How to convert triangular matrix indexes in to row, column coordinates? - c++

I have these indexes:
1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,etc...
Which are indexes of nodes in a matrix (including diagonal elements):
1
2 3
4 5 6
7 8 9 10
11 12 13 14 15
16 17 18 19 20 21
etc...
and I need to get i,j coordinates from these indexes:
1,1
2,1 2,2
3,1 3,2 3,3
4,1 4,2 4,3 4,4
5,1 5,2 5,3 5,4 5,5
6,1 6,2 6,3 6,4 6,5 6,6
etc...
When I need to calculate coordinates I have only one index and cannot access others.

Not optimized at all :
int j = idx;
int i = 1;
while(j > i) {
j -= i++;
}
Optimized :
int i = std::ceil(std::sqrt(2 * idx + 0.25) - 0.5);
int j = idx - (i-1) * i / 2;
And here is the demonstration:
You're looking for i such that :
sumRange(1, i-1) < idx && idx <= sumRange(1, i)
when sumRange(min, max) sum integers between min and max, both inxluded.
But since you know that :
sumRange(1, i) = i * (i + 1) / 2
Then you have :
idx <= i * (i+1) / 2
=> 2 * idx <= i * (i+1)
=> 2 * idx <= i² + i + 1/4 - 1/4
=> 2 * idx + 1/4 <= (i + 1/2)²
=> sqrt(2 * idx + 1/4) - 1/2 <= i

In my case (a CUDA kernel implemented in standard C), I use zero-based indexing (and I want to exclude the diagonal) so I needed to make a few adjustments:
// idx is still one-based
unsigned long int idx = blockIdx.x * blockDim.x + threadIdx.x + 1; // CUDA kernel launch parameters
// but the coordinates are now zero-based
unsigned long int x = ceil(sqrt((2.0 * idx) + 0.25) - 0.5);
unsigned long int y = idx - (x - 1) * x / 2 - 1;
Which results in:
[0]: (1, 0)
[1]: (2, 0)
[2]: (2, 1)
[3]: (3, 0)
[4]: (3, 1)
[5]: (3, 2)
I also re-derived the formula of Flórez-Rueda y Moreno 2001 and arrived at:
unsigned long int x = floor(sqrt(2.0 * pos + 0.25) + 0.5);
CUDA Note: I tried everything I could think of to avoid using double-precision math, but the single-precision sqrt function in CUDA is simply not precise enough to convert positions greater than 121 million or so to x, y coordinates (when using 1,024 threads per block and indexing only along 1 block dimension). Some articles have employed a "correction" to bump the result in a particular direction, but this inevitably falls apart at a certain point.

Related

Matrix with 1d vector

I represent a n*m matrix like chessboard.
1 0 2 0
0 3 0 4
5 0 6 0
0 7 0 8
I don't need to store the zeros in my 1d vector.
vector v = {1, 2, 3, 4.. etc}
I ask the user for a row and column number.
How can i return with i. row j. column element?
if (i+j) % 2 != 0
I return with 0, but i don't know what i need to do when
(i+j) % 2 == 0
Can you help me? (sorry for my bad English)
With regular matrices stored as 1D-vector, coordinate to index would be:
(i + j * width) (or i * height + j depending on convention).
with half case to 0, you just have to divide by 2:
if ((i + j) % 2 != 0) return 0;
else return data[(i + j * width) / 2];

