CODEWITHSUNDEEP
pythonwrapperpython-decorators
Federico Vega
Federico Vega

Reputation: 355

Wrapper for an encapsulated function

Lets say self.sigma is a float number and I define a function that returns a function a.

def f(self):
    a = lambda i: self.sigma*i**2
return a

Now I want to extend the functionality of a, whenever self.sigma is a matrix:

self.sigma = sigma[i,j]

where sigma is a 2-dimensional numpy array.

Is possible to define a wrapper to do it? I would like something like:

def f(self):
    a = lambda i,j: self.sigma[i,j]*i**2
return a

without re-wrtting it explicitly. Is it the best way to do it?

Additional information: In my programme f returns five functions not one. In the context of the program makes sense to encapsulate all this functions in a single function f.

Upvotes: 0

Views: 92

Answers (2)

0xDECAFC0FFEE
0xDECAFC0FFEE

Reputation: 380

Maybe np.vectorize does something similar?

vectorizing f will return a matrix of lambdas. Each takes i as an input.

import numpy as np

def f(sigma):
    a = lambda i: sigma*i**2
    return a

f_vectorized = np.vectorize(f)
sigma = [[0, 1], [2, 3], [4, 5]]

for i in [0, 1, 2]:
    for j in [0, 1]:
        print(f_vectorized(sigma)[i, j](i))

# prints 0, 0, 2, 3, 16, 20

Upvotes: 2

Roman Zh.
Roman Zh.

Reputation: 1049

If your sigma is a numpy array your f works as expected:

import numpy as np

def f(sigma):
    a = lambda i: sigma*i**2
    return a

sigma = np.random.rand(2, 4)

print(sigma)
print(f(sigma)(2))

# [[0.40567906 0.86706289 0.37468664 0.45059327]
#  [0.57024244 0.17452584 0.05394594 0.03218902]]

# [[1.62271624 3.46825154 1.49874657 1.80237307]
#  [2.28096976 0.69810334 0.21578375 0.12875607]]

If your sigma is a list of lists, then you can use vectorize from numpy and get the array of functions:

sigma = [[1, 2], [3, 4]]
vfunc = np.vectorize(f)

print(vfunc(sigma))
#[[<function f.<locals>.<lambda> at 0x000002DC116DE7B8>
#  <function f.<locals>.<lambda> at 0x000002DC116DEC80>]
#
# [<function f.<locals>.<lambda> at 0x000002DC116DEBF8>
# <function f.<locals>.<lambda> at 0x000002DC116DE730>]]

Upvotes: 1

Related Questions