Andrew Haley wrote:
BTW, if your Monte Carlo simulation is affected by such tiny
differences in transcendentals, then your simulation really is broken!
Andrew.
I meant to make that point earlier, but overlooked to do so.
Every time you run a Monte Carlo simulation you can expect slightly different
results - how slight depends on your budget for CPU time. I once had 24
computers working in parallel on Monte Carlo simulations.
http://www.iop.org/EJ/abstract/0031-9155/42/6/016
Since the original poster is concerned about portability, send me a bit of code
and I'll tell you the result on a SPARC processor or two. I would not be
surprised at all if it gave you yet another result.
Dave
