On Friday, March 30, 2012 10:54:56 PM UTC+5:30, Stefan Krastanov wrote: > > > My GSoC Application can be found at > > > https://github.com/sympy/sympy/wiki/GSoC-2012-Application--Bharath-M-R-:-Plotting-Module > . > > Can you please review the application and suggest any changes? > > Before giving my feedback I should repeat yet one more time that I am > not among those that decide which applications get accepted. > > # In synopsis: > I think that you mean a backend and not a whole new module when you > speak about svgfig. > > # In week 1: > What do you mean by parse implicit functions? Anyway, you should not > overcharge lambdify with stuff it was not meant to be. This is the > reason that the old lambdify is such a mess. If you need some > functionality from lambdify just use it in whatever abstraction you > need to build. > There are a few problems with the syntax that you are proposing. You > are using strings instead of expressions, you are using lists instead > of tuples and the order of the arguments is not the one used in other > parts of sympy. > > # In week 3: > The mpmath stuff should be discussed with them *much* in advance. > > # In week 4: > I do not get the example with min and max (and they are not > necessarily binary in python). Moreover, could you elaborate on what > subpixel computation is and why do we need it. Finally, what is the > use for `IntervalSet`. > > # Week 6 > Branch cut plotting? > > # Week 7 > Why is this not possible already by week 1. What is so special about > x**(1/3) > > # Week 8 > About "ploting every 2D function". Be aware that there are some > nontrivial expression (the old lambdify does not work on them, the new > one usually works). Will you be able to plot for example > `Integral(awful_expression_of_x, (x,y,z)) < > real(complex_valued_expression_of_y_and_z`)? > > # Week 9 > Could you elaborate? I do not get a clear idea what you mean here. > > # Week 12 > Elaborate on what do you need this adaptive method for? Is it part of > all that is implemented already? Is it well abstracted, so there is no > code duplication? > > > Thanks for reviewing. I think I was a bit sloppy and also didn't convey my thoughts clearly. I have made the changes. Can you look at it once again and give your views ? :)
> One last important thing. I was left with the impression that you will > rely on experimental_lambdify to return mpmath functions. This is not > what experimental_lambdify does. The only cases when mpmath is used in > this function is when there is no numpy function to translate to > (which happens often), however even then mpmath is hidden in the > evalf() methods of the expressions. > This will only be during the initial implementations. I plan to have a interval arithmetic library for further implementations. > The old lambdify can return mpmath functions but it works on a *very* > limited subset of sympy expressions. > Yeah. I would like to extend mpmath so that more functions can be implemented. > In one sentence: if you rely only on mpmath or numpy for evaluation > you will be able to plot only a very limited subset of expressions. > Maybe a useful idea will be to implement your interval arithmetics > directly in sympy independently of the library that actually does the > computations. > Once I have the extended interval arithmetic library ready, I can extend it to evaluate more functions. The problem with interval arithmetic is that you need to know the characterstics of the function to evaluate it over an interval ie only evaluating at the end points won't suffice. Hence I will not be able to support Integral(awful_expression_of_x, (x,y,z)) < real(complex_valued_expression_of_y_and_z`). If it is possible to simplify the expression to a set of implemented functions, then the expression can be evaluated. -- You received this message because you are subscribed to the Google Groups "sympy" group. To view this discussion on the web visit https://groups.google.com/d/msg/sympy/-/oTU_saPHlq8J. To post to this group, send email to sympy@googlegroups.com. To unsubscribe from this group, send email to sympy+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/sympy?hl=en.