Decreasing Loop Interval by 1 in C/C++

Let's say I have 15 elements. I want to group them such a way that:
group1 = 1 - 5
group2 = 6 - 9
group3 = 10 - 12
group4 = 13 - 14
group5 = 15
This way I'll get elements in each group as below:
group1 = 5
group2 = 4
group3 = 3
group4 = 2
group5 = 1
As you can see loop interval is decreasing.
I took 15 just for an example. In actual programme it's user driven parameter which can be anything (hopefully few thousand).
Now what I'm looking for is:
Whatever is in group1 should have variable "loop" value 0, group2 should have 1, group3 should have 2 and so on... "loop" is an int variable which is being used to calculate some other stuff.
Let's put in other words too
I have an int variable called "loop". I want to assign value to it such a way that:
First n frames loop value 0 next (n -1) frames loop value 1 then next (n - 2) frames loop value 2 all the way to loop value (n - 1)
Let's say I have 15 frames on my timeline.
So n will be 5 ====>>>>> (5 + 4 + 3 + 2 + 1 = 15; as interval is decreasing by 1)
then
first 5 frames(1 - 5) loop is 0 then next 4 frames(6 - 9) loop is 1 then next 3 frames(10 - 12) loop is 2 then next 2 frames(13 - 14) loop is 3 and for last frame(15) loop is 4.
frames "loop" value
1 - 5 => 0
6 - 9 => 1
10 - 12 => 2
13 - 14 => 3
15 => 4
I've tried with modulo(%). But the issue is on frame 12 loop is 2 so (12 % (5 - 2)) remainder is 0 so it increments loop value.
The following lines are sample code which is running inside a solver. #loop is by default 0 and #Frame is current processing frame number.
int loopint = 5 - #loop;
if (#Frame % loopint == 0)
#loop += 1;
If I understand this correctly, then
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
int main(int argc, char *argv[]) {
int n = atoi(argv[1]);
for(int i = 1; i <= n; ++i) {
printf("%d: %f\n", i, ceil((sqrt(8 * (n - i + 1) + 1) - 1) / 2));
}
}
is an implementation in C.
The math behind this is as follows: The 1 + 2 + 3 + 4 + 5 you have there is a Gauß sum, which has a closed form S = n * (n + 1) / 2 for n terms. Solving this for n, we get
n = (sqrt(8 * S + 1) - 1) / 2
Rounding this upward would give us the solution if you wanted the short stretches at the beginning, that is to say 1, 2, 2, 3, 3, 3, ...
Since you want the stretches to become progressively shorter, we have to invert the order, so S becomes (n - S + 1). Therefore the formula up there.
EDIT: Note that unless the number of elements in your data set fits the n * (n+1) / 2 pattern precisely, you will have shorter stretches either at the beginning or in the end. This implementation places the irregular stretch at the beginning. If you want them at the end,
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
int main(int argc, char *argv[]) {
int n = atoi(argv[1]);
int n2 = (int) ceil((sqrt(8 * n + 1) - 1) / 2);
int upper = n2 * (n2 + 1) / 2;
for(int i = 1; i <= n; ++i) {
printf("%d: %f\n", i, n2 - ceil((sqrt(8 * (upper - i + 1) + 1) - 1) / 2));
}
}
does it. This calculates the next such number beyond your element count, then calculates the numbers you would have if you had that many elements.

Linear index upper triangular matrix

If I have the upper triangular portion of a matrix, offset above the diagonal, stored as a linear array, how can the (i,j) indices of a matrix element be extracted from the linear index of the array?
For example, the linear array [a0, a1, a2, a3, a4, a5, a6, a7, a8, a9 is storage for the matrix
0 a0 a1 a2 a3
0 0 a4 a5 a6
0 0 0 a7 a8
0 0 0 0 a9
0 0 0 0 0
And we want to know the (i,j) index in the array corresponding to an offset in the linear matrix, without recursion.
A suitable result, k2ij(int k, int n) -> (int, int) would satisfy, for example
k2ij(k=0, n=5) = (0, 1)
k2ij(k=1, n=5) = (0, 2)
k2ij(k=2, n=5) = (0, 3)
k2ij(k=3, n=5) = (0, 4)
k2ij(k=4, n=5) = (1, 2)
k2ij(k=5, n=5) = (1, 3)
[etc]
The equations going from linear index to (i,j) index are
i = n - 2 - floor(sqrt(-8*k + 4*n*(n-1)-7)/2.0 - 0.5)
j = k + i + 1 - n*(n-1)/2 + (n-i)*((n-i)-1)/2
The inverse operation, from (i,j) index to linear index is
k = (n*(n-1)/2) - (n-i)*((n-i)-1)/2 + j - i - 1
Verify in Python with:
from numpy import triu_indices, sqrt
n = 10
for k in range(n*(n-1)/2):
i = n - 2 - int(sqrt(-8*k + 4*n*(n-1)-7)/2.0 - 0.5)
j = k + i + 1 - n*(n-1)/2 + (n-i)*((n-i)-1)/2
assert np.triu_indices(n, k=1)[0][k] == i
assert np.triu_indices(n, k=1)[1][k] == j
for i in range(n):
for j in range(i+1, n):
k = (n*(n-1)/2) - (n-i)*((n-i)-1)/2 + j - i - 1
assert triu_indices(n, k=1)[0][k] == i
assert triu_indices(n, k=1)[1][k] == j
First, let's renumber a[k] in opposite order. We'll get:
0 a9 a8 a7 a6
0 0 a5 a4 a3
0 0 0 a2 a1
0 0 0 0 a0
0 0 0 0 0
Then k2ij(k, n) will become k2ij(n - k, n).
Now, the question is, how to calculate k2ij(k, n) in this new matrix. The sequence 0, 2, 5, 9 (indices of diagonal elements) corresponds to triangular numbers (after subtracting 1): a[n - i, n + 1 - i] = Ti - 1. Ti = i * (i + 1)/2, so if we know Ti, it's easy to solve this equation and get i (see formula in the linked wiki article, section "Triangular roots and tests for triangular numbers"). If k + 1 is not exactly a triangular number, the formula will still give you the useful result: after rounding it down, you'll get the highest value of i, for which Ti <= k, this value of i corresponds to the row index (counting from bottom), in which a[k] is located. To get the column (counting from right), you should simply calculate the value of Ti and subtract it: j = k + 1 - Ti. To be clear, these are not exacly i and j from your problem, you need to "flip" them.
I didn't write the exact formula, but I hope that you got the idea, and it will now be trivial to find it after performing some boring but simple calculations.
The following is an implimentation in matlab, which can be easily transferred to another language, like C++. Here, we suppose the matrix has size m*m, ind is the index in the linear array. The only thing different is that here, we count the lower triangular part of the matrix column by column, which is analogus to your case (counting the upper triangular part row by row).
function z= ind2lTra (ind, m)
rvLinear = (m*(m-1))/2-ind;
k = floor( (sqrt(1+8*rvLinear)-1)/2 );
j= rvLinear - k*(k+1)/2;
z=[m-j, m-(k+1)];
For the records, this is the same function, but with one-based indexing, and in Julia:
function iuppert(k::Integer,n::Integer)
i = n - 1 - floor(Int,sqrt(-8*k + 4*n*(n-1) + 1)/2 - 0.5)
j = k + i + ( (n-i+1)*(n-i) - n*(n-1) )÷2
return i, j
end
Here is a more efficient formulation for k:
k = (2 * n - 3 - i) * i / 2 + j - 1
In python 2:
def k2ij(k, n):
rows = 0
for t, cols in enumerate(xrange(n - 1, -1, -1)):
rows += cols
if k in xrange(rows):
return (t, n - (rows - k))
return None
In python, the most efficient way is:
array_size= 3
# make indices using k argument if you want above the diagonal
u, v = np.triu_indices(n=array_size,k=1)
# assuming linear indices above the diagonal i.e. 0 means (0,1) and not (0,0)
linear_indices = [0,1]
ijs = [(i,j) for (i,j) in zip(u[linear_indices], v[linear_indices])]
ijs
#[(0, 1), (0, 2)]

