一列中连续浮点数的差

Question

I have a list of floating point numbers in a file in column like this:

123.456

234.567

345.678

How can i generate an output file which is generated by subtracting the value in a line with the value just above it. For the input file above,the output generated should be:

123.456-123.456

234.567-123.456

345.678-234.567

The first value should return zero, but the other values should get subtracted with the value just above it. This is not an homework question. This is a small requirement of my bigger problem and i am stuck at this point. Help much appreciated. Thanks !!

Answer 1

This will work:

diffs = [0] + [j - data[i] for i,j in enumerate(data[1:])]

So, assuming data.txt contains:

123.456
234.567
345.678

then

with open('data.txt') as f:
    data = f.readlines()
    diffs = [0] + [float(j) - float(data[i]) for i,j in enumerate(data[1:])]

print diffs

will yield

[0, 111.111, 111.11099999999999]

This answer assumes you want to keep the computed values for further processing.

If at some point you want to write these out to a file, line by line:

with open('result.txt', 'w') as outf:
    for i in diffs:
        outf.write('{0:12.5f}\n'.format(i))

and adjust the field widths to suit your needs (right now 12 spaces reserved, 5 after the decimal point), written out to file result.txt .

UPDATE : Given (from the comments below) that there is possibly too much data to hold in memory, this solution should work. Python 2.6 doesn't allow opening both files in the same with , hence the separate statements.

with open('result2.txt', 'w') as outf:
    outf.write('{0:12.5f}\n'.format(0.0))
    prev_item = 0;
    with open('data.txt') as inf:
        for i, item in enumerate(inf):
            item = float(item.strip())
            val = item - prev_item
            if i > 0:
                outf.write('{0:12.5f}\n'.format(val))
            prev_item = item

Has a bit of a feel of a hack. Doesn't create a huge list in memory though.

Answer 2

给定值列表：

[values[i] - values[i-1] if i > 0 else 0.0 for i in range(len(values))]

Answer 3

Instead of list comprehensions or generator expressions, why not write your own generator that can have arbitrarily complex logic and easily operate on enormous data sets?

from itertools import imap

def differences(values):
    yield 0  # The initial 0 you wanted
    iterator = imap(float, values)
    last = iterator.next()
    for value in iterator:
        yield value - last
        last = value

with open('data.txt') as f:
    data = f.readlines()

with open('outfile.txt', 'w') as f:
    for value in differences(data):
        f.write('%s\n' % value)

If data holds just a few values, the benefit wouldn't necessarily be so clear (although the explicitness of the code itself might be nice next year when you have to come back and maintain it). But suppose data was a stream of values from a huge (or infinite!) source and you wanted to process the first thousand values from it:

diffs = differences(enormousdataset)
for count in xrange(1000):
    print diffs.next()

Finally, this plays well with data sources that aren't indexable. Solutions that track index numbers to look up values don't play well with the output of generators.