Talk About Network

Google





Science > Crypt Random-numbers > Big linear comp...
Latest [ Topics | Posts ] Archive Post A New Topic Post a Reply
<< Topic < Post Post 1 of 1 Topic 380 of 428
Post > Topic >>

Big linear complexity for SFMT PRNG

by "Cristiano" <cristiano.pi@[EMAIL PROTECTED] > Aug 10, 2007 at 07:24 PM

I calculated the linear complexity of some Mersenne Twister PRNG using two
implementations of the Berlekamp-Massey algorithm.
The sequence to test is generated by taking only the bits in one position
(not the whole 32-bit number). In C, for example, we can do:
    bit= (PRNG() >> 16) & 1
to take the bit 16.

I obtained a strange linear complexity for SFMT19937:
http://www.math.sci.hiro****ma-u.ac.jp/~m-mat/MT/SFMT/
(I checked the correctness of the implementation).
While for MT19937 and WELL19937c I get L=19937 (as expected), for
SFMT19937
I get values from 79860 and 79872 (which is good, but strange).
In the paper
http://www.math.sci.hiro****ma-u.ac.jp/~m-mat/MT/SFMT/M062821.pdf
I saw nothing about that.

Please, could someone explain?

Thanks
Cristiano
 




 1 Posts in Topic:
Big linear complexity for SFMT PRNG
"Cristiano" <  2007-08-10 19:24:34 

Post A Reply:
  Go here to Signup

AddThis Feed Button


About - Advertising - Contact - Frequently Asked Questions - Privacy Policy - Terms of Use - Signup

Contact
localhost-V2008-12-19 Thu Jan 8 0:22:11 PST 2009.