ForwardDiff.jl

ForwardDiff.jl copied to clipboard

one(::type(Dual))

1

Currently, `one` returns a `Dual` number. It seems like it could simply return the underlying value type instead. i.e. change [these lines](https://github.com/JuliaDiff/ForwardDiff.jl/blob/4b143a199f541e7c1248dc5bc29b397be938e81d/src/dual.jl#L356-L357) to ```jl Base.one(d::Dual) = one(typeof(d)) Base.oneunit(d::Dual) = oneunit(typeof(d))...

stevengj

`NaNMath` (and `SpecialFunctions`) as extensions?

5

in particular, this code seems really weird: https://github.com/JuliaDiff/ForwardDiff.jl/blob/226374984ec70b96c89053640b267a5c5543781d/src/dual.jl#L463-L477 in particular because `NaNMath` is loaded beforehand: https://github.com/JuliaDiff/ForwardDiff.jl/blob/226374984ec70b96c89053640b267a5c5543781d/src/ForwardDiff.jl#L10 The same thing can be said about `SpecialFunctions`. I noticed that `DiffRules` also depends...

longemen3000

Differentiation with modulo function yields a Float, even for Rational input. What doesn't work: ``` ForwardDiff.derivative(x->mod(x^2,1), 1//3) 0.6666666666666666 ``` What works: ``` ForwardDiff.derivative(x->x^2, 1//3) 2//3 mod(1//2,1) 1//2 ```

daniel-frisch-kit

Hi, the code in the linked repository gives slightly different optimization results on an Intel machine vs an AMD machine when using ForwardDiff. It does not matter if the operating...

elbert5770

Error differentiating through complex power due to isinteger

6

I have a weird case where a complex power is giving me an inexact error because the `_cpow `helper function determines my object to be an integer. Truncated error message:...

tom-plaa

add back the iteration based seeding for `Array`

2

These were removed in https://github.com/JuliaDiff/ForwardDiff.jl/pull/472 and while I was a bit uncomfortable with it I did not protest. However, now after doing some benchmarks I see that this has a...

KristofferC

Derivative is wrong for this inverse quadratic form

1

The result is correct only if I tag the matrix to be symmetric. ```julia using LinearAlgebra using ForwardDiff function f_backward(x1, x2, rho) cov = [ 1.0 rho; rho 1.0 ]...

colinfang

Is this code a supported use? (single-pass value and derivative)

6

Considering these two methods that compute the value and first derivative of a scalar function in a single pass: ```jl import ForwardDiff import DiffResults @inline function value_and_derivative(f, Y::Type, x::Real) diffresult...

gerlero

One call to get f(x) and f'(x)?

2

Excellent package! Is there a way to get the function value and its derivative in one call? Thanks.

aloispichler

Violating assumptions about Dual comparisons

8

There are already a few issues talking about #481 (#606, #607) and related problems (#480). I've been running into breakage hinted at in [this comment](https://github.com/JuliaDiff/ForwardDiff.jl/pull/481#issuecomment-1171251680) and [this workaround](https://github.com/JuliaDiff/ForwardDiff.jl/issues/607#issuecomment-1314299091) where ```jl...

gpeairs