[Tobyfit-tickets] [TOBYFIT] #97: Investigate Monte Carlo options in fit and simulate

[Tobyfit]

#97: Investigate Monte Carlo options in fit and simulate
--------------------------+-------------------------------------------------
Reporter:  Ronald Fowler  |      Owner:  Ronald Fowler
    Type:  task           |     Status:  new          
Priority:  major          |   Keywords:               
--------------------------+-------------------------------------------------

Comment(by Toby Perring):

 The mc_type option=1 should be made to resemble more closely mc_type=0 in
 its operation: the Sobol sequence should NOT be reset for every pixel, but
 only for each iteration, just as in mc_type=0. This will ensure that
 succesive pixels are evaluated at different points, but that chi-squared
 is alway evaluated at the same points. This will give the same stability
 that mc_type=0 has during fitting.

 The current mc_type=1 can be made mc_type=-1. For the sake of
 completeness, one could add an mc_type=-2, which resets the seed for every
 pixel with the true random number generator.

 At the end of this, the Monte Carlo options will be complete!

-- 
Ticket URL: <http://trac.isis.rl.ac.uk/Tobyfit/ticket/97#comment:1>
Tobyfit <http://tobyfit.isis.rl.ac.uk/>
Tobyfit