Open LebedevRI opened 1 year ago
Thanks for pointing this out. I have to think how best to address it. The issue that errors is the expression when lambdified includes an x^(0.6)
which for a negative, real x
will error. You can avoid that with x=-1.0 + 0im
.
But that is unsatisfying. It seems SymPy shouldn't simplify this the way it does.
julia> @syms x::real;
julia> simplify(abs(x)^(0.6))
0.6
│x│
julia> simplify(1/abs(x)^(0.6))
│ -0.6│
│x │
simplify()
moves exponentiation into the modulo operation, but that apparently changes semantics as per externally-observable side-effects:This is Julia Version 1.9.1, SymPy v1.1.9, on debian amd64 sid.