/**
* Transition function. Currently, we only get the list of probabilities of each item.
*
* @param current - The current state
* @param action - The action
* @param possible - The possible state
* @return The list of probabilities where each index refers to the probability of that item
*/
private List<Double> transition(State current, Action action, State possible) {
List<Double> probs = new ArrayList<Double>(); // Probabilities
Map<Integer, Integer> currentStock, purchase, possibleStock; // Maps
double currentProb; // The current probability
Matrix currentMatrix; // The current probability matrix
int row, column; // The row and column
for (int i = 0; i < current.getState().size(); ++i) {
currentProb = 0.0;
currentStock = current.getState();
purchase = action.getPurchases();
possibleStock = possible.getState();
row = currentStock.get(i) + purchase.get(i);
column = row - possible.getState().get(i);
currentMatrix = this.probabilities.get(i);
if (column < 0
|| column >= currentMatrix.getNumCols()
|| row >= currentMatrix.getNumRows()) { // Invalid state
probs.add(0.0);
continue;
}
if (possibleStock.get(i) > 0
|| (possibleStock.get(i) == 0 && column == 0)) { // Sufficiently provided
currentProb = currentMatrix.get(row, column);
} else if (possibleStock.get(i) == 0 && column > 0) {
// Range of probabilities because user could have eaten plenty
for (int j = column; j < currentMatrix.getNumCols(); ++j) {
currentProb += currentMatrix.get(row, j);
}
}
probs.add(currentProb);
}
return probs;
}
/**
* Check if the given action is valid or not. This means check if the current action causes us to
* over-stock our fridge
*
* @param currentAction - The action we are looking at
* @param currentState - The state we are looking at
* @return true if currentAction leads to state having more than the capacity of the fridge. false
* otherwise
*/
private boolean validAction(Action currentAction, State currentState) {
int sumOfAction = sumOf(currentAction.getPurchases());
int sumOfState = sumOf(currentState.getState());
if ((sumOfAction + sumOfState) > fridge.getCapacity()) {
return false;
}
return true;
}
/**
* Check if the end state is valid. This means check that the difference between the sums of both
* states are not greater than total number of items in fridge * maximum consumption of one item.
*
* <p>Will also check if the difference between the contents of both states for each item is not
* greater than the maximum consumption
*
* @param currentState The current state
* @param endState The end state
* @return true if the end state is valid. false otherwise
*/
private boolean IsValidEndState(State currentState, State endState) {
/* The sums */
int currentSum = sumOf(currentState.getState());
int endSum = sumOf(endState.getState());
if (Math.abs(currentSum - endSum)
> (this.fridge.getMaxTypes() * this.fridge.getMaxItemsPerType())) {
return false;
}
for (int i = 0; i < currentState.getState().size(); ++i) {
if (Math.abs(currentSum - endSum) > this.fridge.getMaxItemsPerType()) {
return false;
}
}
return true;
}
/**
* Make a copy of a policy
*
* @param Map<State, Action> newPolicy - The new policy
*/
private void copyPolicy(Map<State, Action> newPolicy) {
/* Containers for clones */
State current;
Action currentAction;
this.policy = new HashMap<State, Action>();
for (State newState : newPolicy.keySet()) {
current = new State(newState.getState());
current.setCost(newState.getTemporaryCost());
currentAction = new Action(newPolicy.get(newState).getPurchases());
this.policy.put(current, currentAction);
}
}
/** Generate optimal policy */
private void generateOptimalPolicy() {
double startTime, currentTime; // Timer things
boolean alreadyRunThrough = false; // Record if we've already done one run
Set<State> toLookup; // The states to lookup
Map<Action, Double> differentValues =
new HashMap<Action, Double>(); // The different values for the current state
Map<State, Action> newPolicy = new HashMap<State, Action>(); // New policy
State newState; // New state for new policy
Action best; // The best action
startTime = Global.currentTime();
currentTime = Global.currentTime();
while ((currentTime - startTime) <= this.timeRemaining) {
if (alreadyRunThrough) {
toLookup = this.policy.keySet();
} else {
toLookup = this.possibleStates;
}
for (State currentState : toLookup) {
differentValues.clear(); // Reset different values
for (Action currentAction : this.possibleActions) {
if (!validAction(currentAction, currentState)) {
continue;
}
differentValues.put(
currentAction, valueGeneration(currentState, currentAction, toLookup));
}
best = maxArg(differentValues);
if (isBetterPolicy(currentState, differentValues.get(best))) {
newState = new State(currentState.getState());
newState.setTemporaryCost(differentValues.get(best));
newPolicy.put(newState, best);
}
}
if (CheckDifference(0.1, newPolicy)) {
copyPolicy(newPolicy);
newPolicy.clear();
alreadyRunThrough = true;
} else {
currentTime = Global.currentTime();
break;
}
}
}
/**
* Print out important values
*
* @param mode - Indicates the type of values to print out.
*/
private void printImportantValues(int mode) {
Set<?> toPrint = null; // The set to print
int currentIndex = 0; // The current index
switch (mode) {
case 0: // States
System.err.println("Printing out states");
toPrint = this.possibleStates;
break;
case 1: // Actions
System.err.println("Printing out actions");
toPrint = this.possibleActions;
break;
case 2: // Consumption
System.err.println("Printing out consumptions");
toPrint = this.possibleConsumptions;
break;
case 3: // Policies
for (State current : this.policy.keySet()) {
System.out.println(
"State "
+ (currentIndex + 1)
+ ": "
+ current.getState().toString()
+ ". Do Action: "
+ this.policy.get(current).getPurchases().toString());
currentIndex++;
}
return;
default: // Invalid mode
System.err.println("Invalid mode for printing");
System.exit(100);
}
if (toPrint == null) {
System.err.println("Something really horrible went wrong");
System.exit(150);
}
for (Object current : toPrint) {
System.out.println("Value: " + (currentIndex + 1) + current.toString());
currentIndex++;
}
}
/**
* Reward function for a state. Currently, our reward is simply the total number of items in the
* state minus the capacity of the fridge.
*
* <p>This means that we get the highest reward if we fully stock our fridge
*
* @param current - The current state
* @return The immediate reward for being in this state
*/
private double reward(State current) {
int total = sumOf(current.getState()); // The total number of items
return total - this.fridge.getCapacity();
}
