If I understand the problem correctly, a similar algorithm is needed to do feature matching in 2 photos to recover 3D information. A method I particularly like is described here:

HP Labs: Stereo via SVD

Basically, features are found (you can use corners or edges in an image...maybe spikes in your 1D data), and the correlation is performed on the features of each data set using Singular Value Decomposition. So feature detection is used to reduce the data set, and then SVD is used to robustly correlate the data sets.

Quote: Original post by Kylotan
I'm not comparing strings. :) The items are largely judged on two floating point values, one of which dictates the ordering, the other of which just has to 'match'.

Okay, given the clarification in your last two posts, then you are looking at a cross correlation problem. Since you have 2 floats, then you can map your list members into a 2D continuous space and compute the nearest neighbours search I mentioned. The distributions of the distance to the nearest neighbours (intra and inter list) are needed for the cross correlation computation. If it can wait until Saturday, I'll dig out the code I have, remove the relevant sections and make sure it has appropriate documentation... and email it to you.

Cheers,

Timkin