On Sat, Mar 2, 2013 at 2:29 AM, G B <g.c.b.at.w...@gmail.com> wrote: > Ah thanks-- always comforting to know I'm not the first person to > complain... =) And the broadcasting argument makes the current design less > mystifying. > > In my case, I have a sympy equation that I lambdify to a function that I > then want to plot. I wind up with something like this: > xx=linspace(...) > f=lambdify(x,exp,"numpy") > y=f(xx) > plot(xx,y) > > That snippet of code worked fine until it ran on an expression that > simplified to a constant. I could solve the problem by placing special > handling around y before plot, but that feels inelegant. > > I understand the broadcasting argument from a numpy centric view, but is > that the model that lambdify should follow? I feel as though in numpyland > you're more likely to know the shape of your function, and thus would know > from the outset whether y needed to be reshaped. > > In sympyland, the function itself is a variable, so special handling of y > becomes necessary boilerplate. > > Of course, I've explicitly told lambdify to create a "numpy" function, so > it's not unreasonable that it behave in the numpy way. Maybe this is a > place for a keyword argument? > > Maybe I'm giving too much weight to my own particular use case, but is this > a discussion worth reopening?
Yes, if you feel that your use case gives a good argument to change it, then definitely. I personally don't use numpy, so I can't really comment on it much myself. >From what I remember, the reason that it does this is not because of a design choice, but because of the way that zero-argument lambdas work compared to multi-argument ones (basically, multi_argument(array) gives an array, but zero_argument() gives a constant, because it is never broadcast). So without any additional modifications, this is the behavior, but that doesn't mean that it is the right one. Aaron Meurer > > > > On Saturday, March 2, 2013 12:21:20 AM UTC-8, Aaron Meurer wrote: >> >> See https://code.google.com/p/sympy/issues/detail?id=2543 for some >> discussion on this. >> >> Aaron Meurer >> >> On Fri, Mar 1, 2013 at 8:53 PM, G B <g.c.b....@gmail.com> wrote: >> > This is kind of a degenerate case, but personally I think it should >> > behave >> > differently than it does currently. >> > >> > f1=lambdify(x,2*x,"numpy") >> > f2=lambdify(x,S(1),"numpy") >> > >> > f1(array([1,2,3]) --> [2,4,6] >> > f2(array([1,2,3]) --> [1] >> > >> > I feel that f2 should return [1,1,1] in that situation. Am I >> > misinterpreting that? >> > >> > Cheers-- >> > Greg >> > >> > >> > >> > >> > -- >> > You received this message because you are subscribed to the Google >> > Groups >> > "sympy" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> > an >> > email to sympy+un...@googlegroups.com. >> > To post to this group, send email to sy...@googlegroups.com. >> > Visit this group at http://groups.google.com/group/sympy?hl=en. >> > For more options, visit https://groups.google.com/groups/opt_out. >> > >> > > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sympy+unsubscr...@googlegroups.com. > To post to this group, send email to sympy@googlegroups.com. > Visit this group at http://groups.google.com/group/sympy?hl=en. > For more options, visit https://groups.google.com/groups/opt_out. > > -- You received this message because you are subscribed to the Google Groups "sympy" group. To unsubscribe from this group and stop receiving emails from it, send an email to sympy+unsubscr...@googlegroups.com. To post to this group, send email to sympy@googlegroups.com. Visit this group at http://groups.google.com/group/sympy?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
