params_example.sav

This is the example parameter file.
The file should be formatted as
#keyword value

GPZ knows it is reading a keyword because the keyword is preceded by a #
Placing any character before the # will comment out that keyword

I have commented the keywords below, and also included some unused keywords,
in case you want to run with different options

BE VERY CAREFUL not to include #(valid keyword) in your comments, or else GPZ
might accidentally think you are trying to give it a command

#dim 6 -- the number of photometric filters

#prior -- including this keyword will impose the \delta z prior on the bi-modal
          model (see equation 19 in our paper)

#dof -- including this keyword will divide chi^2_model by the number of neighbor
        galaxies when choosing between models (see equation 22 of our paper)

#normbyvar -- including this keyword will tell GPZ to renormalize your input
              data so that the data from each filter has comparable variance
	      over the training set
	      
the keyword below tells GPZ where to find the list of unknown 'test' galaxies
#test /Users/noldor/physics/gaussianprocess/gpQR/data/flux_simulated/f25pct/f25pct_10_test_set.sav

the keyword below tells GPZ where to find the training set of galaxies with
known spectroscopic redshifts
#train /Users/noldor/physics/gaussianprocess/gpQR/data/flux_simulated/f25pct/f25pct_10_training_set.sav

the keyword below tells GPZ where to find the validation set of galaxies with
known spectroscopic redshifts
#valid /Users/noldor/physics/gaussianprocess/gpQR/data/flux_simulated/f25pct/f25pct_10_validation_set.sav


#output results/test_output.sav -- where to put the output

#status results/test_status.sav -- the status file records GPZ's progress as it
goes so that, if the code crashes, you can (hopefully) tell where and maybe even
why.  If you do not specify a name, GPZ defaults to "generic_status_file.sav"


#hyperparams 0 70 
there are two hyperparameters that must be set: the l^2 length scale in the
covariogram (see equation 8 of our paper), and the number of nearest neighbor
galaxies used to generate P(z_p).  If you want to set them by hand, you use this
keyword.  The order of the hyperparameters is the log_10 of l^2 followed by the
number of nearest neighbors

if you would rather GPZ use the validation galaxies to optimize these parameters
(using a simple grid search) you can use the keywords below:

!#llrange -2.0 2.0 4.0 -- the minimum value of log_10(l^2), the maximum value
                           of log_10(l^2), the number of steps to take between
			   them when doing the grid search

!#kkrange 30 90 20 -- the minimum number of nearest neighbors, the maximum
                        number of nearest neighbors, the size of the steps to
			take between them (in this configuration, GPZ will step
			by 20 in the number of nearest neigbhors)


!#only1 -- only allow GPZ to return unimodal P(z_p)
!#prior_width -- a multiplicative factor multiplying K_max in the prior  on the
                 bi-modal model (see equation 19).  This defaults to 1

!#zmax -- set the maximum allowed value of z_p in P(z_p).  Defaults to 5.