Initializing array in cpp and padding with zeros

Question

I'm new a c++, switched from matlab to run simulations faster.
I want to initialize an array and have it padded with zeros.

    # include <iostream>
# include <string>
# include <cmath>
using namespace std;

int main()
{
    int nSteps = 10000;
    int nReal = 10;
    double H[nSteps*nReal];
    return 0;
}

It produces an error:

expected constant expression    
cannot allocate an array of constant size 0    
'H' : unknown size

How do you do this simple thing? Is there a library with a command such as in matlab:

zeros(n);

C-style arrays are mainly there for historical reasons, try to avoid them. They have weird behaviour such as not being able to have the size set at runtime, and causing stack overflows. — M.M, Sep 24 '15 at 07:34
Adding `const`/`constexpr` to both sizes should fix your problem. — Jarod42, Sep 24 '15 at 07:36
@sergej: Probably one that doesn't allow variable length arrays (which are not standard) by default. — Benjamin Lindley, Sep 24 '15 at 07:42
adding {const} indeed sovled the problem, but I also define arrays with lengths defined in the runtime. the {std::vector...} did not do anything — jarhead, Sep 24 '15 at 07:43

score 4 · Accepted Answer · edited May 23 '17 at 11:44

4

Stack-based arrays with a single intializer are zero-filled until their end, but you need to make the array bounds equal to be const.

#include <iostream>

int main()
{       
    const int nSteps = 10;
    const int nReal = 1;
    const int N = nSteps * nReal;
    double H[N] = { 0.0 };
    for (int i = 0; i < N; ++i)
        std::cout << H[i];
}

Live Example

For dynamically allocated arrays, you best use std::vector, which also does not require compile-time known bounds

#include <iostream>
#include <vector>

int main()
{
    int nSteps = 10;
    int nReal = 1;
    int N = nSteps * nReal;
    std::vector<double> H(N);
    for (int i = 0; i < N; ++i)
        std::cout << H[i];
}

Live Example.

Alternatively (but not recommended), you can manually allocate a zero-filled array like

double* H = new double[nSteps*nReal](); // without the () there is no zero-initialization

edited May 23 '17 at 11:44

Community

1
1

answered Sep 24 '15 at 07:42

TemplateRex

69,038
19
164
304

thanks using the vector library helped me with both issus, very helpful for my kind of programming! – jarhead Sep 24 '15 at 07:59
Note that you don't need to use an enum pre-C++11. You can just use `const`. (and since I don't think VS2010 has `constexpr`, `const` is probably what he would use) – Benjamin Lindley Sep 24 '15 at 08:02
@BenjaminLindley thanks, updated. I am so used to using `constexpr` nowadays that I didn't realise this. It's only for non-type template parameters that you can't always use `const` variables (depending on linkage etc.), and could need `constexpr`. – TemplateRex Sep 24 '15 at 10:18

Giorgi Moniava · Answer 2 · 2015-09-24T07:54:48.827

2

If you know length in advance you can just do

#define nSteps 10000
#define nReal 10

Then

double H[nSteps*nReal] = {0};

Or alternatively you can as well add const keyword to your sizes instead of using defines.

edited Sep 24 '15 at 07:54

answered Sep 24 '15 at 07:47

Giorgi Moniava

27,046
9
53
90

1

Better to use `constexpr` over MACRO. – Jarod42 Sep 24 '15 at 07:51

Initializing array in cpp and padding with zeros

2 Answers2