Simplifying floating point exponents sometimes doesn't cancel terms

[MarcBerliner]

E.g.

julia> @variables x y
(x, y)

julia> simplify(x^2.0/(x*y)^2.0)
x ^ 2.0 * inv((x * y) ^ 2.0) ^ 1

compared to

julia> simplify(x^2/(x*y)^2)
inv(y) ^ 2

Removing the parentheses works as well

julia> simplify(x^2.0/x^2.0/y^2.0)
inv(y ^ 2.0) ^ 1

[ChrisRackauckas]

@shashi you know about this one?

[YingboMa]

This is because to simplify this expression, it needs to expand (x*y)^2 to x^2 + y^2

julia> simplify((x*y)^2)
(x ^ 2) * (y ^ 2)

When the exponent is not an integer, we assume that this transformation will be too expensive to be beneficial. This can be fixed by using the isinteger check instead of isa Integer.

[YingboMa]

It's still highly not recommend to write .0 in Julia unnecessarily. It doesn't make anything faster, but sometimes it definitely makes the code slower.