This factor's score, using the neighbors' values from the given assignment, not necessarily their current values.
This factor's score, using the neighbors' values from the given assignment, not necessarily their current values..
Return a record of the current values of this Factor's neighbors.
This factor's contribution to the unnormalized log-probability of the current possible world.
The number of variables neighboring this factor.
The Nth neighboring variable of this factor.
Returns the collection of variables neighboring this factor.
Return this Factor's sufficient statistics for the values in the Assignment.
Return an object that can iterate over all value assignments to the neighbors of this Factor
Return an object that can iterate over all value assignments to the neighbors of this Factor
Return the score and statistics of the current neighbor values; this method enables special cases in which it is more efficient to calculate them together.
Return this Factor's sufficient statistics of the current values of the Factor's neighbors.
In order to two Factors to satisfy "equals", the value returned by this method for each Factor must by "eq".
In order to two Factors to satisfy "equals", the value returned by this method for each Factor must by "eq". This method is overridden in Family to deal with Factors that are inner classes.
True iff the statistics are the values (without transformation), e.
True iff the statistics are the values (without transformation), e.g. valuesStatistics simply returns its argument.
Does this Factor have the given variable among its neighbors?
Does this Factor have any of the given variables among its neighbors?
Return the score for Factors whose values can be represented as a Tensor, otherwise throw an Error.
Return the score for Factors whose values can be represented as a Tensor, otherwise throw an Error. For Factors/Family in which the Statistics are the values, this method simply calls statisticsScore(Tensor).
Given a Tensor representation of the values, return a Tensor representation of the statistics.
Given a Tensor representation of the values, return a Tensor representation of the statistics. We assume that if the values have Tensor representation that the StatisticsType does also. Note that (e.g. in BP) the Tensor may represent not just a single value for each neighbor, but a distribution over values
A single factor in a factor graph. From a Factor you can get its neighboring variables, the factor's score using the neighboring variable's current values, the factor's score using some Assignment to the the neighboring variable's values, sufficient statistics,