|
libmcts
A Monte Carlo Tree Search Library
|
| ▼Nmcts | |
| CAvgBackpropagationStrategy | Adds the backpropagated value to a running stats object that calculates the ev, mean and variance |
| CIBackpropagationStrategy | A backpropagation strategy is used in each node. when a terminal node is simulated, its value is propagated back through every node on the path. every node saves that value according to the implementation of a specific on_backpropagate function |
| CInnerNode | Implements most methods of the inode interface. some methods have to be implemented problem dependend. Inner nodes are all nonterminal nodes |
| CINode | Basic interface for a node in the tree. Every node has to implement this functions |
| CISelectionStrategy | Interface for all selection strategies. A selection strategy selects one child of a parent node according to implementation specific metrics |
| CISimulationStrategy | Interface class for simulation strategies. this strategies are called in terminal nodes to evaluate the context in them |
| CIVisitor | Implements interface for visitor objects |
| CLeafNode | Implements methods of the inode interface that are used for terminal contexts. all leaf nodes are terminal |
| CMaxFunctionSelector | Selector selects the child of a node that maximises a given metric |
| CMaxValueSelector | Selects the child which maximises ev |
| CMCTS | Mainclass to interact with the library. it holds the root of the searchtree |
| CMinFunctionSelector | Selects a node that minimises a given metric |
| CMinSampleSelector | Selects the child with the lowest number of samples |
| CRootNode | Special node that is the root of each tree. Its merely a wrapper that wrapps around a inner node |
| CRunningStats | Computes variance, mean and standart deviation. http://www.johndcook.com/standard_deviation.html http://www.johndcook.com/skewness_kurtosis.html |
| CSamplingSelector | Selector that chooses random children uniformly |
| CSamplingToFunctionSelector | Uses sampling selector until threshold is reached. After that a supplied selection strategy is applied |
| CUCTSelector | Selector that selects the childnode, that maximises the ratio between samples and ev of the parent node |
1.8.15