Implement expr.diff(x, n) where n is an integer

[BlackTeaAndCoffee]

Hi i wanted to calculate the derivative of some big functions, and i thought
that symengine would also use the diff method as in sympy and symengine has a diff method.
But doing so, i get a wrong result. Even for smaller functions i get
wrong results. See following code:

import symengine as si
x = si.symbols('x')
print(si.diff(x**2, 1))

This gives zero. I've got symengine version 0.3.0.

Best Wishes
BlackTea

[bjodah]

You need to specify the symbol which you want to take the derivative with respect to (x in this case) instead of 1.

[BlackTeaAndCoffee]

No i mistyped. I meant

import symengine as si
x = si.symbols('x')
print(si.diff(x**2, x, 1))

1 is for the first derivative. I need higher derivatives. I know that for the n'th derivativ i could just do
the derivation n times, but isn't here a faster way like in sympy? I don't want to sound crude or something, it's just a question.

Best wishes
Black Tea

[bjodah]

I see, yes SymPy's syntax supports multiple ways of doing this. Arguably symengine.py should reproduce that behaviour. In the mean time you can do this:

>>> si.diff(x**2, x, x)
2

or more generally:

>>> si.diff(x**47, *[x]*13)
875997661345462656000*x**34

This would limit the highest derivative to 254th derivative on python <=3.6 (and some much higher number on Python 3.7).

[BlackTeaAndCoffee]

Alright! Thank you very much!

[isuruf]

Arguably symengine.py should reproduce that behaviour

Yes, I agree.

This would limit the highest derivative to 254th derivative on python <=3.6

Not if you put the arguments in a tuple. For eg: si.diff(x**400, *[x]*300) works.

[bjodah]

Interesting that it works, I guess it's only the number of named arguments which are limited.