我遇到了一些奇怪的副作用,我无法用丝毫解释。可能我错过了一些非常明显的东西,但我现在已经找了几个小时的bug,而且代码非常简单,所以我得出结论,我必须对......有一些相当根本的误解。
考虑这个代码,用于计算两个2d矩阵的乘积(我已经更改了set()
函数以将-1
添加到参数单元格,以使调试输出更易于理解。 / p>
template<class T, unsigned column_count, unsigned row_count>
class Matrix
{
private:
const static unsigned row_length = column_count;
const static unsigned column_length = row_count;
using matrix_type = std::array<std::array<T, row_length>, row_count>;
matrix_type matrix;
public:
using value_type = T;
Matrix(const matrix_type& matrix) : matrix(matrix) {}
Matrix() {}
friend std::ostream& operator<<(std::ostream& o, const Matrix& rhs)
{
for (unsigned i = 0; i < column_count; ++i) {
for (unsigned j = 0; j < row_count; ++j) {
o << rhs.matrix[i][j] << ' ';
}
o << '\n';
}
return o;
}
const auto& get_rows() const { return matrix; }
const auto get_columns() const
{
std::array<std::array<T, column_length>, column_count> columns;
for (unsigned i = 0; i < row_length; ++i) {
for (unsigned j = 0; j < column_length; ++j) {
columns[i][j] = matrix[j][i];
}
}
return columns;
}
void set(unsigned i, unsigned j, T v) { matrix[i][j] = -1; }
friend Matrix operator*(const Matrix& m1, const Matrix& m2)
{
auto columns = m1.get_columns();
auto rows = m2.get_rows();
Matrix m3;
std::cout << "before:"
<< "\n";
std::cout << m1 << "\n";
std::cout << m2 << "\n";
std::cout << m3 << "\n";
unsigned i{ 0 };
for (const auto& row : rows) {
i++;
unsigned j{ 0 };
for (const auto& column : columns) {
j++;
value_type v{ 0 };
for (unsigned k = 0; k < column.size(); ++k) {
v += row[k] * column[k];
}
m3.set(i, j, v);
}
}
std::cout << "after:"
<< "\n";
std::cout << m1 << "\n";
std::cout << m2 << "\n";
std::cout << m3 << "\n";
return m3;
}
};
如您所见,getter函数要么返回副本,要么返回常量引用。 operator*
函数采用常量参数。
我现在构建两个矩阵,如下:
std::array<int, 3> c1{ { 1, 2, 3 } };
std::array<int, 3> c2{ { 4, 5, 6 } };
std::array<int, 3> c3{ { 7, 8, 9 } };
std::array<std::array<int, 3>, 3> m1{ { c1, c2, c3 } };
std::array<std::array<int, 3>, 3> m2 = m1;
Matrix<int, 3, 3> matrix1(m1);
Matrix<int, 3, 3> matrix2(m2);
现在我以不同的方式调用operator*
:
matrix1* matrix2;
结果:
before:
1 2 3
4 5 6
7 8 9
1 2 3
4 5 6
7 8 9
0 0 0
0 0 0
0 0 183238709
after:
-1 -1 -1
4 5 6
7 8 9
1 2 3
4 5 6
7 8 9
0 0 0
0 -1 -1
-1 -1 -1
matrix2* matrix1;
结果:
before:
1 2 3
4 5 6
7 8 9
1 2 3
4 5 6
7 8 9
0 0 0
0 0 0
0 0 -1823473620
after:
1 2 3
4 5 6
7 8 9
-1 -1 -1
4 5 6
7 8 9
0 0 0
0 -1 -1
-1 -1 -1
matrix1* matrix1;
结果:
before:
1 2 3
4 5 6
7 8 9
1 2 3
4 5 6
7 8 9
1385085408 32767 401515081
1 1385085440 32767
1385085440 32767 1385085464
after:
-1 -1 -1
4 5 6
7 8 9
-1 -1 -1
4 5 6
7 8 9
1385085408 32767 401515081
1 -1 -1
-1 -1 -1
如您所见,作为第一个参数传递的矩阵将被更改。这对我来说毫无意义,因为它作为const传递,set()
仅在m3
上运行。不知怎的m3
得到了部分&#34;绑定&#34;到作为operator*
的第一个参数的矩阵。为什么呢?
答案 0 :(得分:0)
您在循环中写出越界,因为i
和j
都过早递增。
代码应如下所示:
for (const auto& row : rows) {
// i++;
unsigned j{ 0 };
for (const auto& column : columns) {
// j++;
value_type v{ 0 };
for (unsigned k = 0; k < column.size(); ++k) {
v += row[k] * column[k];
}
m3.set(i, j, v);
j++; // <--
}
i++; // <--
}