http://code.activestate.com/recipes/tags/argsort/new/2012-03-14T15:14:11-07:00ActiveState Code Recipes2012-03-14T15:14:11-07:00Garretthttp://code.activestate.com/recipes/users/4181290/http://code.activestate.com/recipes/578074-fast-sorting-of-n-dimensional-array-by-first-dimen/

<p style="color: grey"> Python recipe 578074 by <a href="/recipes/users/4181290/">Garrett</a> (<a href="/recipes/tags/argsort/">argsort</a>, <a href="/recipes/tags/lists/">lists</a>, <a href="/recipes/tags/numpy/">numpy</a>, <a href="/recipes/tags/python/">python</a>, <a href="/recipes/tags/sorting/">sorting</a>). </p> <p>I have looked far and wide for code for fast sorting of n dimensional arrays by the first element, for example if I had the array: <br /> ray = [[1,2,3,7,5][10,11,12,13,14]] </p> <p>I would want it to come out as ray = [[1,2,3,5,7][10,11,12,14,13]]</p> <p>There are several ways to do this. One is zipped = zip(<em>ray) zipped.sort() ray = zip(</em>zipped)</p> <p>but this is extremely slow. Numpy has a much faster way to do it, but it wasn't immediately apparent.</p> <p>if the above were a numpy array you could simply do the following: indexes = numpy.argsort(ray[0]) for n in xrange(len(ray)) ray[n] = ray[n][indexes]</p> <p>I did a time test of the two methods below.</p>