David Widmann
David Widmann
Hi! The changes in Roots 2 broke root finding with non-standard number types in Bijectors: https://github.com/TuringLang/Bijectors.jl/pull/219 The reason seems to be that in Roots < 2 `find_zero(f, (a, b))` was...
I couldn't find an existing issue so I open a new one here. The ColVecs/RowVecs abstraction would be useful in other packages as well that otherwise don't want or need...
As it says :wink: The PR implements a modification of the suggestion in https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/issues/384 that is non-breaking and (hopefully) avoids most of the method ambiguity issues that I think could...
There are mainly two things that can be improved in the current implementations of `show`: one can remove string interpolation and implement `show(::IO, ::MIME"text/plain", x)` for more verbose multi-line output...
This PR implements https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/issues/299#issuecomment-943761371. See the comment and the additional explanations in the issue for more details. This PR is still WIP and tests will fail (I assume).
This PR adds the inverse multiquadric kernel ```math k(x, x'; alpha, c) = (c + d(x, x')^2)^(- alpha) ``` with parameters `alpha, c > 0`. In recent years, in particular...
It's late and I haven't thought much about it - but maybe we could unify the different kernel matrix functions by using ```julia # maybe the default dispatch should depend...
If one defines `isequal` one has to define `hash` as well since otherwise the two are not consistent. In general, if one does not care about floating point comparisons one...
Copied from Slack: BTW I propose to remove the IndependentMOKernel ane the IntrinsicCoregionMOKernel since they are just KernelTensorProducts with a WhiteKernel and a yet to define MatrixKernel (maybe a bad...
It is unclear to me why `set!` and `duplicate` are needed. Could they be removed?