An official website of the United States government
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
Secure .gov websites use HTTPS
A lock (
) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.
Generates a new sample file (MTGS64) that contains values of heritabilities (fraction of phenotypic variance for other variances) and correlations
Reads MTGSPREP output to determine model so that calculation of phenotypic variance is determined without user intervention
Linked with MTGSSUB.FOR, MTGSRSB.FOR, and timing subroutines
Batch and make files for automated compiling and linking included
LAGCOR.FOR
Calculates lags of samples written to MTGS61 or MTGS64; user can specify additional burn-in rounds if MTGSRUN value is considered insufficient
Requests gap between lag values calculated and number of correlations to calculate; e.g., if a value of 5 is used for lag and 20 correlations are requested, correlation of samples would be written 5, 10, 15, ..., 100 records apart in sample file
Remember that samples in sample files may already be thinned depending on MTDFRUN option chosen; this thinning rate should be considered when determining any additional burn-in and when deciding on gap between values calculated.
User specifies original frequency of sample writing used in MTGSRUN so that correct lags can be calculated based on original (unthinned) chain and lags listed in output file correspond to original chain; e.g., with a MTGSRUN thinning rate of 20 (every 20th sample written to MTGS61) and a requested gap of 5 records for calculating a total of 10 correlations, correlations of samples would be written 5 apart corresponding to lags of 5, 10, 15, ..., 50 records apart in thinned chain and lags of 100, 200, ..., 1,000 samples apart in original chain
MNSAMP.FOR
Calculates mean of sampled values in MTGS61 or MTGS64, which is especially useful to monitor a Gibbs chain for a run that has not completed
Allows user to specify additional burn-in rounds if MTGSRUN value is considered insufficient
PULLDAT.FOR
Used to read sampling data from MTGS61, 62, or 64 and copy that information to a user-specified text file; user specifies number of variables to copy to text file (with a shortcut for writing all variables).
Allows user to specify additional burn-in rounds if MTGSRUN value is considered insufficient
Writes sample information for a user-specified interval between samples read from sample file
PULLMAT.FOR
Used similarly to PULLDAT.FOR except that output file is a user-specified MATLAB file
Resulting dataset can be loaded into MATLAB using "load filename" command; file name must have .mat as file extension
Matrix will be called gs (lower case), and rows will contain string of values for each parameter specified (rather than columns) to minimize memory usage (the only way to write the data file with columns containing the variable is to read the whole data set in, which will often be impossible); result is that number of samples in sample file (MTGS61 or 64) must be known in advance
Fortran program written by Raftery and Lewis; description comes from program header
Calculates number of iterations required in a Markov-chain Monte Carlo (e.g., Gibbs sampling) run; user has to specify precision required
Returns number of iterations required to estimate posterior cdf of q-quantile of quantity of interest (function of parameters) to within ±r with probability s
Also gives number of "burn-in" iterations required for conditional distribution given any starting point (of derived 2-state process) to be within ε of actual equilibrium distribution
Not distributed with utilities program but available directly from StatLib.
Contact
Please direct questions, comments, and program problems to Curt Van Tassell
