Description
Evaluation Details
We will test your code electronically. You will be supplied with a testing script that will run a subset of the tests (tests.py). If your code fails all of the tests performed by the script (using Python version 3.5.2), you will receive zero marks. It’s up to you to create test cases to further test your code—that’s part of the assignment!
When your code is submitted, we will run a more extensive set of tests. You have to pass all of these more elaborate tests to obtain full marks on the assignment.
We will set a timeout of 60 seconds per board during marking. This 60 second time limit includes the time to create a model and find a solution. Solutions that time-out are considered incorrect. Ensure that your implementations perform well under this limit, and be aware that your computer may be quicker than teach.cs (where we will be testing). We will not conduct end to end tests on boards with constraints whose domain is larger than six.
Your code will not be evaluated for partial correctness, it either works or it doesn’t. It is your responsibility to hand in something that passes at least some of the tests in the provided testing script.
Make certain that your code runs on teach.cs using Python3 (version 3.5.2) using only standard imports. This version is installed as “python3” on teach.cs. Your code will be tested using this version and you will receive zero marks if it does not run using this version.
Do not add any non-standard imports from within the Python file you submit (the imports that are already in the template files must remain). Once again, non-standard imports will cause your code to fail the testing and you will receive zero marks.
Do not change the supplied starter code. Your code will be tested using the original starter code, and if it relies on changes you made to the starter code, you will receive zero marks.
Introduction
There are two parts to this assignment.
-
Propagators. You will implement two constraint propagators—a Forward Checking constraint prop-agator, and a Generalized Arc Consistence (GAC) constraint propagator—and three heuristics— Minimum-Remaining-Value (MRV), Degree (DH), and Least-Constraining-Value (LCV).
-
Models. You will implement three different CSP models: two grid-only KenKen models, and one full KenKen puzzle model (adding cage constraints to grid).
What is supplied
cspbase.py. Class definitions for the python objects Constraint, Variable, and BT.
propagators.py. Starter code for the implementation of your two propagators. You will modify this file with the addition of two new procedures prop FC and prop GAC.
heuristics.py. Starter code for the implementation of the variable ordering heuristic, MRV; and the value heuristic, LCV. You will modify this file with the addition of the new procedures ord mrv, ord dh, and val lcv.
|
Assignment 2, University of Toronto, CSC384 – Intro to AI, Winter 2018 |
3 |
-
11+
2/
20x
6x
3-
3/
240x
6x
6x
7+
30x
6x
9+
8+
2/
|
5 |
6 |
3 |
4 |
1 |
2 |
|
6 |
1 |
4 |
5 |
2 |
3 |
|
4 |
5 |
2 |
3 |
6 |
1 |
|
3 |
4 |
1 |
2 |
5 |
6 |
|
2 |
3 |
6 |
1 |
4 |
5 |
|
1 |
2 |
5 |
6 |
3 |
4 |
Figure 1: An example of a 6 6 KenKen grid with its start state (left) and solution (right).
kenken csp.py. Starter code for the CSP models. You will modify three procedures in this file: binary ne grid, nary ad grid, and kenken csp model.
tests.py. Sample test cases. Run the tests with “python3 tests.py”.
KenKen Formal Description
The KenKen puzzle3 has the following formal description:
KenKen consists of an n n grid where each cell of the grid can be assigned a number 1 to n. No digit appears more than once in any row or column. Grids range in size from 3 3 to 9 9.
KenKen grids are divided into heavily outlined groups of cells called cages. These cages come with a target and an operation. The numbers in the cells of each cage must produce the target value when combined using the operation.
For any given cage, the operation is one of addition, subtraction, multiplication or division. Values in a cage can be combined in any order: the first number in a cage may be used to divide the second, for example, or vice versa. Note that the four operators are “left associative” e.g., 16/4/4 is interpreted as (16/4)/4 = 1 rather than 16/(4/4) = 16.
A puzzle is solved if all empty cells are filled in with an integer from 1 to n and all above constraints are satisfied.
An example of a 6 6 grid is shown in Figure 1. Note that your solution will be tested on n n grids where n can be from 3 to 9.
3http://thinkmath.edc.org/resource/introducing-kenken-puzzles
|
Assignment 2, University of Toronto, CSC384 – Intro to AI, Winter 2018 |
4 |
Question 1: Propagators (worth 60/100 marks)
You will implement Python functions to realize two constraint propagators—a Forward Checking (FC) constraint propagator and a Generalized Arc Consistence (GAC) constraint propagator. These propagators are briefly described below. The files cspbase.py, propagators.py, and heuristics.py provide the complete input/output specification of the two functions you are to implement.
