[Gate-users] Different seeds for GATE in a cluster

[Bryan McIntosh]

Naser,

One of the summer students in our lab just dealt with this very
problem.  Under GATE 5.0.0 (and 6.0.0 as well, presumably) not only do
you need to specify which random number generator to use, you also
need to specify its starting seed.  For example:

/gate/random/setEngineName MersenneTwister
/gate/random/setEngineSeed auto

The second command's possible values are:

default - Uses the default random number seed from CLHEP.  This is
probably why all your simulations give the same results

auto - Sets a random number seed based on the system clock time and
the process ID of the GATE instance that's being run.  Usually gives
staistically unique results in my experience.

1754756 (or any other number) - Sets a random number seed manually by
entering a number.  Not a bad way of doing it, so long as you ensure
that you don't use the same seed twice for simulations that you want
to get statistically unique results from.

This should solve your problem; we've run the simulations with ASCII
and ROOT output disabled and only the sinogram output enabled, and
everything executes quite well.  Our biggest problem has been the fact
that our 80 crystal ring scanner that we're simulating gives output
files that are 250 megabytes each before we use SSRB to trim them down
with an external program!

Cheers,

-Bryan