Lab 11: Playing Tic-Tac-Toe

Overview

In this lab, we will implement a HashTable for use in a learning algorithm for Tic-Tac-Toe.

Materials

IntelliJ
Lab partner

Setup

Download the skeleton for this project.
Unpack the code into a new IntelliJ Java project.

Description

HashTables are an efficient implementation of the Map interface, with the ability to lookup items in constant time on average. For this lab, we will be implementing a HashTable to be used in the context of an Artificial Intelligence game player.

In particular, we will be learning how to play the game Tic-Tac-Toe. After a game is played, each board state along the path from the initial board to the final board will be recorded as a win for the winning player. If the board state has never been encountered before, then a new entry is created in a HashTable with equal win/loss/draw counts, and then updated based on this play. Later board states will be influenced more than earlier board states, to account for the recency of the position and opportunity for alternate play choices.

As more games are recorded, the AI will build up a probabilistic picture of the game tree. The user can elect to train multiple iterations at once with the Train button, which will play random games of the computer vs itself. Using the learned knowledge, the AI Move button will select the move that has shown the highest probability of success.

Besides the HashTable to store board states, another HashTable is used to record a GUI Rectangle for each position on the board. To guide the player, those moves more likely than average to succeed will be shaded green, while those less likely to succeed will be shaded red.

In your code below, you should follow good coding practices, for example, noticing when you are writing essentially the same code twice and abstracting this pattern into a private method.

Step 1 - index

The first method we must implement for the HashTable is the index method. When passed a key, it will return the appropriate HashTable index for that key using the hashCode() of the key modulo by the length of the array, then taking the absolute value.

A successful solution for this step will pass the testIndex test case.

Step 2 - put

The second method we must implement for the HashTable is the put method. When passed a key-value pair, first, the associated index into the allocated array must be found using the index method from Step 1.

If there is no entry at the found location, a new HashNode is created to store the key and value. However, if there already exists a HashNode at that location, resolve the collision by walking the linked list of HashNodes at that location, looking for either a matching key or the end of the chain.

If a match is found, replace the old value for that node with the new value and exit the loop. If the end of the chain is reached without finding a match for the key, add a new HashNode at the beginning of the list, using the current array entry as the next in the constructor.

Only when a new HashNode is created should the size counter be incremented.

Note: You will not pass any tests at this point. Both put and get are needed to really test out your code.

Step 3 - get

This method finds the associated index into the allocated array using the index method from Step 1. Then, following the separate chaining collision resolution method described above, we walk through the HashNodes found, until either a matching key is identified and the method returns the value of this node, or we encounter a null position and the method returns Optional.empty().

A successful solution for Steps 2 and 3 will pass testPutGet, testDuplicate, and testVeryBad.

Step 4 - allKeys

While the keys cannot be returned in order using a HashTable, it is feasible to loop through the array and collect all the keys in this unordered fashion. Note that you must collect all the keys in the linked list, not just the key of the first element. Store each key when found in an ArrayList, and return this list.

A successful solution will pass testKeys.

Step 5 - resizing

As we add more items to our Map, the efficiency will degrade. We need to revise the put method to account for this.

When the load factor of the HashTable, determined by the number of items stored in the HashTable divided by the capacity of the array, exceeds MAX_LOAD, the put method should double the length of the array.

Once the array length is doubled, ‘reput’ all the entries into the new array (you will need to index() them again since the length of the array has changed).

A successful solution will pass testCapacityIncrease.

Step 6 - Playing Tic-Tac-Toe

Once your implementation passes the unit tests, you are ready to play Tic-Tac-Toe. Record the following in a Evaluation Document.

Experimentally determine the number of training iterations necessary for the AI to never be defeated when you play against it.
Discuss the strategy you attempted to use to defeat the AI.
Characterize the strategy the AI manifests.

Grading

To Partially Complete this lab, complete Steps 1, 2, 3, and 4.
To Complete this lab, do the above and Steps 5 and 6.