Christopher Rackauckas

I think the best thing to do would be to use a DiscreteCallback and basically do event location yourself inside of it. The function that you need to do the...

>The event handling support in DifferentialEquations.jl is very robust and so far so great! Unfortunately it's not 😢 and we're working on it at least. There are some weird bracketing...

Yeah it works for most cases. People do weird stuff with floating point numbers though... library writing teaches you crazy stuff about things like `1e-6 - 1f-6`...

`integrator(t)`

The straddle interval is `[integrator.t,integrator.t+integrator.dt]`. >Also is there a way to get the parameters of the interpolant? That will be necessary to do this in closed-form. A lot of the...

>However I have a situation in which my dynamics function is very expensive to compute. Therefore I'd like to minimize the number of root finding steps to the bare minimum....

Ahh, this is really cool! A few thoughts on this. Sorry this got long, but let me split this. ## Extrapolation in ODEs: OrdinaryDiffEq.jl `perform_step` The first thought would probably...

>That is, how would you set the timestep for DGLDDRK84_F now? Presumably you're going to halve the timestep until the answer stops changing to your desired tolerance. If you're going...

I see I was thinking steps, and that has different use cases. You're thinking the whole solve, which would be used to get global error estimates and global tolerances. It...

I created GlobalDiffEq.jl with your starter code in there an enough to make it conform to the common interface, and opened a bunch of issues to essentially ready it for...