best practices request for np.where() that obviates the double dereferencing of the np.where object-CodePudding

I would like to evaluate

x=np.array([1,2,3])
imax=np.where(x==max(x))

Clearly imax=2 but that is not what np.where returns. To obtain imax=2, I must use a two-step process that involves doubly dereference the np.where object:

np_where_object=np.where(x==max(x))
imax = np_where_object[0][0]

But this is inelegant. Is there a best practice that avoids the double dereferencing and obtains imax=2 in one pass? And I don't mean np.where(x==max(x))[0][0], which is still inelegant.

CodePudding user response：

Yes, there is. It's argmax:

>>> x = np.array([1,2,3])
>>> imax = x.argmax()
>>> imax
2