Pointers and multi dimensional arrays

Hi,
I have troubles understanding the relationship between pointers (to double) and multidimensional arrays (of doubles).

I have a C++ routine that returns pointers to doubles, and a class object that I'm creating computes multidimensional arrays, because I like to think of the data as matrices.

Then when the computation is done all I have to do is to copy the array values into the location pointed by the pointer, increment it and so on.

So this is what I have:

Code:

---

void STDCALL FIESUB ( double *dfddis, double *dfdvel )
{
...
}

---

In this routine I instantiate an object of the class Field:

Code:

---

    class msField
    {
    private:
        double m_dfdx[6][6];
        double m_dfdxdot[6][6];
    public:
        void getForceDerivatives(double *fddist, double *fdvel);
    };

---

The idea being that in getForceDerivatives I would do something like this:

Code:

---

void msField::getForceDerivatives(double *fddist, double *fdvel)
{
    // retrieves the force derivatives from the field member datas
    int i, j;
    for ( i = 0; i < 6; i++)
    {
        for ( j = 0; j < 6; j++)
        {
            *fddist++=m_dfdx[i][j];
            *fdvel++=m_dfdxdot[i][j];
        }
    }
}

---

That function is called like this (inside of FIESUB):

Code:

---

rField.getForceDerivatives(dfddis, dfdvel);

---

The result?
Mayhem.
Obviously I am forgetting something.
Can anyone explain the relationship between pointers to double and arrays of doubles (multidimensional arrays, that is...).
Any help much appreciated!!
And72.

A pointer to a double is just that, a pointer to a double. It can be used as different things, including a pointer to a single double somewhere in memory, or the pointer to a double that is part of a single-dimensional array of doubles.

In your code, you have fddist and fdvel, which are pointers to double. You use them as if they were pointing to single dimensional arrays of double. You increment them 36 times in your nested for loops, so you are assuming that they each originally pointed to an array that is at least 36 doubles long.

The fact that you are also using multi-dimensional arrays here should not be a problem. It appears that you are using those correctly, assuming you have initialized them to valid values.

The potential source of mayhem is if fddist and/or fdvel do not point to an array of at least 36 doubles. If they point to single instances of a double, or if they point to nowherem or if they point to an array whose size is smaller than 36, then you will likely get problems.

It might help if you explained what you mean by mayhem, and if you post how dfddis and dfdvel are originally created or assigned to, since those are the pointers you are passing to the function.

Hi and thanks for your answer.
The problem is that I am writing an extension: in other words I do not control upstream from the routine in question.
So by mayhem I mean that the main application (in development) is getting a bunch of NAN (not a number?) as a consequence of my assignment.
Now looking at the code i suspect that I am doing something wrong in the computation, not in the assignment.
I trust that whoever wrote the main application reserved enough memory for storing 36 values, but I will double check that.
Lots of blathering on my part, but thanks for double checking my code.
And72.

you might try using meaningful variable names too! what the hell is dfddis supposed to mean? It looks like your left hand went nuts on the keyboard! :D

Quote:

---

Originally Posted by ChaosEngine

you might try using meaningful variable names too! what the hell is dfddis supposed to mean? It looks like your left hand went nuts on the keyboard! :D

---

Those variable names kind of make sense to me...

You mean forward_distance / forward_velocity? Although the speed of development is important, typing speed is not really a factor here.

(One thing I dislike about C is its love for acronyms - I'd take size over strlen every day :) Is it because C doesn't have namespaces - obfuscating the names makes naming collisions less probable?)