Sudoku Solver - Scilab - algorithm

I wrote a program in SciLab that solves sudoku's.
But it can only solve sudoku's that always have a square with 1 possible value.
Like very easy and easy sudoku's on brainbashers.com .
Medium sudoku's always reach a point that they do not have a square with 1 possible value.
How can I modify my code to solve these, more difficult sudoku's?
///////////////////////////////////////////////////////////////////////////
////////////////////////// Check Sudoku ///////////////////////////////
///////////////////////////////////////////////////////////////////////////
function r=OneToNine(V) // function checks if the given vector V contains 1 to 9
r = %T // this works
u = %F
index = 1
while r == %T & index < 10
for i=1 : length(V)
if V(i)==index then
u = %T
end
end
index=index+1
if u == %F then r = %F
else u = %F
end
end
if length(V) > 9 then r = %F
end
endfunction
function y=check(M) // Checks if the given matrix M is a solved sudoku
y = %T // this works too
if size(M,1)<>9 | size(M,2)<>9 then // if it has more or less than 9 rows and columns
y = %F // we return false
end
for i=1 : size(M,1) // if not all rows have 1-9 we return false
if OneToNine(M(i,:)) == %F then
y = %F
end
end
endfunction
function P=PossibilitiesPosition(board, x, y)
// this one works
// we fill the vector possibilites with 9 zeros
// 0 means empty, 1 means it already has a value, so we don't need to change it
possibilities = [] // a vector that stores the possible values for position(x,y)
for t=1 : 9 // sudoku has 9 values
possibilities(t)=0
end
// Check row f the value (x,y) for possibilities
// we fill the possibilities further by puttin '1' where the value is not possible
for i=1 : 9 // sudoku has 9 values
if board(x,i) > 0 then
possibilities(board(x,i))=1
end
end
// Check column of the value (x,y) for possibilities
// we fill the possibilities further by puttin '1' where the value is not possible
for j=1 : 9 // sudoku has 9 values
if board(j, y) > 0 then
possibilities(board(j, y))=1
end
end
// Check the 3x3 matrix of the value (x,y) for possibilities
// first we see which 3x3 matrix we need
k=0
m=0
if x >= 1 & x <=3 then
k=1
else if x >= 4 & x <= 6 then
k = 4
else k = 7
end
end
if y >= 1 & y <=3 then
m=1
else if y >= 4 & y <= 6 then
m = 4
else m = 7
end
end
// then we fill the possibilities further by puttin '1' where the value is not possible
for i=k : k+2
for j=m : m+2
if board(i,j) > 0 then
possibilities(board(i,j))=1
end
end
end
P = possibilities
// we want to see the real values of the possibilities. not just 1 and 0
for i=1 : 9 // sudoku has 9 values
if P(i)==0 then
P(i) = i
else P(i) = 0
end
end
endfunction
function [x,y]=firstEmptyValue(board) // Checks the first empty square of the sudoku
R=%T // and returns the position (x,y)
for i=1 : 9
for j=1 : 9
if board(i,j) == 0 & R = %T then
x=i
y=j
R=%F
end
end
end
endfunction
function A=numberOfPossibilities(V) // this checks the number of possible values for a position
A=0 // so basically it returns the number of elements different from 0 in the vector V
for i=1 : 9
if V(i)>0 then
A=A+1
end
end
endfunction
function u=getUniquePossibility(M,x,y) // this returns the first possible value for that square
pos = [] // in function fillInValue we only use it
pos = PossibilitiesPosition(M,x,y) // when we know that this square (x,y) has only one possible value
for n=1 : 9
if pos(n)>0 then
u=pos(n)
end
end
endfunction
///////////////////////////////////////////////////////////////////////////
////////////////////////// Solve Sudoku ///////////////////////////////
///////////////////////////////////////////////////////////////////////////
function G=fillInValue(M) // fills in a square that has only 1 possibile value
x=0
y=0
pos = []
for i=1 : 9
for j=1 : 9
if M(i,j)==0 then
if numberOfPossibilities(PossibilitiesPosition(M,i,j)) == 1 then
x=i
y=j
break
end
end
end
if x>0 then
break
end
end
M(x,y)=getUniquePossibility(M,x,y)
G=M
endfunction
function H=solve(M) // repeats the fillInValue until it is a fully solved sudoku
P=[]
P=M
if check(M)=%F then
P=fillInValue(M)
H=solve(P)
else
H=M
end
endfunction
//////////////////////////////////////////////////////////////////////////////
So it solves this first one
// Very easy and easy sudokus from brainbashers.com get solved completely
// Very Easy sudoku from brainbashers.com
M = [0 2 0 0 0 0 0 4 0
7 0 4 0 0 0 8 0 2
0 5 8 4 0 7 1 3 0
0 0 1 2 8 4 9 0 0
0 0 0 7 0 5 0 0 0
0 0 7 9 3 6 5 0 0
0 8 9 5 0 2 4 6 0
4 0 2 0 0 0 3 0 9
0 1 0 0 0 0 0 8 0]
But it doens't solve this medium:
M2= [0 0 6 8 7 1 2 0 0
0 0 0 0 0 0 0 0 0
5 0 1 3 0 9 7 0 8
1 0 7 0 0 0 6 0 9
2 0 0 0 0 0 0 0 7
9 0 3 0 0 0 8 0 1
3 0 5 9 0 7 4 0 2
0 0 0 0 0 0 0 0 0
0 0 2 4 3 5 1 0 0]
Error code when trying to solve medium sudoku:
-->solve(M2)
!--error 21
Invalid index.
at line 14 of function PossibilitiesPosition called by :
at line 3 of function getUniquePossibility called by :
at line 20 of function fillInValue called by :
at line 182 of function solve called by :
at line 183 of function solve called by :
at line 183 of function solve called by :
at line 183 of function solve called by :
at line 183 of function solve called by :
solve(M2)
at line 208 of exec file called by :
_SCILAB-6548660277741359031.sce', 1
while executing a callback

