## UAI 2016 Inference EvaluationOrganizer: Vibhav Gogate (Email: vgogate at hlt dot utdallas dot edu)
## Result File FormatThe first line must contain only the task solved: PR <SOLUTION> -BEGIN- <SOLUTION> The solution format are as follows depending on the task **Partition function, PR**: Line with the value of the`log10`of the partition function. For example, an approximation`log10 Pr(e) = -0.2008`which is known to be an upper bound may have a solution line:
-0.2008 **Most probable explanation, MPE**: A space separated line that includes:the number *n*of model variables, andthe MPE instantiation, a list of value indices for all *n*variables.
For example, an input model with 3 binary variables may have a solution line: 3 0 1 0 **Marginals, MAR**: A space separated line that includes:The number of variables in the model. A list of marginal approximations of all the variables. For each variable its cardinality is first stated, then the probability of each state is stated. The order of the variables is the same as in the model, all data is space separated.
For example, a model with 3 variables, with cardinalities of 2, 2, 3 respectively. The solution might look like this: 3 2 0.1 0.9 2 0.3 0.7 3 0.2 0.2 0.6 **Marginal MAP, MMAP**: A space separated line that includes:The number *q*of query variablesthe most probable instantiation, a list of variable value pairs for all *q*variables.
For example, if the solution is an assignment of 0, 1 and 0 to three query variables indexed by 2 3 and 4 respectively, the solution will look as follows 3 2 0 3 1 4 0 If a solver does not produce a solution by the given time, it would be considered as having failed on the instance. |