Christian Schilling
Christian Schilling
Do you mean `SupportFunction
Okay. I created #707 for my understanding :man_shrugging:
The line segments would store every vertex twice. A more efficient representation would be with only the vertices in the right order (`Plots` accepts the points both in clockwise and...
The reason for the difference is that we use different implementations of the support function. ```julia julia> ρ([1.], (X ∩ (Y ∩ Z))) 5.0 ``` `X` is an `AbstractPolytope`. Hence...
The original issue in the OP is resolved, but the problem in the second post is still present. Below is another (simpler) example where the support-function approximation is strongly underestimating...
The algorithm can return a certificate, which is useful when applied in a loop with similar sets. > If the positions in the new frame are close to those in...
An alternative is to use the functionality for controlling the backend through `LazySets`. ```julia julia> ReachabilityAnalysis.LazySets.set_exponential_backend!(ExponentialUtilities) Val{:ExponentialUtilities}() julia> ReachabilityAnalysis.LazySets.set_exponential_backend!(Expokit) Val{:Expokit}() ``` I have prepared #704. The only thing to consider...
I do not see the benefit here. The expression on the right looks more complicated to me.
Wow, 80 dimensions. I am impressed that the method can handle that. > I tried saving after over-approximating with `Zonotope`′s and it is better ... but still 14.28 gb lol...
@willsharpless: If you want the infinity norm, then the box approximation is what you want. There is a function `norm` of a set, but it is rather naive and uses...