Colin B. Macdonald

Would be a nice feature. 1. Currently, the pretty printing is done at the time when the sym is created and cached in `x.unicode`. Instead, would need to generate it...

should we allow symfun as indept var in another symfun?

2

Some number of @latot's PRs/issues depend on this working: ``` syms g(x) f(g) = 2*x ``` and/or: ``` syms g(x) f(g) = 2*g ``` That is, having a symfun as...

limit(1/x,x,0) without direction should be NaN

4

or an error or something. SymPy's default to right-hand limit is not very intuitive. I supposed I should just call SymPy with both `-` and `+` when user doesn't specify......

improve new double functions nan behaviour

3

``` >> zeta(nan) ans = NaN - NaNi ``` (probably involves dealing with real and complex cases separately (which might be a good idea for other reasons).

diff(symfun) should respect `@symfun/symvar` ordering

3

`@sym/symvar` and `@symfun/symvar` are subtly different (as documented). `diff(f)` where `f` is a symfun should use `@symfun/symvar`, but it looks like it does not: ``` syms f(g, h) diff(f) ans(g,...

In PyTave and Java `class(x)` gives the Python or Java class. That might be nice for us too. E.g., ``` a = interval(1, 5); class(a) ans = sympy.Interval ``` For...

Make the sym class a new-style classdef class

3

Possibly this allows fixing #13. Good chance to learn more about Octave OO programming. Maybe @latot is interested?

eye(3,'sym'), zeros, ones, etc are missing

1

One way is probably with static methods, which i think needs classdef. So wait for octave to get classdef?

isinf is tristate in sympy, SMT not so

4

which should we follow? isinf(x + oo). SMT 2014a says true. Sympy says none. This is pretty corner-case stuff. I'm inclined to follow upstream behaviour (i.e., sympy) rather than trying...

pytave dict hacking

2

Current implementation may not work correctly if only some of the values are SymPy objects - [x] confirmed - [ ] make test for that Also, dicts with SymPy object...