Writing a function to allocate an upper triangular matrix

Go To StackoverFlow.com


I am trying to write a function to allocate an upper triangular matrix. it should return a pointer to the first element of the allocated array. I also need to use dynamic allocation to ensure that the exact amount of required memory is allocated but I am not quite sure how to do that... Any tips or advice would be greatly appreciated! I am a noob to c++.. Anyway here is my code if that helps!

#include <iostream>

using namespace std;

int main ()
int a[3][3],i,j;  //creating two dimensional array
int d; 
int * p;  
cout<<"Please Enter the 9 elements of matrix (with spaces): ";  

cout<<"\nThe matrix is\n ";  

cout<<"\nSetting zero in upper triangular matrix\n";  

  return 0;  
2012-04-03 20:11
by CornucopiaBeebee
I'm not sure what you're expecting things like cin>>d,&a[i][j] to do.. - Oliver Charlesworth 2012-04-03 20:14
The program is compiling by the way but its only printing out the last element of the array entered and the zeros highlighting the upper matrix.. - CornucopiaBeebee 2012-04-03 20:19
I think you need to explain what you think cin>>d,&a[i][j] does. It surely doesn't do what you think it does - David Heffernan 2012-04-03 20:22
cin>>d,&a[i][j]; is a funny construct - you read into d and then get address of a[i][j] promptly discarding it - Anycorn 2012-04-03 20:23


As per oli's comment I think you are looking to do

cin >> d; a[i][j] = d;



I suggest reading something like .... http://www.cplusplus.com/doc/tutorial/basic_io/

theres your first issue

dynamic allocation is done through code like new and malloc try reading...


as for how to store your upper matrix... I would recommend just using a normal 2d matrix chances are it will work better with most matrix libraries out there.

Good luck with your homework.

2012-04-03 20:26
by Medran


Normally when allocating an upper triangular matrix of size N you allocate N + N-1 ... + 1 (look up sum of integers) elements, and then you need to either create access mechanisms (likely what is intended here) so that when you want element M(i,j) you get the element in row i, column j, despite the fact that (nearly) half the elements are missing; or do that manually when using the matrix in whatever matrix manipulation code uses it.

You treat the one dimensional array as if the first N elements are the first row, the next N-1 elements are the (right N-1 elements of the) second row, and so on.

Because this question looks and smells like homework, I think that's about as much hint as appropriate.

2012-04-03 20:20
by DRVic
I shouldn't say "normally". The major linear algebra packages support triangular matrices in both this packed representation as well as the wasteful square-matrix-using-only-half-of-the-entries, which may often be just as good or better in terms of performance - leftaroundabout 2012-04-03 20:25
The only reason you would want the compact representation is if the matrix is large. But at the same time, the question specified "the exact amount of required memory". From which I infer that the intent is to teach getting that right - DRVic 2012-04-03 20:32