Well, one of the easiest way to program a Sudoku solver (not the most efficient) could be to solve each cell with all the possible options recursively (which could be similar to the "Backtracking" algorithm) until a full answer is found.
Another options (I would say it's better) is to iterate trough all the squares solving all the "simple" squares and storing the possible answers in the others squares, then repeat (now you have some more solved), repeat the process until the Sudoku is solved or no more squares can be solved directly. Then you could try the rest with brute-force or Backtracking (maybe half or more of the Sudoku is already solved, so it may be relatively efficient)
Anyway,with a quick search I found this Wikipedia page where some Sudoku solving algorithms are explained with pseudo-code examples, hopefully these will be useful to you

Related

Quick way of finding complementary vectors in MATLAB

I have a matrix of N rows of binary vectors, i.e.
mymatrix = [ 1 0 0 1 0;
1 1 0 0 1;
0 1 1 0 1;
0 1 0 0 1;
0 0 1 0 0;
0 0 1 1 0;
.... ]
where I'd like to find the combinations of rows that, when added together, gets me exactly:
[1 1 1 1 1]
So in the above example, the combinations that would work are 1/3, 1/4/5, and 2/6.
The code I have for this right now is:
i = 1;
for j = 1:5
C = combnk([1:N],j); % Get every possible combination of rows
for c = 1:size(C,1)
if isequal(ones(1,5),sum(mymatrix(C(c,:),:)))
combis{i} = C(c,:);
i = i+1;
end
end
end
But as you would imagine, this takes a while, especially because of that combnk in there.
What might be a useful algorithm/function that can help me speed this up?
M = [
1 0 0 1 0;
1 1 0 0 1;
0 1 1 0 1;
0 1 0 0 1;
0 0 1 0 0;
0 0 1 1 0;
1 1 1 1 1
];
% Find all the unique combinations of rows...
S = (dec2bin(1:2^size(M,1)-1) == '1');
% Find the matching combinations...
matches = cell(0,1);
for i = 1:size(S,1)
S_curr = S(i,:);
rows = M(S_curr,:);
rows_sum = sum(rows,1);
if (all(rows_sum == 1))
matches = [matches; {find(S_curr)}];
end
end
To display your matches in a good stylized way:
for i = 1:numel(matches)
match = matches{i};
if (numel(match) == 1)
disp(['Match found for row: ' mat2str(match) '.']);
else
disp(['Match found for rows: ' mat2str(match) '.']);
end
end
This will produce:
Match found for row: 7.
Match found for rows: [2 6].
Match found for rows: [1 4 5].
Match found for rows: [1 3].
In terms of efficiency, in my machine this algoritm is completing the detection of matches in about 2 milliseconds.

