Is it possible to convert a regular vector to a ComponentArray with the specified structure?

[ForceBru]

The README shows that a ComponentArray can be indexed like a vector and converted to a vector as well:

https://github.com/jonniedie/ComponentArrays.jl/blob/f3a1b6d9e6d1549f7513579632d66597067eb93b/README.md?plain=1#L44-L58

Is it possible to reverse this operation: convert a regular vector like [1, 2] to, say, a ComponentArray(a=1, b=2) with the given fields and their types? I want something like this to convert a vector to model parameters:

function loss(params_vec::AbstractVector)
    params = toComponentArray(params_vec, example=ComponentArray(a=1, b=2, c=[3, 4]))
    @. params.a + params.b + params.c
end

optimize(loss, [1, 2, 3, 4])

I know that one can pass the ComponentArray as the parameter because ComponentVector <: AbstractVector, but that won't work if the optimizer always calls loss with regular vectors for some reason.

Basically, I'd like to have a function like toComponentArray which receives the source vector and the expected structure of the ComponentArray and returns the ComponentArray that contains the data from the vector. Is this possible?

[ForceBru]

Looks like this is done by ComponentArrays.Axis:

https://github.com/jonniedie/ComponentArrays.jl/blob/f3a1b6d9e6d1549f7513579632d66597067eb93b/src/axis.jl#L20-L27

---

Yep, this seems to work and even has similar interface to what I envisioned:

julia> example = ComponentArray(a=1, b=5, c=[600,1000])
ComponentVector{Int64}(a = 1, b = 5, c = [600, 1000])

julia> ax, = getaxes(example)
(Axis(a = 1, b = 2, c = 3:4),)

julia> ComponentArray([2,3,9,8], ax)
ComponentVector{Int64}(a = 2, b = 3, c = [9, 8])

[ForceBru]

Thanks, me!

[NilsWildt]

Thank you ;)