CODEWITHSUNDEEP
pythonnumpyoptimizationscipysolver
abcd
abcd

Reputation: 10751

Solve for the argument to a function that will produce a particular return value

Suppose I have a Python function y of x, where x must be a number, and the return value is always a number. y is a mathematical function of one variable.

Is there a function in Python (i.e., in numpy, scipy, etc.) that would be able to solve for the particular value(s) for x that produces a desired return value -- using, e.g., gradient descent?

I'm looking for a magic_function that would do something like this:

>>> from my_module import y
>>> from magic_package import magic_function
>>> desired = 10
>>> x = magic_function(y, desired)
>>> y(x)
10

Upvotes: 0

Views: 336

Answers (1)

PM 2Ring
PM 2Ring

Reputation: 55469

You can wrap your function y(x) so that it's offset by the desired value. Here's a simple demo of the long way to do that:

def y(x):
    return x*x

def offset_function(f, desired=0):
    def newf(x):
        return f(x) - desired
    return newf

y9 = offset_function(y, 9)

for x in range(5):
    print x, y(x), y9(x)

output

0 0 -9
1 1 -8
2 4 -5
3 9 0
4 16 7

However, it's simpler just to use a lambda:

y9 = lambda x: y(x) - 9

You can pass that to your root-finder like this:

scipy.optimize.root(lambda x: y(x) - 9, xguess)

Upvotes: 2

Related Questions