Error in inner product of vectors and index - c++

Question

I found a bug in my code and can't figuring out the error. I tried debugging by showing the output of each variable step by step but I can't find my error. Here is what I have and what I want to do:
I have a matrix A:
0000
0101
1010
1111
And I have a matrix B:
10000
21000
30100
41100
20010
21010
40110
41110
30001
41001
30101
41101
40011
41011
40111
41111
The matrix B has 16 rows and 5 coloumns. The matrix A has 4 rows and 4 coloumns. Now I declare a matrix C that has 4 rows and 16 coloumns.
What I want to do is to calculate the inner product of each row from B with a corresponding row from A. With corresponding I mean that the first coloumn of B shoud define the row from A that I want to multiply. So the B matrix has in fact also four-dimensional vectors and the first element corresponds to the row of A. One could say this first coloumn of B is an index for choosing the row of A. Because C++ start counting by zero I substract one for my index. Here is my code:
std::vector< std::vector<int> > C(4, std::vector<int>(16));
std::vector<int> index(4);
std::vector<int> vectorA(4);
std::vector<int> vectorB(4);
for( int y = 0; y < 16; y++)
{
for(int i=0; i<4; ++i){
vectorA[i] = A[ B[y][0]-1 ][i];
}
for( int x = 1; x < 4; x++)
{
vectorB[x -1] = B[y][x];
}
C[ B[y][0] -1][index[ B[y][0] -1] ] = inner_product(vectorA.begin(), vectorA.end(), vectorB.begin(), 0);
index[B[y][0]-1] += 1;
}
This results in my matrix C:
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
2 2 3 1 2 1 2 2 3 0 0 0 0 0 0 0
The first two rows are correct but row three and four are false.
The correct solution has to be (maybe except of ordering in row 3 and 4):
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
4 3 3 2 3 2 3 2 2 0 0 0 0 0 0 0
Where is my problem? Please help, it drives me crazy :( I tried showing each variable by step but can't find why is it false.
Thanks and greetings.

Answer 1

I have to agree with the other comments: Your code is kind of confusing. You should really simplify the access of vectors by index.
First simple thing you should do is to change the first column of B to be zero-based. All stuff in C++ is zero-based. Adopt it. Do not start adjusting it in your code by substracting one. (This does not gain much simplicity, but is is symptomatic for your code.)
Another source of confusion is that you use the first column of B as an index into A. This might be an implication from the problem you'd like to solve, but it makes things unclear: first column of B has a totally different meaning, always code in a way that objects are seperated by their meaning.
For me the most confusing thing is, that I really do not get what you're up to. With inner product you mean dot product, right? You have 2 sets of vectors you want to calculate the dot product of. This should result in a set of scalars, a 1D vector not a 2D matrix. You do some special stuff with this index vector, which makes the result being a 2D matrix. But you haven't explained the purpose/system behind it. Why do you need a vector for index, not just a scalar??
Vector index is the most ugly/complex part of your code. Without having a clue what you are up to, I would still guess that you find out what is going wrong when you start printing out the full vector index on every iteration and check if it is changing the way you expect.

Answer 2

I don't know what's the rationale behind OP choices, so I can't properly comment the design of code provided, but for what I can understand there are some mistakes with the example input too.
Given A and B matrices as presented, the inner product of the lower rows of A with the corresponding in B is always 0:
B[1] { 2, 1, 0, 0, 0 },
row "2" or A[1] is { 0, 1, 0, 1 } <- B[4] { 2, 0, 0, 1, 0 },
B[5] { 2, 1, 0, 1, 0 },
The same for the succesive row. Only if swapped, the expected output can be obtained and so I did in my code.
vectorA and vectorB and the corresponding copy loops aren't really neccessary and probably are the cause of the wrong output:
for( int x = 1; x < 4; x++)
{ // ^^^^^ this should be <= to reach the last element
vectorB[x -1] = B[y][x];
}
My code, with the updated input and the direct use of A and B is:
#include <iostream>
#include <vector>
#include <numeric>
using vec_t = std::vector<int>; // I assume a C++11 compliant compiler
using mat_t = std::vector<vec_t>;
using std::cout;
int main() {
mat_t A{
{ 0, 0, 0, 0 },
{ 1, 0, 1, 0 }, // <-- those lines are swapped
{ 0, 1, 0, 1 }, // <--
{ 1, 1, 1, 1 }
};
mat_t B{
{ 1, 0, 0, 0, 0 },
{ 2, 1, 0, 0, 0 },
{ 3, 0, 1, 0, 0 },
{ 4, 1, 1, 0, 0 },
{ 2, 0, 0, 1, 0 },
{ 2, 1, 0, 1, 0 },
{ 4, 0, 1, 1, 0 },
{ 4, 1, 1, 1, 0 },
{ 3, 0, 0, 0, 1 },
{ 4, 1, 0, 0, 1 },
{ 3, 0, 1, 0, 1 },
{ 4, 1, 1, 0, 1 },
{ 4, 0, 0, 1, 1 },
{ 4, 1, 0, 1, 1 },
{ 4, 0, 1, 1, 1 },
{ 4, 1, 1, 1, 1 }
};
mat_t C(4, vec_t(16));
vec_t pos(4);
for ( int i = 0; i < 16; ++i )
{
int row = B[i][0] - 1;
int col = pos[row];
int prod = std::inner_product( A[row].begin(), A[row].end(),
++(B[i].begin()), 0 );
// ^^^ skip the first element
C[row][col] = prod;
if ( prod )
++pos[row];
}
for ( auto & r : C )
{
for ( int x : r ) {
cout << ' ' << x;
}
cout << '\n';
}
return 0;
}
The output is:
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 2 0 0 0 0 0 0 0 0 0 0 0 0 0
2 2 3 2 3 2 3 3 4 0 0 0 0 0 0 0
I don't know if the ordering of the last row is as expected, but it mimics the logic of OP's code.