On Monday, 24 May 2021 at 14:44:09 UTC-7 ... wrote: > On Sunday, 9 May 2021 at 02:50:06 UTC-6 vdelecroix wrote: > >> What is annoying in the above scenario is that it breaks >> badly the rule "same input parent" -> "same output parent". >> Here are three distinct output parents for the same substitution >> command >> > > I completely agree with Vincent on this point. If you are going to do > something further with the output in your code, having potentially > different objects is very inconvenient. Since partial substitution (which > is supported by "subs" method) almost certainly has to live in the same > polynomial ring as the starting polynomial, I strongly favour the result of > subs to be in this ring whenever possible. Perhaps if there is a full > substitution and another natural ring to live in, as is the case with ring > homomorphisms, it can be there, but do not drop to the ring of coefficients > just because it is possible for a particular case. >
+1 on letting the codomain only depend on the parents of the argument; not on the values themselves. Note, though, that currently sage: R.<x,y>=QQ[] sage: f=x+y sage: f.subs(x=1.0) y + 1.00000000000000 sage: f.subs(x=1.0).parent() Multivariate Polynomial Ring in x, y over Real Field with 53 bits of precision I don't think you'd want to break that. The parent here gets discovered because the coercion system can figure out that adding/multiplying elements from RR and from QQ['x','y'] can be performed using a base change and produce elements of RR['x','y'] > On the other hand, call method does not support partial substitution and > acts as an evaluation where it makes a lot of sense to use the coefficient > ring by default, but again one can consider natural subcases. > It does: sage: f(x=1.0) y + 1.00000000000000 I agree it's a bit redundant to have __call__ as well as subs, but that's a historic artefact by now. I'd say making them completely synonymous is the easiest way of dealing with this. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/1691ffe1-e631-4bb8-b8c0-2680e2e61303n%40googlegroups.com.
