Option allowing the user to import the prior distribution from another model via its posterior mode results file and inserting the prior data into the posterior mode results file of the current model. The posterior mode results file is located under the "mode" directory of the current model directory (see the output selection help page) and is a mat-file with the main results from the posterior mode estimation.
The idea behind this option is that the learning parameter(s) which is (are) estimated need not be well identified. One such parameter is the autoregressive parameter used by the Kalman filter for the belief coefficients and if it is not well identified a uniform prior may not work well when estimating the posterior mode. The beta prior equivalent to a standard uniform has mean 0.5 and standard deviation somewhat below 0.288675. By lowering the standard deviation of this distribution one obtains some curvature of the prior which may help to obtain an estimate of the autoregressive parameter which is at least well behaved. By importing the prior from a model which has the same priors for all other estimated parameters and a standard uniform for the autoregressive parameter, the desired prior may then be used for posterior sampling and other computations which involve the prior distribution.
Page url: http://www.texlips.net/yada/help/index.html?import_prior_data.htm