[Question] Combining priors

[alvations]

Thank you for the useful and extremely user-friendly tool!

If we have multiple priors saved from multiple subsets of an large dataset, is there some way to combine the priors?

E.g. we have a corpus of 1 billion, we split it up into 1000 parts and ran eflomal-align and then eflomal-makepriors for each of them, any way to combine the 1000 different prior files into 1?

[robinp]

For me it seems the priors contain keyed counts (not probabilities), so they can be trivially merged by adding up the counts of the same keys. Key is effectively all columns but the last, which is the count.