Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 118 views

    Sudoku Report

    Uploaded by

    wdusln
    This document summarizes an approach to solving Sudoku puzzles using constraint satisfaction problems. It describes Sudoku puzzles and represents them as constraint satisfaction problems. It then summarizes four approaches to solving Sudoku-CSPs: backtracking, backtracking with the MRV heuristic, backtracking with MRV and forward checking, and backtracking with MRV and constraint propagation. It provides pseudocode for the approaches and discusses implementations. It also provides results comparing the four approaches on three test Sudoku puzzles.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Sudoku Report

    Uploaded by

    wdusln
    118 views5 pages
    This document summarizes an approach to solving Sudoku puzzles using constraint satisfaction problems. It describes Sudoku puzzles and represents them as constraint satisfaction problems. It then summarizes four approaches to solving Sudoku-CSPs: backtracking, backtracking with the MRV heuristic, backtracking with MRV and forward checking, and backtracking with MRV and constraint propagation. It provides pseudocode for the approaches and discusses implementations. It also provides results comparing the four approaches on three test Sudoku puzzles.

    Original Description:

    sudoku report

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes an approach to solving Sudoku puzzles using constraint satisfaction problems. It describes Sudoku puzzles and represents them as constraint satisfaction problems. It then summarizes four approaches to solving Sudoku-CSPs: backtracking, backtracking with the MRV heuristic, backtracking with MRV and forward checking, and backtracking with MRV and constraint propagation. It provides pseudocode for the approaches and discusses implementations. It also provides results comparing the four approaches on three test Sudoku puzzles.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    118 views5 pages

    Sudoku Report

    Uploaded by

    wdusln
    This document summarizes an approach to solving Sudoku puzzles using constraint satisfaction problems. It describes Sudoku puzzles and represents them as constraint satisfaction problems. It then summarizes four approaches to solving Sudoku-CSPs: backtracking, backtracking with the MRV heuristic, backtracking with MRV and forward checking, and backtracking with MRV and constraint propagation. It provides pseudocode for the approaches and discusses implementations. It also provides results comparing the four approaches on three test Sudoku puzzles.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 5

    Solving Sudoku with Constraint Satisfaction Problem

    Ali Haider Zaveri - 108661497


    Jagat Sastry - 108721027

    Sudoku explained
    Sudoku is a number placement puzzle. In this puzzle you are given an N x N grid
    of cells. The grid itself is composed of M x K sub-grids. You can place a single
    digit, drawn from 1 to N, in any cell. Initially the grid will have some of its cells
    partially filled. The objective of the puzzle is to complete the grid so that:
    1. Every cell contains a digit.
    2. No digit appears twice in any row, column of the N x N grid or in any row,
    column of any of the M x K sub-grid.

    Sudoku as a Constraint Satisfaction Problem


    Each cell on the N x N grid can be thought of as a variable. Constraint on each
    variable is that no two variables in the same row, column or M x k sub-grid can
    have the same number([1-N]) assigned to it. In this problem each constraint is a
    binary constraint as only two variables are involved in each constraint.
    We have solved Sudoku by modeling it as a Constraint Satisfaction Problem by
    using the four methods which are explained in detail below.
    Pseudo Code:
    SudokuBT(board){
    inputs: board, the Sudoku board-each cell of the board is a variable
    local variable: val-domain values of variables
    cell-board cell
    If AC-3 fails, return false //For CSP arc consistency
    select cell //in order of minimum remaining values except for basic backtracking
    for each val in Domain[cell] do
    if(valid assignment)
    assign val to cell
    If fwd checking fails, continue //Only for fwd checking case
    if(call SudokuBT(board))
    return true
    else
    reset cell to 0
    end for
    return false
    }

    Backtracking
    SudokuGridDFS.java
    Backtracking is a recursive algorithm. Using it without any heuristic is a nave
    approach for solving any constraint satisfaction problem.
    From the above pseudo code it can be seen that the backtracking approach
    makes recursive calls until it finds the board inconsistent. When it finds the board
    inconsistent it backtracks to find another assignment of variables for which the
    boards will be consistent. Pseudo code for this has been given above.

    Backtracking + MRV heuristic


    In the nave backtracking approach there was no logic behind choosing next
    variable for assignment.
    The MRV (Minimum Remaining Values or Most Constrained Variable) heuristic is
    used with backtracking to help choose the next variable for assignment. The
    most restricted variable is the one which has the least number of domain values
    remaining to which it can be assigned to.
    The pseudo code is similar to normal backtracking with the only difference being
    in selecting the next variable for assignment.
    Implementation specifics and optimizations:
    SudokuGridMRV.java
    We have created a class called Cell.java the objects of which signify the variables
    of the problem
    Cell.java has the following attributes
    - m_val: stores the value of the variable
    - m_constraints: A list of values that the variable cannot take.
    - m_row,m_col: Position of cell on the grid
    Instead of storing all the values that a cell can take, we store only those that it
    cannot take, thus minimizing the memory required and constraint checking time.

