Can you simplify the input file to about 10 trials or so (but where it still produces the same result) and then post it as an attachment? I'd be happy to take a look at it.