Count the frequency of matrix values including 0

I have a vector
A = [ 1 1 1 2 2 3 6 8 9 9 ]
I would like to write a loop that counts the frequencies of values in my vector within a range I choose, this would include values that have 0 frequencies
For example, if I chose the range of 1:9 my results would be
3 2 1 0 0 1 0 1 2
If I picked 1:11 the result would be
3 2 1 0 0 1 0 1 2 0 0
Is this possible? Also ideally I would have to do this for giant matrices and vectors, so the fasted way to calculate this would be appreciated.
Here's an alternative suggestion to histcounts, which appears to be ~8x faster on Matlab 2015b:
A = [ 1 1 1 2 2 3 6 8 9 9 ];
maxRange = 11;
N = accumarray(A(:), 1, [maxRange,1])';
N =
3 2 1 0 0 1 0 1 2 0 0
Comparing the speed:
K>> tic; for i = 1:100000, N1 = accumarray(A(:), 1, [maxRange,1])'; end; toc;
Elapsed time is 0.537597 seconds.
K>> tic; for i = 1:100000, N2 = histcounts(A,1:maxRange+1); end; toc;
Elapsed time is 4.333394 seconds.
K>> isequal(N1, N2)
ans =
1
As per the loop request, here's a looped version, which should not be too slow since the latest engine overhaul:
A = [ 1 1 1 2 2 3 6 8 9 9 ];
maxRange = 11; %// your range
output = zeros(1,maxRange); %// initialise output
for ii = 1:maxRange
tmp = A==ii; %// temporary storage
output(ii) = sum(tmp(:)); %// find the number of occurences
end
which would result in
output =
3 2 1 0 0 1 0 1 2 0 0
Faster and not-looping would be #beaker's suggestion to use histcounts:
[N,edges] = histcounts(A,1:maxRange+1);
N =
3 2 1 0 0 1 0 1 2 0
where the +1 makes sure the last entry is included as well.
Assuming the input A to be a sorted array and the range starts from 1 and goes until some value greater than or equal to the largest element in A, here's an approach using diff and find -
%// Inputs
A = [2 4 4 4 8 9 11 11 11 12]; %// Modified for variety
maxN = 13;
idx = [0 find(diff(A)>0) numel(A)]+1;
out = zeros(1,maxN); %// OR for better performance : out(maxN) = 0;
out(A(idx(1:end-1))) = diff(idx);
Output -
out =
0 1 0 3 0 0 0 1 1 0 3 1 0
This can be done very easily with bsxfun.
Let the data be
A = [ 1 1 1 2 2 3 6 8 9 9 ]; %// data
B = 1:9; %// possible values
Then
result = sum(bsxfun(#eq, A(:), B(:).'), 1);
gives
result =
3 2 1 0 0 1 0 1 2