Finding Neighbourhood in Matrices [duplicate]

This question already has an answer here:
Conway's Game of Life, counting neighbors [closed]
(1 answer)
Closed 9 years ago.
I am working on project containing cellular automat methods. What I am trying to figure is how to write function helping to find all the neighbours in a 2d array.
for example i ve got size x size 2d array [size = 4 here]
[x][n][ ][n]
[n][n][ ][n]
[ ][ ][ ][ ]
[n][n][ ][n]
Field marked as x [0,0 index] has neighbours marked as [n] -> 8 neighbours. What Im trying to do is to write a function which can find neighbours wo writting tousands of if statements
Does anybody have an idea how to do it ?
thanks
For the neighbours of element (i,j) in NxM matrix:
int above = (i-1) % N;
int below = (i+1) % N;
int left = (j-1) % M;
int right = (j+1) % M;
decltype(matrix[0][0]) *indices[8];
indices[0] = & matrix[above][left];
indices[1] = & matrix[above][j];
indices[2] = & matrix[above][right];
indices[3] = & matrix[i][left];
// Skip matrix[i][j]
indices[4] = & matrix[i][right];
indices[5] = & matrix[below][left];
indices[6] = & matrix[below][j];
indices[7] = & matrix[below][right];
Suppose you are in cell (i, j). Then, on an infinite grid, your neighbors should be [(i-1, j-1), (i-1,j), (i-1, j+1), (i, j-1), (i, j+1), (i+1, j-1), (i+1, j), (i+1, j+1)].
However, since the grid is finite some of the above values will get outside the bounds. But we know modular arithmetic: 4 % 3 = 1 and -1 % 3 = 2. So, if the grid is of size n, m you only need to apply %n, %m on the above list to get the proper list of neighbors: [((i-1) % n, (j-1) % m), ((i-1) % n,j), ((i-1) % n, (j+1) % m), (i, (j-1) % m), (i, (j+1) % m), ((i+1) % n, (j-1) % m), ((i+1) % n, j), ((i+1) % n, (j+1) % m)]
That works if your coordinates are between 0 and n and between 0 and m. If you start with 1 then you need to tweak the above by doing a -1 and a +1 somewhere.
For your case n=m=4 and (i, j) = (0, 0). The first list is [(-1, -1), (-1, 0), (-1, 1), (0, -1), (0, 1), (1, -1), (1, 0), (1, 1)]. Applying the modulus operations you get to [(3, 3), (3, 0), (3, 1), (0, 3), (0, 1), (1, 3), (1, 0), (1, 1)] which are exactly the squares marked [n] in your picture.
Add and subtract one from the coordinates, in all possible permutations. Results outside the boundaries wrap around (e.g. -1 becomes 3 and 4 becomes 0). Just a couple of simple loops needed basically.
Something like
// Find the closest neighbours (one step) from the coordinates [x,y]
// The max coordinates is max_x,max_y
// Note: Does not contain any error checking (for valid coordinates)
std::vector<std::pair<int, int>> getNeighbours(int x, int y, int max_x, int max_y)
{
std::vector<std::pair<int, int>> neighbours;
for (int dx = -1; dx <= 1; ++dx)
{
for (int dy = -1; dy <= 1; ++dy)
{
// Skip the coordinates [x,y]
if (dx == 0 && dy == 0)
continue;
int nx = x + dx;
int ny = y + dy;
// If the new coordinates goes out of bounds, wrap them around
if (nx < 0)
nx = max_x;
else if (nx > max_x)
nx = 0;
if (ny < 0)
ny = max_y;
else if (ny > max_y)
ny = 0;
// Add neighbouring coordinates to result
neighbours.push_back(std::make_pair(nx, ny));
}
}
return neighbours;
}
Example use for you:
auto n = getNeighbours(0, 0, 3, 3);
for (const auto& p : n)
std::cout << '[' << p.first << ',' << p.second << "]\n";
Prints out
[3,3]
[3,0]
[3,1]
[0,3]
[0,1]
[1,3]
[1,0]
[1,1]
which is the correct answer.

