CODEWITHSUNDEEP
pythonnumpysympy
not-a-user
not-a-user

Reputation: 4327

In sympy, how do I get the coefficients of a rational expression?

I have a rational (here: bilinear) expression and want I sympy to collect the coefficients. But how?

from sympy import symbols, Wild, pretty_print

a, b, c, d, x, s = symbols("a b c d x s")

def coeffs(expr):
    n0 = Wild("n0", exclude=[x])
    n1 = Wild("n1", exclude=[x])
    d0 = Wild("d0", exclude=[x])
    d1 = Wild("d1", exclude=[x])
    match = expr.match((n0 + n1*x) / (d0 + d1*x))
    n0 = n0.xreplace(match)
    n1 = n1.xreplace(match)
    d0 = d0.xreplace(match)
    d1 = d1.xreplace(match)
    return [n0, n1, d0, d1]

if __name__ == '__main__':
    pretty_print(coeffs((a + b*x) / (c + d*x)))
    pretty_print(coeffs(2 * (a + b*x) / (c + d*x)))
    pretty_print(coeffs(s * (a + b*x) / (c + d*x)))

I tried this using match, but it fails almost always, for example in the last line (the one with a symbolic prefactor of "s") I get

Traceback (most recent call last):
  File "...", line 20, in <module>
    pretty_print(coeffs(s * (a + b*x) / (c + d*x)))
  File "...", line 11, in coeffs
    n0 = n0.xreplace(match)
  File "/usr/lib/python3/dist-packages/sympy/core/basic.py", line 1626, in xreplace
    return rule.get(self, self)
AttributeError: 'NoneType' object has no attribute 'get'

so the match did not work.

Upvotes: 3

Views: 1773

Answers (1)

Stelios
Stelios

Reputation: 5521

If you know that the expressions you will consider are rational, you could extract their numerator and denominator and find their coefficients independently.

Here's one approach to do so:

import sympy as sp

def get_rational_coeffs(expr):
    num, denom = expr.as_numer_denom()

    return [sp.Poly(num, x).all_coeffs(), sp.Poly(denom, x).all_coeffs()]

a, b, c, d, x, s = sp.symbols("a b c d x s") 
expr = (a + b*x) / (c + d*x)

# note the order of returned coefficients
((n1, n0), (d1, d0)) = get_rational_coeffs(s*expr)
print(((n0, n1), (d0, d1)))

((a*s, b*s), (c, d))

The above approach is also faster than coeffs. I get the following timings (based on Jupyter's %timeit magic) for the case where the coefficients of expr are required (case where coeffs succeeds):

%timeit get_rational_coeffs(expr)
%timeit coeffs(expr)

1000 loops, best of 3: 1.33 ms per loop

1000 loops, best of 3: 1.99 ms per loop

Upvotes: 4

Related Questions