Efficiently unpack a vector into binary matrix Octave

On Octave I'm trying to unpack a vector in the format:
y = [ 1
2
4
1
3 ]
I want to return a matrix of dimension ( rows(y) x max value(y) ), where for each row I have a 1 in the column of the original digits value, and a zero everywhere else, i.e. for the example above
y01 = [ 1 0 0 0
0 1 0 0
0 0 0 1
1 0 0 0
0 0 1 0 ]
so far I have
y01 = zeros( m, num_labels );
for i = 1:m
for j = 1:num_labels
y01(i,j) = (y(i) == j);
end
end
which works, but is going get slow for bigger matrices, and seems inefficient because it is cycling through every single value even though the majority aren't changing.
I found this for R on another thread:
f3 <- function(vec) {
U <- sort(unique(vec))
M <- matrix(0, nrow = length(vec),
ncol = length(U),
dimnames = list(NULL, U))
M[cbind(seq_len(length(vec)), match(vec, U))] <- 1L
M
}
but I don't know R and I'm not sure if/how the solution ports to octave.
Thanks for any suggestions!
Use a sparse matrix (which also saves a lot of memory) which can be used in further calculations as usual:
y = [1; 2; 4; 1; 3]
y01 = sparse (1:rows (y), y, 1)
if you really want a full matrix then use "full":
full (y01)
ans =
1 0 0 0
0 1 0 0
0 0 0 1
1 0 0 0
0 0 1 0
Sparse is a more efficient way to do this when the matrix is big.
If your dimension of the result is not very high, you can try this:
y = [1; 2; 4; 1; 3]
I = eye(max(y));
y01 = I(y,:)
The result is same as full(sparse(...)).
y01 =
1 0 0 0
0 1 0 0
0 0 0 1
1 0 0 0
0 0 1 0
% Vector y to Matrix Y
Y = zeros(m, num_labels);
% Loop through each row
for i = 1:m
% Use the value of y as an index; set the value matching index to 1
Y(i,y(i)) = 1;
end
Another possibility is:
y = [1; 2; 4; 1; 3]
classes = unique(y)(:)
num_labels = length(classes)
y01=[1:num_labels] == y
With the following detailed printout:
y =
1
2
4
1
3
classes =
1
2
3
4
num_labels = 4
y01 =
1 0 0 0
0 1 0 0
0 0 0 1
1 0 0 0
0 0 1 0

Count the number of rows between each instance of a value in a matrix

Assume the following matrix:
myMatrix = [
1 0 1
1 0 0
1 1 1
1 1 1
0 1 1
0 0 0
0 0 0
0 1 0
1 0 0
0 0 0
0 0 0
0 0 1
0 0 1
0 0 1
];
Given the above (and treating each column independently), I'm trying to create a matrix that will contain the number of rows since the last value of 1 has "shown up". For example, in the first column, the first four values would become 0 since there are 0 rows between each of those rows and the previous value of 1.
Row 5 would become 1, row 6 = 2, row 7 = 3, row 8 = 4. Since row 9 contains a 1, it would become 0 and the count starts again with row 10. The final matrix should look like this:
FinalMatrix = [
0 1 0
0 2 1
0 0 0
0 0 0
1 0 0
2 1 1
3 2 2
4 0 3
0 1 4
1 2 5
2 3 6
3 4 0
4 5 0
5 6 0
];
What is a good way of accomplishing something like this?
EDIT: I'm currently using the following code:
[numRow,numCol] = size(myMatrix);
oneColumn = 1:numRow;
FinalMatrix = repmat(oneColumn',1,numCol);
toSubtract = zeros(numRow,numCol);
for m=1:numCol
rowsWithOnes = find(myMatrix(:,m));
for mm=1:length(rowsWithOnes);
toSubtract(rowsWithOnes(mm):end,m) = rowsWithOnes(mm);
end
end
FinalMatrix = FinalMatrix - toSubtract;
which runs about 5 times faster than the bsxfun solution posted over many trials and data sets (which are about 1500 x 2500 in size). Can the code above be optimized?
For a single column you could do this:
col = 1; %// desired column
vals = bsxfun(#minus, 1:size(myMatrix,1), find(myMatrix(:,col)));
vals(vals<0) = inf;
result = min(vals, [], 1).';
Result for first column:
result =
0
0
0
0
1
2
3
4
0
1
2
3
4
5
find + diff + cumsum based approach -
offset_array = zeros(size(myMatrix));
for k1 = 1:size(myMatrix,2)
a = myMatrix(:,k1);
widths = diff(find(diff([1 ; a])~=0));
idx = find(diff(a)==1)+1;
offset_array(idx(idx<=numel(a)),k1) = widths(1:2:end);
end
FinalMatrix1 = cumsum(double(myMatrix==0) - offset_array);
Benchmarking
The benchmarking code for comparing the above mentioned approach against the one in the question is listed here -
clear all
myMatrix = round(rand(1500,2500)); %// create random input array
for k = 1:50000
tic(); elapsed = toc(); %// Warm up tic/toc
end
disp('------------- With FIND+DIFF+CUMSUM based approach') %//'#
tic
offset_array = zeros(size(myMatrix));
for k1 = 1:size(myMatrix,2)
a = myMatrix(:,k1);
widths = diff(find(diff([1 ; a])~=0));
idx = find(diff(a)==1)+1;
offset_array(idx(idx<=numel(a)),k1) = widths(1:2:end);
end
FinalMatrix1 = cumsum(double(myMatrix==0) - offset_array);
toc
clear FinalMatrix1 offset_array idx widths a
disp('------------- With original approach') %//'#
tic
[numRow,numCol] = size(myMatrix);
oneColumn = 1:numRow;
FinalMatrix = repmat(oneColumn',1,numCol); %//'#
toSubtract = zeros(numRow,numCol);
for m=1:numCol
rowsWithOnes = find(myMatrix(:,m));
for mm=1:length(rowsWithOnes);
toSubtract(rowsWithOnes(mm):end,m) = rowsWithOnes(mm);
end
end
FinalMatrix = FinalMatrix - toSubtract;
toc
The results I got were -
------------- With FIND+DIFF+CUMSUM based approach
Elapsed time is 0.311115 seconds.
------------- With original approach
Elapsed time is 7.587798 seconds.

