Why a variable's value changes without altering it?

[xianwen]

Hi there,

I'm writing a function to solve polynomial problems. A strange thing happened, that, the variable b, in the program, changes to another value, though I didn't alter it. Can someone give me a hint?

#include <stdio.h>
#include <math.h> 

void polinomial(double a, double b, double c, double * x1, double * x2) {
    double bSqrMns4ac;
    printf("{a, b, c} = {%f, %f, %f}\n", a, b, c);
    bSqrMns4ac = pow(b, 2) - 4 * a * c;
    printf("bSqrMns4ac = %f\n", bSqrMns4ac);
    if (bSqrMns4ac < 0) {
        *x1 = 0;
        *x2 = 0;
    } else {
        *x1 = (-b - sqrt(bSqrMns4ac))/ (2 * a);
        *x2 = (-b + sqrt(bSqrMns4ac))/ (2 * a);
        printf("{x1, x2} = {%f, %f}\n", (-b - sqrt(bSqrMns4ac))/ (2 * a), (-b + sqrt(bSqrMns4ac))/ (2 * a));
    }
}

int main (void) {
    double beta1 = -1.88138742; // beta1: price coefficient
    double beta2 = 0.07182447; // beta2: price^2 coefficient; Note that beta2 is a in the polinomial.
    double beta3 = 0.30941525; // beta3: coefficient of MSC labelling effect on wild salmon
    double beta4 = 0.70772077; // beta4: coefficient of AB labelling effect on farmed cod
    double beta5 = 0.69330401; // beta5: coefficient of AB labelling effect on farmed salmon
    double betaLabel[3] = {beta3, beta4, beta5}; // We have three c for the polinomial; each labelling effect coefficient reflects a c.
    int priceStar = 8;
    double b, wtpMSC[2], wtpABCod[2], wtpABSalmon[2];
    
    printf("\nbeta1 = %f, \nbeta2 = %f, \nbeta3 = %f (%f), \nbeta4 = %f (%f), \nbeta5 = %f (%f)\n\n", beta1, beta2, beta3, betaLabel[0], beta4, betaLabel[1], beta5, betaLabel[2]);
    
    b = 2 * beta2 * priceStar + beta1;    
    polinomial(beta2, b, betaLabel[0], & wtpMSC[1], & wtpMSC[2]);
    printf("WTP(MSC) = %f, %f\n\n", wtpMSC[1], wtpMSC[2]);
     
    polinomial(beta2, b, betaLabel[1], & wtpABCod[1], & wtpABCod[2]);
    printf("WTP(AB on cod) = %f, %f\n\n", wtpABCod[1], wtpABCod[2]);
     
    polinomial(beta2, b, betaLabel[2], & wtpABSalmon[1], & wtpABSalmon[2]);
    printf("WTP(AB on salmon) = %f, %f\n\n", wtpABSalmon[1], wtpABSalmon[2]);
    
    return 0;
}

[TheBigH]

You've declared it twice.

[MK27]

I'm writing a function to solve polynomial problems. A strange thing happened, that, the variable b, in the program, changes to another value, though I didn't alter it.

Which "b" is that?

        double a = -1.88138742;
        double b = 0.07182447;
        double c[3] = {0.30941525, 0.70772077, 0.69330401};
        double b, x1[2], x2[2], x3[2];

Presuming this didn't result in a compiler error (it probably should), I'd guess b #2 counts, which is uninitialized, and hence, a random value.

[xianwen]

Hi.

Thank you. You were right. After removing second b, it worked. However, it didn't solve the original problem ('cause the code I put on-line was a simplified version).

I've now put the original code in the post. Can you have another look at it?

You've declared it twice.

[xianwen]

Hi MK27,

Thank you for your hints.

I've updated the codes to the one I'm actually running. Can you have another look at it? It doesn't have duplicated declaration problem; but still, the value of b changes.

Which "b" is that?

        double a = -1.88138742;
        double b = 0.07182447;
        double c[3] = {0.30941525, 0.70772077, 0.69330401};
        double b, x1[2], x2[2], x3[2];

Presuming this didn't result in a compiler error (it probably should), I'd guess b #2 counts, which is uninitialized, and hence, a random value.

[TheBigH]

Aha. I found it.

The problem is that you've declared these:
double b, wtpMSC[2], wtpABCod[2], wtpABSalmon[2];

But then you call this:
polinomial(beta2, b, betaLabel[0], & wtpMSC[1], & wtpMSC[2]);

wtpMSC[2] does not exist. You are accessing and modifying memory that you shouldn't, and by bad luck this is messing with b.

[xianwen]

Hi TheBigH,

Thank you very much for finding one of the errors. However, the value of b still changes.

