[ I have posted this to both "sci.crypt" and "sci.crypt.random-numbers". I have posted separately to each group instead of cross-posting ]
Today I forked the "Dieharder" repository on Github, and I've added a new test.
My new test is Test No. 18, and it is for predicting the output of the 32-Bit Mersenne Twister.
You can see my code changes here:
https://github.com/eddelbuettel/dieharder/compare/master...healytpk:dieharder_predict_mersenne:master
If you compile it and run it as follows:
./dieharder/dieharder -d 18 -g 401
./dieharder/dieharder -d 18 -g 402
./dieharder/dieharder -d 18 -g 403
./dieharder/dieharder -d 18 -g 404
./dieharder/dieharder -d 18 -g 405
You'll see that it comes back with "PASS" except for generator No. 403 (i.e. the 32-Bit Mersenne Twister).
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)