Accessing belief inputs

Element expectations

 
Usage

• ex  operator; one argument N in parenthesis. N is the name of an element.
• ex  operator; two arguments S, N in parenthesis. S is a expectation store number, N is the name of an element.

The first form of the usage returns the expectation of the element , with the expectation taken from the current default expectation store. The second form of the usage returns the expectation for the element N with the expectation taken explicitly from expectation store S.

If N is not a built element, then the list of functionally specified expectations is searched to determine (and return if present) whether an expectation has been specified functionally via the FE:  command.

Notice that you must be careful when using the first form in defining a function, as the default expectation store may be changed (by you) between definition of the function and evaluation of it. The default expectation store here means the default expectation store at the time of evaluation; this might be different to the default at the time of definition. We recommend that you only use the second form when defining functions.  

Correlations between elements

 
Usage

• corr  operator; two or three arguments I, , and optionally in parenthesis and separated by commas. I is an integer representing a valid belief store, and and are the names of elements.

This is used to return a belief input from belief store I in correlation form. For example, returns the correlation between the elements and in belief store 1. The abbreviated form (allowed only to preserve compatibility with the related operator var ) with only one argument will always return a value of unity.

Correlations declared implicitly via the FVAR:  command may also be accessed using this operator: the built store of names is searched firstly, and the virtual store secondly. For such usage, use of the correlation form assumes that you have made all necessary definitions. For example, use of implies that all the necessary quantities [x with x], [y with y] and [x with y] have been specified beforehand using the FVAR:  command. An error will occur otherwise. Note that for both operators var  and corr , the built store of names is searched firstly, and the virtual store secondly. Note also that the beliefs involved are not checked for coherence. Under these circumstances, reported correlations greater than unity are feasible.  

Variances and covariances between elements

 

Usage

• var  operator; two or three arguments I, , and optionally in parenthesis and separated by commas. I is an integer representing a valid belief store, and and are the names of elements.
• var  operator; one or two arguments , and optionally in parenthesis and separated by commas. and are the names of elements.

The first form of the usage is used to return a belief input from belief store I in covariance form. For example, returns the covariance specified between the elements and in belief store 1, . The abbreviated form with only one argument returns a variance. For example, is the abbreviation of and returns the variance specified for in belief store 2, .

Covariances declared functionally via the FVAR:  command may also be accessed using this operator: the built store of names is searched firstly, and the functionally defined store secondly.

The second form of the usage does not require explicitly a belief store number. Instead, the default variance store number is used. This store number is 1 by default, but can be changed by using the v  argument to the CONTROL:  command. Notice that you must be careful when using the second form in defining a function, as the default variance store may be changed (by you) between definition of the function and evaluation of it. The default variance store here means the default variance store at the time of evaluation; this might be different to the default at the time of definition. We recommend that you only use the first form when defining functions.