用python对字典进行排序和汇总

Question

I want to normalize a directed graph. Currently I am doing this:

def normalize_distribution(in_degree_dist):
    '''
    normalize the distribution so all values sum up to 1
    '''
    # calcuate normalization factor
    factor=1.0/sum(in_degree_dist.itervalues())

    # sort the dictionary
    sorted_in_degree_dist = OrderedDict(sorted(in_degree_dist.iteritems(), key= lambda srt: srt[0]))

    # apply the factor to every value
    for key in sorted_in_degree_dist:
        sorted_in_degree_dist[key] *= factor
    return sorted_in_degree_dist

I found out that I am doing one iteration too much. While iterating over the dictionary keys I could sum up the values. So I would only do one iteration instead of two a huge saver if the graph gets big.

So I replaced the lambda with my own function. But the sorting is kind of smart so the key function isn't called on every element which is kind of necessary for sorting.

def sort_sum(*args, **kwargs):
    '''
    Sorting and summing up
    '''
    print args
    return args[0][0]

For a graph with n = 20 nodes and p = 0.5 the output is:

Random graph with 20 nodes and probability 0.5 created.
Degree distribution calculated.
((4, 1),)
((8, 4),)
((9, 1),)
((10, 5),)
((11, 4),)
((12, 3),)
((13, 1),)
((14, 1),)
Degree distribution normalized.

Only eight calls for 20 elements, which is good for sorting but bad for summing.

I thought of list comprehension to do this

[key, val for key, val in in_degree_distribuion.iteritems()]

but I can't figure out how to sum up.

Do I have to write my own sorting and summing algorithm to do this in one step?

Answer 1

You could always apply the multiplication factor whilst sorting (just after really), instead of doing the sum.

eg.

# calcuate normalization factor
factor = 1.0 / sum(in_degree_dist.itervalues())

# sort the dictionary
sorted_in_degree_dist = OrderedDict((key, in_degree_dist[key] * factor)
    for key in sorted(in_degree_dist))

# or 
sorted_in_degree_dist = OrderedDict((key, value * factor)
    for key, value in sorted(in_degree_dist.iteritems()))