Sum of submatrices of bigger matrix

I have a big matrix as input, and I have the size of a smaller matrix. I have to compute the sum of all possible smaller matrices which can be formed out of the bigger matrix.
Example.
Input matrix size: 4 × 4
Matrix:
1 2 3 4
5 6 7 8
9 9 0 0
0 0 9 9
Input smaller matrix size: 3 × 3 (not necessarily a square)
Smaller matrices possible:
1 2 3
5 6 7
9 9 0
5 6 7
9 9 0
0 0 9
2 3 4
6 7 8
9 0 0
6 7 8
9 0 0
0 9 9
Their sum, final output
14 18 22
29 22 15
18 18 18
I did this:
int** matrix_sum(int **M, int n, int r, int c)
{
int **res = new int*[r];
for(int i=0 ; i<r ; i++) {
res[i] = new int[c];
memset(res[i], 0, sizeof(int)*c);
}
for(int i=0 ; i<=n-r ; i++)
for(int j=0 ; j<=n-c ; j++)
for(int k=i ; k<i+r ; k++)
for(int l=j ; l<j+c ; l++)
res[k-i][l-j] += M[k][l];
return res;
}
I guess this is too slow, can anyone please suggest a faster way?
Your current algorithm is O((m - p) * (n - q) * p * q). The worst case is when p = m / 2 and q = n / 2.
The algorithm I'm going to describe will be O(m * n + p * q), which will be O(m * n) regardless of p and q.
The algorithm consists of 2 steps.
Let the input matrix A's size be m x n and the size of the window matrix being p x q.
First, you will create a precomputed matrix B of the same size as the input matrix. Each element of the precomputed matrix B contains the sum of all the elements in the sub-matrix, whose top-left element is at coordinate (1, 1) of the original matrix, and the bottom-right element is at the same coordinate as the element that we are computing.
B[i, j] = Sum[k = 1..i, l = 1..j]( A[k, l] ) for all 1 <= i <= m, 1 <= j <= n
This can be done in O(m * n), by using this relation to compute each element in O(1):
B[i, j] = B[i - 1, j] + Sum[k = 1..j-1]( A[i, k] ) + A[j] for all 2 <= i <= m, 1 <= j <= n
B[i - 1, j], which is everything of the sub-matrix we are computing except the current row, has been computed previously. You keep a prefix sum of the current row, so that you can use it to quickly compute the sum of the current row.
This is another way to compute B[i, j] in O(1), using the property of the 2D prefix sum:
B[i, j] = B[i - 1, j] + B[i, j - 1] - B[i - 1, j - 1] + A[j] for all 1 <= i <= m, 1 <= j <= n and invalid entry = 0
Then, the second step is to compute the result matrix S whose size is p x q. If you make some observation, S[i, j] is the sum of all elements in the matrix size (m - p + 1) * (n - q + 1), whose top-left coordinate is (i, j) and bottom-right is (i + m - p + 1, j + n - q + 1).
Using the precomputed matrix B, you can compute the sum of any sub-matrix in O(1). Apply this to compute the result matrix S:
SubMatrixSum(top-left = (x1, y1), bottom-right = (x2, y2))
= B[x2, y2] - B[x1 - 1, y2] - B[x2, y1 - 1] + B[x1 - 1, y1 - 1]
Therefore, the complexity of the second step will be O(p * q).
The final complexity is as mentioned above, O(m * n), since p <= m and q <= n.