python itertools组合，包括自动相关

Question

Is there an elegant/pythonic way to retrieve both the cross- and auto-correlations of elements of a list?

I could do this manually, though I'd like to try and use itertools.combinations , which by default doesn't seem to produce auto-correlations [(1,1), (2,2), etc.].

This question actually deals with the cross -correlations only:

Calculating correlations between every item in a list

Try this for a MWE:

import itertools
data = range(4); crosscorrs=[]
# Get the cross-correlations:
for (i,j) in itertools.combinations(data, 2):
    crosscorrs.append((i,j))
# How to get the auto-correlations?
# What about the (j,i) correlations even?!
# Result
print crosscorrs

What about the (j,i) correlations even!?

Thanks for all help

Answer 1

There is a number of things wrong with the question, including a piece of example code which will raise an Exception , but I will attempt to address what you seem to be asking.

As you name the collection "crosscorrs" I take it you want separate lists for "crosscorrelation" indices and "autocorrelation" indices. To get the (j,i) indices, you can just refer to them as such. The same-index-stuff can be added to a set, to avoid having duplicates.

from itertools import combinations
data = range(4)
crosscorrs=[]
autocorrs = set()

for (i,j) in combinations(data,2):
    crosscorrs.extend([(i,j),(j,i)])
    autocorrs.add((i,i))
    autocorrs.add((j,j))

print(crosscorrs, '\n')
print(list(autocorrs))

If you would want all of this to go into 1 collection, take a look at itertools.combinations_with_replacement and the option to reverse data by data[::-1] . Quite possibly, a set can come in handy, as it supports union operations.