On Wed, 11 Feb 2015 15:24:04 +0100, Joerg Sonnenberger
<jo...@britannica.bec.de> wrote:
On Wed, Feb 11, 2015 at 02:13:10PM +0100, j. van den hoff wrote:
for the netbsd repo (presuming there are half a million checkins in it)
Hash conflicts likely count all artifacts, so it would be more like 2.1
do they?
million :)
OK, if yes -- and with your figure of ≈2e6 artifacts --, we cannot use the
first order taylor approximation p(n)=n^2/(2m) but
are better off with p(n) = 1 - exp(-n^2/(2m)) (still relying on n << m):
p(2e6) ≈ 1 - exp(-4e12/2/1.1e12) ≈ 1 - exp(-2) ≈ 0.86
so that indeed would lead to something like 0.86^4 ≈ 0.56 for prob. of 4
collisions (and the fifth one to be expected "soon").
this indeed makes the empirical finding consistent with expectations.
thank you!
joerg (yes, me too ;-))
Joerg
_______________________________________________
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
--
Using Opera's revolutionary email client: http://www.opera.com/mail/
_______________________________________________
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
