Gaurav Arya

It seems like it's getting stuck in between `DiscreteCallback`'s, so that I can't catch the infinite loop in the `DiscreteCallback`. It'll take me a bit to make an MWE, but...

Seconding the idea of allowing a single function to return multiple quantities. Also, to throw another spanner into the works: in my use case, it would be most efficient to...

Perhaps it would be best to just have a separate kwarg? E.g. the user can specifiy `rate_and_bounds(u,p,t)` returning the triplet of `(rate(u,p,t), rateinterval(u,p,t), urate(u,p,t))` (and `lrate` optionally e.g. as a...

I encountered this issue because I was making my own aggregator where I tracked some stateful information within the aggregator and directly extracted `integrator.cb.condition.{some state}` after running `solve!` on the...

Reading around, it seems is that it's expected behaviour in SciML for all the caches etc. to be created upon initialization of an integrator, and if one wanted to avoid...

I'm hitting the same error doing when trying to install `juliaup` because of my cluster disallows lock in the home directory, e.g. here is the FAQ entry describing the behaviour...

For tests to pass with this PR, we also need an `AbstractFFTs` release due to https://github.com/JuliaMath/AbstractFFTs.jl/pull/116. In addition, the (half incorrect) Zygote rules that would be removed here do apply...

That's right. The first call to the primal, as well as subsequent calls, all agree with Zygote's result for Zygote's _subsequent calls_ (i.e. out = `0.177...`)

But if I make **five** calls to the primal without reseting the seed, I get what Zygote's first call gives us! ```julia julia> using Distributions, Random julia> function f(x) out...

I am able to reproduce similar behaviour without `Distributions`: ```julia julia> using Zygote, Random julia> function f(x) #= The sqrt actually makes this not equivalent to the previous sampling procedure....