Marcelo Forets

Results 234 comments of Marcelo Forets

> It worked, however, the file is ~15 gb. Is there a better way to save the objects? whoa, that's huge. my first suggestion would indeed have been to export...

> (happy to share, omitted for brevity), if it is only omitted for reason of brevity, then i encourage you to share it. we could take a look and potentially...

> I expanded the dimension to 40 and the analysis was able to complete at some point over night (wow, well done). i thought the number of taylor series variables...

If the system is 9-dimensional, don't you have out of bounds in `x[N+1]` and other terms? EDIT: Ah, you pass `X0 × D0` as initial states so that should be...

I'm getting some results with a slightly modified version: Code: ```julia using LinearAlgebra, Plots using ReachabilityAnalysis N = 9; nd = N; @taylorize function L96!(dx, x, p, t) # F...

yes, the interval `tspan=(1., 0.)` was meant to compute backwards approximation. iirc DiffEq has a similar behavior.

A simple way to implement this method is to use `IntervalArithmetic.mince` for interval boxes (just think of the IntervalMatrix as an interval box in dimension m x n ). It...

Thanks for your insight. > I think it would be better to separate the domain of the parameters from the parametric structure. Yes; we can do that through struct composition...

once matmul methods are included, we may add the matrix benchs in @lucaferranti's blog

True, that implementation is more or less what i had in mind. It may be convenient to subtype `