more-dimensional arrays

program dim

integer matrix (20,30)

=

int main(){

int matrix[20][30];

?

Let’s say it’s very similar when you use it. However, conceptually this is different. The C++ code declares “a vector of length 20 of pointers to a vector of int of length 30”. So “matrix” is a pointer to a pointer to int. There is no length/dimension checking done anywhere.

In Fortran, you can only get a “rectangular blob” of memory, that is, the above will always have 20*30 elements. In C++ you can (dynamically) construct arrays with “ragged” boundaries, where each row has a different number of columns.

int main(){

int **matrix;

matrix = new int *[20]; // a vector of length 20 of pointers to int

for (int j=0, jង j++) *matrix[j] = new int [30];

// do something… this is the same as matrix[20][30];

for (int j=0, jង j++) delete [] *matrix[j];

delete [] matrix;

}

int main(){

int **matrix;

matrix = new int *[20]; // a vector of length 20 of pointers to int

for (int j=0, jង j++) matrix[j] = new int [j+10];

// do something… this is the same as matrix[20][30];

for (int j=0, jង j++) delete [] matrix[j];

delete [] matrix;

}

Previous slide

Next slide

Back to first slide

View graphic version