As I changed the code to

#include <stdio.h>
#include <math.h> 

void polinomial(double a, double b, double c, double * x1, double * x2) {
    double bSqrMns4ac;
    printf("{a, b, c} = {%f, %f, %f}\n", a, b, c);
    bSqrMns4ac = pow(b, 2) - 4 * a * c;
    printf("bSqrMns4ac = %f\n", bSqrMns4ac);
    if (bSqrMns4ac < 0) {
        *x1 = 0;
        *x2 = 0;
    } else {
        *x1 = (-b - sqrt(bSqrMns4ac))/ (2 * a);
        *x2 = (-b + sqrt(bSqrMns4ac))/ (2 * a);
        printf("{x1, x2} = {%f, %f}\n", (-b - sqrt(bSqrMns4ac))/ (2 * a), (-b + sqrt(bSqrMns4ac))/ (2 * a));
    }
}

int main (void) {
    double beta1 = -1.88138742; // beta1: price coefficient
    double beta2 = 0.07182447; // beta2: price^2 coefficient; Note that beta2 is a in the polinomial.
    double beta3 = 0.30941525; // beta3: coefficient of MSC labelling effect on wild salmon
    double beta4 = 0.70772077; // beta4: coefficient of AB labelling effect on farmed cod
    double beta5 = 0.69330401; // beta5: coefficient of AB labelling effect on farmed salmon
    double betaLabel[3] = {beta3, beta4, beta5}; // We have three c for the polinomial; each labelling effect coefficient reflects a c.
    int priceStar = 8;
    double b;
    double wtpMSC[2], wtpABCod[2], wtpABSalmon[2];
    
    printf("\nbeta1 = %f, \nbeta2 = %f, \nbeta3 = %f (%f), \nbeta4 = %f (%f), \nbeta5 = %f (%f)\n\n", beta1, beta2, beta3, betaLabel[0], beta4, betaLabel[1], beta5, betaLabel[2]);
    
    b = 2 * beta2 * priceStar + beta1;    
    polinomial(beta2, b, betaLabel[0], & wtpMSC[0], & wtpMSC[1]);
    printf("WTP(MSC) = %f, %f\n\n", wtpMSC[0], wtpMSC[1]);
    
    polinomial(beta2, b, betaLabel[1], & wtpABCod[0], & wtpABCod[1]);
    printf("WTP(AB on cod) = %f, %f\n\n", wtpABCod[0], wtpABCod[1]);
    
    polinomial(beta2, b, betaLabel[2], & wtpABSalmon[0], & wtpABSalmon[1]);
    printf("WTP(AB on salmon) = %f, %f\n\n", wtpABSalmon[0], wtpABSalmon[1]);
    
    return 0;
}

, the value of b changes to x2 when polinomial was called for the first time.

I don't really know why it behaves this way; but I guess we're closer to finding out the last bug.

[TheBigH]

Hmmm, this is very strange. I can't reproduce the bug now. When I compile and run it, b does not change at all.

[xianwen]

Then it may be the problem of my compiler. Which compiler do you use?

[anduril462]

I can't replicate that behavior with the code you just posted. How are you determining that b changes to x2? What print statement is showing that? I added code to print b immediately after you calculate it, then again after every call to polinomial. Here's the output:

$ ./fish


beta1 = -1.881387,
beta2 = 0.071824,
beta3 = 0.309415 (0.309415),
beta4 = 0.707721 (0.707721),
beta5 = 0.693304 (0.693304)


b = -0.732196
{a, b, c} = {0.071824, -0.732196, 0.309415}
bSqrMns4ac = 0.447216
{x1, x2} = {0.441726, 9.752514}
b = -0.732196
WTP(MSC) = 0.441726, 9.752514


{a, b, c} = {0.071824, -0.732196, 0.707721}
bSqrMns4ac = 0.332784
{x1, x2} = {1.081257, 9.112983}
b = -0.732196
WTP(AB on cod) = 1.081257, 9.112983


{a, b, c} = {0.071824, -0.732196, 0.693304}
bSqrMns4ac = 0.336926
{x1, x2} = {1.056343, 9.137897}
b = -0.732196
WTP(AB on salmon) = 1.056343, 9.137897

Note that the first one is immediately after you calculate b = 2 * beta2 * priceStar + beta1. The rest are after each call to polinomial. It never changes. Perhaps the code wasn't rebuilt properly to include the most recent changes?

[anduril462]

Ahh, I see TheBigH already got to it (I'm slow today). I use GCC 4.4.5. Which compiler do you use?

[xianwen]

Thank you very much anduril462. I was using tcc TCC : Tiny C Compiler under Windows 'cause I was using Windows in the school.

Now I'm at home and the code is running fine with GCC. I've sent and email to tcc mailing list and hopefully they'll be able to answer me.

Have a good night

xianwen

Ahh, I see TheBigH already got to it (I'm slow today). I use GCC 4.4.5. Which compiler do you use?