[fricas-devel] Bugs in TransSolvePackage

[Riccardo GUIDA]

This is an improved revival of a 2007 bug [1] which seems still in good health


(1) -> solve(tan(z)=0,z) -- OK

(1) [z = 0]
Type: List(Equation(Expression(Integer)))
(2) -> solve(sin(z)/cos(z)=0,z) -- OK

(2) [z = 0]
Type: List(Equation(Expression(Integer)))
(3) -> solve(cot(z)=0,z) -- OK

%pi
(3) [z = ───]
2
Type: List(Equation(Expression(Integer)))
(4) -> solve(cos(z)/sin(z)=0,z) -- ???

(4) []
Type: List(Equation(Expression(Integer)))
(5) -> solve(tanh(z)=0,z) -- ??? I'm wondering why fricas does not know that atanh(0) = 0

(5) [z = atanh(0)]
Type: List(Equation(Expression(Integer)))
(6) -> solve(sinh(z)/cosh(z)=0,z) -- ??? AWFUL KO

(6) [z = 0, z = log(- 1)]
Type: List(Equation(Expression(Integer)))
(7) -> solve(coth(z)=0,z) -- ? Error???
Invalid argument

(7) -> coth(z)=0

(7) coth(z) = 0
Type: Equation(Expression(Integer))
(8) ->







PS

The semantics of [] is not really crystal-clear...

(8) -> solve(cos(z)/sin(z)=cot(z),z)

(8) []
Type: List(Equation(Expression(Integer)))
(9) -> solve(sin(z)/cos(z)=tan(z),z)

(9) []
Type: List(Equation(Expression(Integer)))


It seems to me that the natural output for general solvers should be a set, see [2].


[1] http://axiom-wiki.newsynthesis.org/104BranchCutsDiffer?root=IssueTracker
[2] http://docs.sympy.org/latest/modules/solvers/solveset.html

[Riccardo GUIDA]

PS

Post by Riccardo GUIDA
(6) -> solve(sinh(z)/cosh(z)=0,z) -- ??? AWFUL KO
(6) [z = 0, z = log(- 1)]
Type: List(Equation(Expression(Integer)))

I mean, if log it is defined also on the axis |z| e^{i pi } than z=i pi should be ok, but what is bad is that one finds 0,1,2 solutions according to the representation of the question.