Simplifying Symbolics.jl expression

[Alseidon]

I’m adding to the list of posts about simplifying a Symbolics.jl expression, but I wanted to check if I did something wrong or if it is an internal limitation. I want to check if two (lengthy) expressions are identical, and am trying to simplify their difference. However, simplify(expr) doesn’t simplify things such as this:

(I ran simplify twice, and a third pass didn’t change anything).

Edit: I found a simplified example.

Maybe the fact that the computed expression’s (b1) terms are not reordered is the problem ?

Edit 2: a MWE

julia> using Symbolics
[ Info: Precompiling Symbolics [0c5d862f-8b57-4792-8d23-62f2024744c7]

julia> @variables a b c d e f
6-element Vector{Num}:
 a
 b
 c
 d
 e
 f

julia> (a^2 + b^2) * cos(c + d) * e + e * cos(c + d) * (-a^2 - b^2)
(a^2 + b^2)*e*cos(c + d) + (-(a^2) - (b^2))*e*cos(c + d)

julia> (a^2 + b^2) * cos(c + d) * e + e * cos(c + d) * (-a^2 - b^2) |> simplify
0

julia> (a^2 + b^2) * cos(c + d) * e * f + e * cos(c + d) * (-a^2 - b^2) * f |> simplify
(a^2 + b^2)*e*f*cos(c + d) + (-(a^2) - (b^2))*e*f*cos(c + d)

I suspect the problem here is the terms in the computed expression (b1) are not reordered.

[greatpet]

@Alseidon

julia> (a^2 + b^2) * cos(c + d) * e * f + e * cos(c + d) * (-a^2 - b^2) * f |> simplify
(a^2 + b^2)*e*f*cos(c + d) + (-(a^2) - (b^2))*e*f*cos(c + d)

Just change simplify to expand, and the output becomes 0.
It’ll be great to have a smarter simplify, but in this case, expansion is clearly the solution.

[Alseidon]

@greatpet

Just change simplify to expand, and the output becomes 0.

This was indeed the solution for the two last examples, thank you!

For the first one (quoted below), it didn’t work: differences of identical terms do not cancel out.

@Alseidon

However, simplify(expr) doesn’t simplify things such as this:

I got it working with the same expression on another Pluto notebook though, it might just be an error I didn’t spot.