In this algebra, it’s possible to compute on a mesh of arbitrary 5 dimensional conformal geometric algebra simplices, which can be represented by a bundle of nested dyadic tensors. julia> using Grassmann, StaticArrays; basis"+-+++" (⟨+-+++⟩, v, v₁, v₂, v₃, v₄, v₅, v₁₂, v₁₃, v₁₄, v₁₅, v₂₃, v₂₄, v₂₅…

This doesn’t actually matter for my use case though. I don’t need the output to be ===. It only generates some symbols which get pasted into a function.

The @pure annotation tells the compiler that you guarantee the output to be ===. If you don’t have or need that guarantee, then you should let compiler know by removing the annotation.

Alright then, I will remove that annotation in this particular case, but there many valid use cases of it in other places of my package.

Julia has gotten significantly better at constant propagation since you first wrote this, so I would be surprised if many of them actually give you a performance benefit anymore in any case.

[image] chakravala: Alright then, I will remove that annotation in this particular case, but there many valid use cases of it in other places of my package. At the risk of sounding like a breaking record, this is probably not true, mostly just because there isn’t really a way to use @pure cor…

[image] Keno: The point people are trying to make here is that it’s best to avoid @pure because the compiler makes zero guarantees about what it means, so if you don’t 100% need it, you’re better off without it. Even if nothing breaks right now, future versions of Julia can and will break and cr…

[image] Per: I’m only guessing, but I think that the reason why "return value allocation is definitely not pure " is that it depends on where heap memory is available, and this is is a global state. Example: f() = [] f() === f() # returns false The function f is not pure, since calling it wit…

[image] Mason: This seems especially worrisome since you do have a tenancy to advertise Grassmann.jl as something people should adapt their project to make use of, so if someone did act on that it could result in an unfortunate situation. It is clearly stated in my repository that there is no…

[image] chakravala: Those who are serious about using Grassmann.jl should consider consulting with me. Hum, too bad. I was really excited by your post, because I have to solve systems like A\b in two occasions: interpolating orbit and inside the Kalman filter (which applies to the attitude an…

@Ronis_BR , I don’t think you will encounter a major issue with the A\b method, would be interesting to see what you try out! Another interesting aspect would be using bivectors and quaternions for that also. As @Mason mentioned, what you probably don’t want to do is overload internal Grassmann metho…

Grassmann.jl A\b 3x faster than Julia's StaticArrays.jl

General Usage Performance

dpsanders June 19, 2020, 9:31pm 32

It seems easy enough to do a global find and replace to remove the @pure and see what happens to the benchmarks.

Topic		Replies	Views
[ANN] Grassmann.jl : Differential geometric algebra Package Announcements package , announcement , differentiation , array , linearalgebra	23	10156	December 22, 2019
Computing Inverse of a stack of matrices General Usage performance , linearalgebra	19	3945	August 29, 2020
[ANN] CliffordAlgebras.jl Package Announcements math , algebra , geometry , geometric-algebra	19	2662	November 30, 2021
Paper for discussion: The Linear Algebra Mapping Problem Internals & Design	13	1680	November 25, 2019
Static array multiplied by its adjoint seems to allocate Performance linearalgebra , memory-allocation , staticarrays , allocations	12	438	December 2, 2024

Grassmann.jl A\b 3x faster than Julia's StaticArrays.jl

Related topics