Backgound

I want to generate random numbers with distributions(np.random.normal, np.random.poisson, etc.), passing seveal keyword parameters(loc, scale, size, etc. each one is a list) into it.

loc = [1,2,3,6,10]
scale = [4,6,7,8,5]
size = [10,9,7,8,5]

# When I know which kwargs is in use, this lambda function works
list(map(lambda x,y: np.random.normal(loc = x, size = y), loc,size))

# however, the number of kwargs may change and the kwargs themselves may change. It won't work with codes below. How to generlize the function above?
list(map(lambda **params: np.random.normal(**params),**{'loc':loc, 'scale': scale, 'size':size}))
list(map(lambda x,y: np.random.poisson(loc = x, size = y), loc,size))

Outputs:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-47-dfea011463fd> in <module>
----> 1 list(map(lambda **params: np.random.normal(**params),**{'loc':loc, 'size':size}))

TypeError: map() does not take keyword arguments



---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-48-1f3449886ea1> in <module>
----> 1 list(map(lambda x,y: np.random.poisson(loc = x, size = y), loc,size))

<ipython-input-48-1f3449886ea1> in <lambda>(x, y)
----> 1 list(map(lambda x,y: np.random.poisson(loc = x, size = y), loc,size))

mtrand.pyx in numpy.random.mtrand.RandomState.poisson()

TypeError: poisson() got an unexpected keyword argument 'loc'

Question

Is there a way to use built-in/numpy to iterate over kwargs's elements?

CodePudding user response：

You can try this:

import numpy as np

params = dict(loc=[1,2,3,6,10],
              scale=[4,6,7,8,5],
              size=[10,9,7,8,5])

ds = (dict(zip(params.keys(), vals)) for vals in zip(*params.values()))
list(np.random.normal(**d) for d in ds)

CodePudding user response：

If you want to use map, you could do:

params = (loc, scale, size)
names = ('loc', 'scale', 'size')

list(map(lambda p: np.random.normal(**dict(zip(names, p))),
         zip(*params)))