Differences

This shows you the differences between two versions of the page.

--- fp:homework04 [2022/05/03 18:10]
pdmatei
+++ fp:homework04 [2023/05/09 15:27] (current)
pdmatei
@@ Line 1: / Line 1: @@
-===== H04. Final homework =====
+===== Homework 4. Five in a row AI =====
-For the final homework, you **may** choose between two options:
+Homework 4 consists in developing a complete five-in-a-row game, together with an AI and a very basic interface.
-  * **A. Lab homework (0.5p)**. All students which **did not submit** Homework 3, (submitted a solution which did not compile or got under 30p), may choose this homework. The **maximum** number of points that may be achieved is 0.5p. See details below.
-  * **(B) Tic Tac Toe part 2 (1p)**. All students which **submitted** Homework 3 or a consistent part of it may choose this homework. See details below.
+  * For this stage, you will decide on the structure of your project, as well as write tests to guide your implementation.
+  * There are three parts of this homework, which will be evaluated independently;
+  * **The evaluation will take into account your working code, AS WELL AS THE TESTS THAT YOU WRITE** (If you write insufficient tests, an otherwise working code may not be graded).
+  * There will be no other partial points, other than those mentioned explicitly in the statement.
-===== A. Lab homework (0.5p) =====
+==== 4.1. Board analysis (0.2p) ====
-The lab homework consists in providing **complete** solutions to labs [[fp:lab04|4]], [[fp:lab05|5]], [[fp:lab06|6]], [[fp:lab07|9]] and [[fp:lab08|10]], except for exercises marked with **(!)**.
+For this part, you need to implement a function ''sequences'', that returns a map of the form: ''(5,a), (4,b), (3,c), (2,d)'' where:
-  * **Each completed lab will be graded with 0.1p**
+  * ''a'' is the number sequences of length 5 that the player has established (on lines, columns or diagonals).
-  * **There will be no partial points for partially solved labs** - each lab has to be entirely complete to be graded.
+  * ''b'' is the number of sequences of length 4 **which can be filled-out in order to achieve a 5-sequence**. For example: ''XX XX'' is such a sequence for player ''One'', however ''XX0XX'' is not, since this cannot be filled-out in order to achieve a 5-sequence.
-  * **Solutions will be presented live and graded during lab** - no solutions sent by email or MS Teams will be graded.
+  * ''c'' (resp. ''d'') is the number of sequences of length 3 (resp. 2) which can be filled-out in order to achieve a 5-sequence.
-  * **Solutions should be solved 100% individually**. If some exercises or **parts** of them are taken from other sources, the **entire homework** will not be graded.
-  * Deadline: **Friday 27 May**
+  * Proper testing is **essential** for this function, as the AI will take decisions based on it. It will be harder to spot problems with ''sequences'' in a later stage of the project.
+  * You can replace the implementation of ''winner'' with a call to this function, to make the code more compact and avoid partial re-implementations.
-===== B. Tic Tac Toe part 2 (1p) =====
+<code scala>
+def sequences(p: Player)(b: Board): Map[Int,Int] = ???
+</code>
-==== B.1. Board analysis (0.3p) ====
+==== 4.2. Functionality (0.35p) ====
-==== B.2. Functionality (0.3p) ====
+For this part, you need to implement a functional project which consists of a running loop:
+  * Asks for the player to supply a move (expressed as a pair (x,y) of coordinates to play, or in any other manner that you prefer).
+  * Plays the opponent move via the AI
+  * Displays the board, and repeats, until either player wins.
-==== B.3. Game AI (0.4p) ====
+More details will be available during lecture.
-.2p bonus for the best AI.
+** For this particular part, tests are not necessary**.
+==== 4.3. Game AI (0.45p) ====
+This part is at the very core of the project, and essentially consists of a function ''play'', which implements the AIs decision of selecting a move.
+<code scala>
+def play(p: Player, b: Board): Board = ???
+</code>
+**You may slightly change the signature of this function**, to accommodate your project.
+You may start with a very basic (trivial) implementation, which will allow you to move on with **4.2.** and then refine it after **B.2.** is complete.
+We offer no guidelines about how this function should be implemented, this is entirely up to you, but you may consider:
+**Possible strategies:**
+  * You may implement ''play'' by simply evaluating the sequences that are available in the current play, as well as those of the opponent. This **local** strategy may become complicated, require a lot of code, as the game may unfold in different ways. It is also important to have a good strategy for starting the game (e.g. a play in either corner of the board is a bad way to start).
+  * The //standard// approach, is to implement a [[https://en.wikipedia.org/wiki/Minimax|Minimax Tree]] which allows you to inspect the game unfolding for a few moves in advance (e.g. 3 moves). More details will be given during lecture.
+  * Very basic, but functional AI: **0.1p**
+  * Intermediate AI (which may win in some cases): **0.2p**
+  * Minimax-based AI **0.45p**
+==== 4.4. Bonus (0.2p) ====
+During the last lecture, we will play a tournament of Five-in-a-row, in which all your AI programs will compete.
+**The author the best AI implementations will be awarded a bonus of 0.2p**.
+** The winner of the tournament will receive a maximal grade of 10 for the entire lecture (no attendance to the exam will be necessary)**.