Marcelo Forets

One has to extract the vector field from the initial-value problem and construct the corresponding `ODEProblem`. Additional arguments and problem options can just be passed through. With the idea of...

Exactly, the idea of using `MathematicalSets` was that it should provide the necessary functionality; then eg. LazySets implements the higher-level interface defining its `universe` function.

> After thinking everything through more again more carefully, I think you are right. > > In order for me to understand what is going on, I like to write...

`Distributions.jl` has a macro for argument checking, which is triggered on/off using a keyword argument: see [macro](https://github.com/JuliaStats/Distributions.jl/blob/345d1e085916134901d6ffac9ed31aeee0d0ee83/src/utils.jl#L3), and [example]( https://github.com/JuliaStats/Distributions.jl/blob/master/src/univariate/continuous/uniform.jl#L32).

This is also a source of bugs, e.g. in the following code, the struct is built, but the guard is ignored (!): ```julia julia> guard_1 = Universe(3); julia> t1 =...

I agree to add a linearization routine, ideally through existing tools as you propose. I have previously written some code that was using TaylorSeries for the computations, but i would...

And how about > At the same time, it has less structure, because it does not speak about "constraint on what" (state or input). ?

We were interested in floating-point computations and discarded using the Jordan normal form because it is ill-advised (ref. to the comments in that thread). However, to compute that integral for...

Very neat! Yesterday I thought it would be more involved, but you are right, maybe we just have to add one new condition in line 289, ie. be sure that...

Nice! There is no rush. I think adding unit tests as you propose is more than enough. > Are there any other corner cases that are not working? Do you...