Add methods taking machine floats for Sinh, Cosh, Tanh, Asinh, Acosh, Atanh, CubeRoot, Erf, Gamma

[fingolfin]

• A bunch of operations for floating point numbers had no implementation for
  machine floats, even though C99 provides suitable functions.

• Most of the math functions we checked for in configure are actually
  guaranteed to be there by C99. The sole exception is exp10, which is a
  non-standard extension.

• Add a bunch of super-trivial tests to ensure those new functions are
  actually callable. These are not good tests, just a bare minimum.

Resolves #3983

[fingolfin]

Actually as far as I can tell, not even the float package implements this, for any float type -- correct, @laurentbartholdi ?

So perhaps we should just remove this, there seems little value in keeping an operation with no methods around?

[laurentbartholdi]

Well, I just followed the standard (I don't remember where I read it, though); so in my opinion it's best to leave it there, so as to reserve the word "Zeta".

[fingolfin]

@laurentbartholdi could you please clarify which "standard" you are referring to here?

[mohamed-barakat]

Thanks @fingolfin.

[coveralls]

Coverage increased (+0.005%) to 84.371% when pulling 853fa92 on fingolfin:mh/floats into 7941f3d on gap-system:master.

[laurentbartholdi]

No, sorry :) I just remember having looked up something and copied it into the file, but can't locate it anymore. A quick Google search brings up Library TR1 ISO/IEC TR 19768:2007, then published as an independent ISO standard, ISO/IEC 29124:2010 which seems never to have been properly implemented, except perhaps in C++17.

…

On Mon, Apr 27, 2020, 11:18 Max Horn ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In tst/testinstall/float.tst <#3986 (comment)>: > +0. +gap> Square(0.); +0. + +# +gap> FrExp(0.); +[ 0., 0 ] +gap> LdExp(0.,0); +0. +gap> Norm(0.); +0. +gap> SinCos(0.); +[ 0., 1. ] +gap> Erf(0.); +0. +gap> #Zeta(0.); # TODO: not implemented for machine floats @laurentbartholdi <https://github.com/laurentbartholdi> could you please clarify which "standard" you are referring to here? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#3986 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AARAQUDFSS3FBH22E75WKMTROVEXXANCNFSM4MQ7RZ3Q> .

[fingolfin]

Well, C99 implements most of the functions you added to GAP, except for zeta. And also gamma was renamed to tgamma and with a companion lgamma that apparently often is more useful. See https://en.cppreference.com/w/c/numeric/math

I looked a bit around and found no standard the includes zeta, other than indeed C++17 which seems to include everything and the kitchen sink, but that's not something we can reasonably aim for :-)