understanding dynamic matrix declaration

**Elnaz** · 01-14-2011

Dear all,
It's been a long time that I haven't done C and I've forgot almost all about it. I'm trying to catch up with it!
Well, first I need to understand one part i.e. dynamic matrix declaration; specifically, I am trying to understand these lines of code:

Code:

// 0. write Data matrix
	int size_x = 210;
	int size_y = 7;
	double **Data, *pDataData;
	Data = (double **)malloc(size_x*sizeof(double *)+size_x*size_y*sizeof(double));
	for (i = 0, pDataData = (double *)(Data+size_x); i < size_x; i++, pDataData += size_y)
	Data[i]=pDataData;

	for(i=0;i<size_x;i++)
		for(j=0;j<size_y;j++){
			Data[i][j] = *(Data_in+size_y*i+j);
			//printf("%f        ", Data[i][j]);
		}	//printf("\n\n\n\n\n");

Especially, I have problem understanding the bold lines. Could you try to explain it in a very clear way what exactly is going on here. I mean I know (by book) what malloc does and all that I just don't understand the details like what this initialization does:

Code:

pDataData = (double *)(Data+size_x)

and how a pointer to pointer equals a pointer in this line:

Code:

Data[i]=pDataData

Thank you very much in advance
Elnaz

**tabstop** · 01-14-2011

If you're going to talk about bold lines, you should probably make some lines bold.

Anyway, to take the second part first, Data[i] is a pointer, and pDataData is a pointer, consequently any worry about pointers to pointers is irrelevant because you don't have any.

In the first part, Data+size_x is pointer arithmetic: you take the pointer Data, and you move it forward by size_x things-that-it-points-to. Data points to double*, so we move forward size_x * size-of-a-pointer-to-double-on-your-system bytes. That then gets cast to a double* before being assigned to pDataData.

**Elnaz** · 01-14-2011

Thank you, I got your points; the code is supposed to allocate a 210 by 7 matrix. I get that it first allocates a 210 by 7 memory block using malloc but why then it further adds a 210 length array to that block?
It then goes on to write the addresses of each row into Data[i] (I assume) but again why going forward by 210 at first?
I'm kind of not getting the structure here.

**nimitzhunter** · 01-14-2011

What do you in malloc is allocating data for 210 double * and 210*7 double. It looks like the first 210 elements are double *. You need pdatadata to point to the double, not double*, so you add 210 to Data to access the 211st elements on the heap.

**Elnaz** · 01-16-2011

In other words, what I still don't get is that in the 2nd block of code:

Code:

for(i=0;i<size_x;i++)
		for(j=0;j<size_y;j++){
			Data[i][j] = *(Data_in+size_y*i+j);
			}

we're feeding the values to the Data matrix; so, why do we feed Data[i] with the addresses of the rows of the matrix in the first step because eventually it will be filled with the values anyway. Why to bother fill it with some addresses that will be overwritten by values eventually??
I'm sure I'm missing something here.

**tabstop** · 01-16-2011

You have to provide Data[i] addresses, otherwise you won't be able to even contemplate Data[i][j]. If you don't know where Data[i] is, how can you do "j spaces after Data[i]" (which is what Data[i][j] means, after all)?

**Elnaz** · 01-16-2011

So, the first

Code:

size_x*sizeof(double *)

memory block is allocated for this? for Data[i]s to include some addresses? We need more space in memory than the actual size of the matrix. We need the actual size of the matrix plus a vector to include the addresses of the rows separately. Is that it?

**tabstop** · 01-16-2011

Originally Posted by Elnaz

So, the first

Code:

size_x*sizeof(double *)

memory block is allocated for this? for Data[i]s to include some addresses? We need more space in memory than the actual size of the matrix. We need the actual size of the matrix plus a vector to include the addresses of the rows separately. Is that it?

Got it in one.

Thread: understanding dynamic matrix declaration

Thread Tools

Search Thread

Display

understanding dynamic matrix declaration

Similar Threads

matrix multipilation with mpi

C program - inverse of a matrix

problem with matrix size declaration

Weird errors.

how to search for something in the matrix which is( 2-D array)