    Backtracking + MRV + Forward Checking


    We use the pseudo code we had used for backtracking but with the following
    modifications.
    As seen in the results section, backtracking used with the MRV heuristic does
    help in considerable reduction in the number of expanded states (cells). Forward
    checking helps to determine inconsistent states earlier. After assigning a value to
    the most restricted variable, if any of its neighbours (that have not been
    assigned a value as yet), i.e. the variables with which it is involved in a
    constraint cannot be assigned a value, we backtrack. This helps in determining
    inconsistencies earlier as we do not have to wait for the assignment to actually
    fail.

    Implementation specifics:
    SudokuGridFwdChecking.java
    After assigning a value to the most restricted cell we check the constraint list of
    each of its neighbours that have not been assigned a value as yet. If the
    constraint list of any of these neighbours is equal to N i.e. if the neighbouring cell
    cannot take any more values then we backtrack.

    Backtracking + MRV + Constraint Propagation


    We use the pseudo code we had used for backtracking but with the following
    modifications.
    We have used constrain propagation as a predecessor to every assignment. In
    constraint propagation we are seeing whether the arc between two neighbours is
    consistent or not. The arc between two neighbours is consistent if for every value
    in the domain of variable X there is a value in the domain of variable Y that it can
    take without compromising the constraint test.
    Implementation Specifics:
    SudokuGridMRVCP.java
    We use the AC-3 algorithm to check arc consistency. This helps to detect failures
    early even before doing an assignment.

    Contd in the next page...

    Results:
    For Test 1
    6 2 3
    _,_,6,_,5,_
    _,1,_,2,_,_
    _,_,1,_,_,_
    _,_,_,3,_,_
    _,_,4,_,1,_
    _,2,_,4,_,_

    Method

    Consistency Checks

    Time(ms)

    Memory(MB
    )

    Backtracking

    474

    ~0

    1.23

    Backtracking + MRV

    31

    ~0

    0.62

    Backtracking + MRV + Fwd


    check

    30

    ~0

    0.62

    Backtracking + MRV + CP

    27

    ~0

    1.25

    For test case 2


    9 3 3
    _,8,_,9,_,_,_,2,_
    9,_,_,_,6,_,_,_,8
    _,4,6,_,_,2,1,9,_
    _,_,5,_,1,_,_,_,3
    _,7,_,2,_,8,_,6,_
    6,_,_,_,4,_,7,_,_
    _,3,4,7,_,_,9,5,_
    7,_,_,_,8,_,_,_,2
    _,6,_,_,_,4,_,3,_

    Method

    Consistency Checks

    Time(ms)

    Memory(MB)

    Backtracking

    61386

    ~0

    3.7

    Backtracking +
    MRV

    50

    ~0

    0.62

    Backtracking +
    MRV + Fwd Check

    51

    ~0

    0.62

    Backtracking +
    MRV + CP

    51

    ~0

    3.7

    For test case 3


    12 3 4
    _,_,_,_,8,3,_,_,6,_,7,_
    _,4,_,7,_,_,_,_,5,2,_,12
    9,_,_,11,_,5,7,_,_,_,10,4
    11,7,_,2,_,10,_,_,_,_,_,5
    _,_,12,_,_,_,_,_,_,7,_,_
    _,_,_,_,12,_,_,5,_,11,_,_
    _,_,3,_,9,_,_,1,_,_,_,_
    _,_,4,_,_,_,_,_,_,12,_,_
    6,_,_,_,_,_,5,_,3,_,8,2
    12,2,_,_,_,9,6,_,4,_,_,3
    1,_,8,4,_,_,_,_,2,_,12,_
    _,9,_,3,_,_,12,2,_,_,_,_

    Method

    Consistency
    Checks

    Time(ms)

    Memory(MB)

    Backtracking

    163104392

    ~54596

    0.62

    Backtracking +
    MRV

    11576

    ~0

    4,32

    Backtracking +
    MRV + Fwd Check

    12990

    ~0

    6.16

    Backtracking +
    MRV + CP

    6278

    5041

    21.5

    Running Tests
    Run the code in the following way
    Change the method as required in TestSudoku.METHOD java code and call the
    java code in the following manner
    java ai.TestSudoku test_case_file
    where test_case_file has entry like
    6 2 3
    _,_,6,_,5,_
    _,1,_,2,_,_
    _,_,1,_,_,_
    _,_,_,3,_,_
    _,_,4,_,1,_
    _,2,_,4,_,_
    000

    You might also like