How to make this algorithm of pattern finding?

I have a matrix and I need to find a pattern inside this matrix.
Matrix is:
1 0 0 1 1 1 0 0 0 1
0 0 0 1 1 0 1 0 0 1
0 1 1 1 0 0 0 1 0 1
1 0 1 0 0 1 1 0 1 0
1 1 1 0 0 0 1 1 0 1
0 1 0 0 1 1 0 1 0 1
1 1 1 0 0 0 1 0 0 1
1 0 0 1 0 1 1 1 0 1
Rules:
We choose one number from every row.
The next choosen number from second row must be an opposite of the precedent.
Positions of the numbers choosed by the 1 and 2 rules, must be a precise pattern.
So the question would be:
Find the best pattern that respect the 3 rules.
Example from the matrix shown:
Choosed a number: 0(2) //what is in "()" represents the position of the value..position start from 1 to 10 on rows.
1(4)
For the positions 2 and 4 to be a pattern must support rules 1 and 2 for the rest of the matrix.
So we go further on the 3rd row and we check 2nd position:1. We go 4th row, we check 4th position:0. Seems to respect the rules. There are opposite numbers on 2nd and 4th position, so we continue: 5th row, 2nd position:, and so on, but you will see on 7th row 2nd position:1 and 8th row 4th position:1; so the pattern of positions 2-4 is not good.
How could I make an algorithm based on these rules?
Maybe this will help (motivated by the comment to your question). This is a C++ sort of answer. This answer assumes 0 is always the number you pick, but you can easily edit this to allow 1 to be first.
int firstPos, secondPos;
for(int i = 0; i < 10; ++i)
if(matrix[0][i] == 0)
firstPos = i;
for(int i = 0; i < 10; ++i)
if(matrix[0][i] == 1)
secondPos= i;
bool success = true;
for(int i = 0; i < 10/2; ++i)
if(matrix[2*i][firstPos] == matrix[2*i][secondPos])
success == false;
if(success)
cout << "success" << endl;
else
cout << "failure" << endl;
I would define a pattern by index of the first item (F) and index of the second item (S). I'll also assume that indices begin with 0 (instead of 1 as in your example). Both F and S can take a value between 0 and 9. Solution is simple. Have a double nested loop that runs F and S from 0 to 9, and in third innermost loop just verify that current F and S form a pattern.

Resources