Twan Koolen

OK, so this appears to be because: ```julia julia> Tracker.track(identity, rand(3)) |> typeof |> supertype AbstractArray{Float64,1} ``` while ```julia julia> Tracker.track(identity, rand(3)) |> eltype Tracker.TrackedReal{Float64} ``` So in a sense,...

After #571, I think the next issue is with ```julia using Tracker using StaticArrays x = Tracker.track(identity, 1.0) y = [Tracker.track(identity, 2.0), Tracker.track(identity, 2.0)] @show typeof(x * y) ``` which...

Also wanted to point to https://github.com/tkoolen/RigidBodyDynamicsDiff.jl, which, though experimental, has moderately optimized gradients w.r.t. `q` (not `v` yet) that I'd expect to be significantly faster than what reverse-mode AD in...

(away from a computer) I actually think that particular error is just due to the `println` you added (nothing in RBD.jl core code should use any kind of `IOBuffer`). I...

With MKL instead of OpenBLAS this is not an issue. I suspect https://github.com/xianyi/OpenBLAS/issues/1622.

@brada4, thank you for looking at this. Here's what I did (verbose, mostly just to make this reproducible for myself). I created a `test.jl` file containing ```julia using RigidBodyDynamics using...

OK, thanks, I'll track that issue.

OK, here's an even simpler example, tracking a sinusoidal reference trajectory for a single body moving along the y-axis, simulated using `RigidBodyDynamics.simulate` and animated using `MeshCatMechanisms` (showing the inertial ellipsoid)....

Not currently, no. It's been in the back of my mind for a while, but it's not a super high priority item for me personally right now. A PR would...

I started exploring SDF support a bit, but I found that the behavior of SDF's `` tags is currently very ill-defined. I think I'm going to wait at least until...