README file for using Professor John Rust's GAUSS programs
that compute NLS, FGLS or ML esimates of binary choice model
for Econ551 Problem Set 5, Q4

You should have the following files:
1)  estimate.gpr
2)  log_mle.g (ML estimator for logit)
3)  log_nls.g (NLS estimator for logit)
4)  log_fgls.g (FGLS estimator for logit)
5)  prb_mle.g (ML estimator for probit)
6)  prb_nls.g (NLS estimator for probit)
7)  prb_fgls.g (FGLS estimator for probit)
8)  drvchk.g (procedure to check if numerical
	      and analytical gradients match)
9)  hesschk.g (procedure to check if numberical 
	       and analytical hessian match)
10) max.g (procedure for maximization)

Note 1: File 1) is a shell program that estimate one of the models 
        2)-7) for the purpose of 8)-10) (derivative check, hessian
        check, or estimation).  When the shell program is run, 
        you would be asked to indicate which model you want to run
        and for what action.  

Note 2: Each of files 2)-7), name.g, when called would create an 
	output file name.out that stores all output generated in 
	the last iteration, and save the estimate as a vector name.fmt

Note 3: To do MLE, write the procedures log/prb_mle.g and run estimate.gpr.
	To do NLS, write the procedures log/prb_nls.g and run estimate.gpr.
	To do FGLS, first do NLS (write log/prb_nls.g and run estimate.gpr),
	            which would produce the first stage estimate. 
		    Then write log/prb_fgls.g, and run estimate.gpr again.

GOOD LUCK AND ENJOY!