Using delete[] in a function: Invalid address specified to RtlValidateHeap

Question

I'm writing a program to solve the Bloch equations for a magnetic field that changes in time. The values of my magnetic field vary over too great a range to use a simple Runge Kutta method, but I have a standard 4th order with a constant step size set up to work with a constant uniform magnetic field. Now I'm trying to rig an a program that uses an embedded 4-5 order method to use the truncation error, and adjust my stepsize (ala the methods in Numerical Recipes in C (2nd ed))

The function as I have it is thus:

void rungekuttaKC(double a[6], double b[6][6], double c[6], double cs[6], double h, double gammaN, double bx, double by, double bz, double *mu) {
double *mu5 = new double[3];
mu5 = mu;
double *mu4 = new double[4];
mu4 = mu;
double *kx = new double[6];
double *ky = new double[6];
double *kz = new double[6];
double add[3] = { 0.0, 0.0, 0.0 };
double addmu[3] = { 0.0, 0.0, 0.0 };
double addmus[3] = { 0.0, 0.0, 0.0 };
for (int i = 0; i < 6; i++) {
    for (int j = 0; j < i; j++) {
        add[0] = add[0] + kx[j] * b[i][j];
        add[1] = add[1] + ky[j] * b[i][j];
        add[2] = add[2] + kz[j] * b[i][j];
    }

    kx[i] = h*gammaN*((mu[1] + h*add[1]) * bz - (mu[2] + h*add[2]) * by);
    ky[i] = h*gammaN*((mu[2] + h*add[2]) * bx - (mu[0] + h*add[0]) * bz);
    kz[i] = h*gammaN*((mu[0] + h*add[0]) * by - (mu[1] + h*add[1]) * bx);

}
for (int l = 0;l < 6;l++){
    addmu[0] = addmu[0] + kx[l] * c[l];
    addmu[1] = addmu[1] + ky[l] * c[l];
    addmu[2] = addmu[2] + kz[l] * c[l];

    addmus[0] = addmu[0] + kx[l] * cs[l];
    addmus[1] = addmu[1] + ky[l] * cs[l];
    addmus[2] = addmu[2] + kz[l] * cs[l];
}
for (int m = 0;m <=2; m++) {
    mu4[m] = mu4[m] + addmus[m];
    mu5[m] = mu5[m] + addmu[m];
    mu[m] = mu4[m];
}

//delete[] mu5;
//delete[] mu4;
//delete[] kx;
//delete[] ky;
//delete[] kz;
//delete[] add;
//delete[] addmu;
//delete[] addmus;

}

a, b, c, and cs are the values in the Bucher tableau for the Cash-Karp method, h is the step size,1.0e-8.

static double a[6] = { 1.0 / 5.0, 3.0 / 10.0, 3.0 / 5.0, 1.0, 7.0 / 8.0 };
static double bij[6][6] = { {0.0,0.0,0.0,0.0,0.0,0.0},
    {1.0 / 5.0, 0.0, 0.0, 0.0, 0.0, 0.0},
    {3.0 / 40.0, 9.0/40.0, 0.0, 0.0, 0.0, 0.0},
    {3.0/10.0, -9.0/10.0, 6.0/5.0, 0.0, 0.0, 0.0},
    {-11.0/54.0, 5.0/2.0, -70.0/27.0, 35.0/27.0, 0.0},
    {1631.0/55296.0, 175.0/512.0, 575.0/13824.0, 44275.0/110592.0, 253.0/4096.0} };
static double c[6] = { 37.0 / 378.0, 0.0, 250.0/621.0, 125.0/594.0, 0.0, 512.0/1771.0 };
static double cs[6] = {2825.0/27648.0, 0.0, 18575.0/48384.0, 13525.0/55296.0, 277.0/14336.0, 1.0/4.0 };

the magnetic field values are set up outside the function by an interpolator, and passed in, but just to make sure it's producing oscillations, I am testing it with a constant field, where bx and by are 0, and bz is 5. The mu that gets passed in starts as {1,0,0}

Now, the whole program runs in the debugger when I don't use any of the delete commands at the end of the function, but there's a pretty obvious memory leak.

When I use delete with only the arrays I initialize with new, a window pops up with the error "Main.exe has triggered a breakpoint." and no other information. In the output window I get " HEAP[Main.exe]: Invalid address specified to RtlValidateHeap( 00470000, 0016F514 ) Main.exe has triggered a breakpoint."

And when I delete all of them at the end of the function, I get the same error, but a different address in the RtlValidateHeap.

I've tried running in release mode, and switching my runtime library and I get "Critical error detected c0000374 Main.exe has triggered a breakpoint."

Am I misusing delete[] or is there something else I'm doing wrong? what does this error mean?

Answer 1

You are deleting add, addmu, and addmus. These were allocated on the stack (not on the heap with new). Therefore, they cannot be delete[]'d.

Since you know all the sizes of your arrays at compile-time and they are small, you should probably allocate them all on the stack. For dynamic arrays, in C++ the best practice is to use a container like vector, which frees you from the responsibility of using delete manually.

Additionally, as Borgleader said, the lines

double *mu5 = new double[3];
mu5 = mu;
double *mu4 = new double[4];
mu4 = mu;

are not doing what you expect. First, a double[3] is allocated on the heap, and its address is given to mu5. Then, you erase that address and overwrite it with mu. When you delete[] mu5, you are freeing the memory pointed to by mu (which is probably not intended), and the new double[3] is still standing. To correct it now with a minimum of changes, you could replace that snippet with

double *mu5 = new double[3];
std::copy(mu, mu + 3, mu5);
double *mu4 = new double[4];
std::copy(mu, mu + 4, mu4);

However, writing this is C style, not C++ style (although it uses C++ syntax).