Brief implementation description: The propagator functions take as input a CSP object csp and (option-ally) a Variable newVar representing a newly instantiated Variable, and return a tuple of (bool,list) where bool is False if and only if a dead-end is found, and list is a list of (Variable, value) tuples that have been pruned by the propagator. ord mrv and ord dh take a CSP object csp as input, and return a Variable object var. val lcv takes a CSP object csp and a Variable object var as input, and returns a list of values in the domain of that variable. In all cases, the CSP object is used to access variables and constraints of the problem, via methods found in cspbase.py.
You must implement:
prop FC (worth 20/100 marks)
A propagator function that propagates according to the FC algorithm that check constraints that have exactly one uninstantiated variable in their scope, and prune appropriately. If newVar is None, forward check all constraints. Otherwise only check constraints containing newVar.
prop GAC (worth 20/100 marks)
A propagator function that propagates according to the GAC algorithm, as covered in lecture. If newVar is None, run GAC on all constraints. Otherwise, only check constraints containing newVar.
ord mrv (worth 5/100 marks)
A variable ordering heuristic that chooses the next variable to be assigned according to the Minimum-Remaining-Value (MRV) heuristic. ord mrv returns the variable with the most constrained current domain (i.e., the variable with the fewest legal values).
ord dh (worth 5/100 marks)
A variable ordering heuristic that chooses the next variable to be assigned according to the Degree heuristic (DH). ord dh returns the variable that is involved in the largest number of constraints on other unassigned variables.
val lcv (worth 10/100 marks)
A value heuristic that, given a variable, chooses the value to be assigned according to the Least-Constraining-Value (LCV) heuristic. val lcv returns a list of values. The list is ordered by the value that rules out the fewest values in the remaining variables (i.e., the variable that gives the most flexibility later on) to the value that rules out the most.
|
Assignment 2, University of Toronto, CSC384 – Intro to AI, Winter 2018 |
5 |
Question 2: Models (worth 40/100 marks)
You will implement three different CSP models using three different constraint types. The three different constraint types are (1) binary not-equal; (2) n-ary all-different; and (3) KenKen cage. The three models are
-
binary grid-only KenKen; (b) n-ary grid-only KenKen; and (c) full KenKen. The CSP models you will build are described below. The file kenken csp.py provides the complete input/output specification.
Brief implementation description: The three models take as input a valid KenKen grid, which is a list of lists, where the first list has a single element, N, which is the size of each dimension of the board, and each following list represents a cage in the grid. Cell names are encoded as integers in the range 11; :::; nn and each inner list contains the numbers of the cells that are included in the corresponding cage, followed by the target value for that cage and the operation (0=+, 1=-, 2=/, 3=*). If a list has two elements, the first element corresponds to a cell, and the second one—the target—is the value enforced on that cell.
For example, the model ((3); (11; 12; 13; 6; 0); (21; 22; 31; 2; 2); ::::) corresponds to a 3×3 board4 where
-
cells 11, 12 and 13 must sum to 6, and
-
the result of dividing some permutation of cells 21, 22, and 31 must be 2. That is, (C21/C22)/C23 = 2 or (C21/C23)/C22 = 2, or (C22/C21)/C23 = 2, etc…
All models need to return a CSP object, and a list of lists of Variable objects representing the board. The returned list of lists is used to access the solution. The grid-only models do not need to encode the cage constraints.
You must implement:
binary ne grid (worth 10/100 marks)
A model of a KenKen grid (without cage constraints) built using only binary not-equal constraints for both the row and column constraints.
nary ad grid (worth 10/100 marks)
A model of a KenKen grid (without cage constraints) built using only n-ary all-different constraints for both the row and column constraints.
kenken csp model (worth 20/100 marks)
A model built using your choice of (1) binary binary not-equal, or (2) n-ary all-different constraints for the grid, together with (3) KenKen cage constraints. That is, you will choose one of the previous two grid models and expand it to include KenKen cage constraints.
Notes: The CSP models you will construct can be space expensive, especially for constraints over many variables, (e.g., for cage constraints and those contained in the first binary ne grid grid CSP model). Also be mindful of the time complexity of your methods for identifying satisfying tuples, especially when coding the kenken csp model.
HAVE FUN and GOOD LUCK!
-
Note that cell indexing starts from 1, e.g. 11 is the cell in the upper left corner.
