mzgubic