Created
August 7, 2015 00:58
-
-
Save nwertzberger/eaebd778dd3116ec873a to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
public class OptimizedTransitionCalculator { | |
private static final Logger logger = LoggerFactory.getLogger( | |
OptimizedTransitionCalculator.class | |
); | |
public Policy generateNewPolicy( | |
Set<Action> actions, | |
Map<State, Double> expectedUtilities, | |
Map<StateAction, ? extends Transition> transitions) { | |
Map.Entry<State, Double>[] expectedUtilityEntries = expectedUtilities | |
.entrySet() | |
.toArray(new Map.Entry[0]); | |
Action[] actionArray = actions.toArray(new Action[0]); | |
return new Policy( | |
expectedUtilities.keySet().parallelStream().collect( | |
toConcurrentMap( | |
(state) -> state, | |
(state) -> maxAction( | |
state, actionArray, transitions, expectedUtilityEntries | |
) | |
) | |
) | |
); | |
} | |
/** | |
* This function is the most used function in this program. | |
* Every cycle saved here counts for a lot. Optimizations are welcome. | |
* | |
* I have smashed all subroutines into this function. I've commented the | |
* delineation. | |
* | |
* @param state | |
* @param actions | |
* @param transitions | |
* @param expectedUtilities | |
* @return | |
*/ | |
private Action maxAction( | |
final State state, | |
final Action[] actions, | |
final Map<StateAction, ? extends Transition> transitions, | |
final Map.Entry<State, Double>[] expectedUtilities) { | |
double maxExpectedUtility = Double.NEGATIVE_INFINITY; | |
Action maxAction = null; | |
for (Action act : actions) { | |
// Calculate the action's expected utility in this state. | |
Transition t = transitions.get(new StateAction(state, act)); | |
double expectedUtility = 0.0; | |
for (Map.Entry<State, Double> entry : expectedUtilities) { | |
expectedUtility += t.getStateProbability(entry.getKey()) * entry.getValue(); | |
} | |
// Update maximal expected utility | |
if (expectedUtility > maxExpectedUtility) { | |
maxAction = act; | |
maxExpectedUtility = expectedUtility; | |
} | |
} | |
return maxAction; | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment