Problem with code

The problem with my code arises when I try putting in a value for quadmin and quadmax of around 60000. It starts saying all the results are negative numbers, though they can't be because of the numbers I'm using.

Code:

			short a;
			short bc;
			short cc;
			unsigned short iterations;
			cout << " " << endl;
			cout << " " << endl;
			cout << "----------------------------------------------" << endl;
			cout << " " << endl;
			cout << " " << endl;
			cout << "How many quadratics? (from n=min to n=max)" << endl;
			cout << "min?" << endl;
			cin >> quadmin;
			cout << "max?" << endl;
			cin >> quadmax;
			cout << "How many iterations? (from 0 to 65,535)" << endl;
			cin >> iterations;
			cout << "a?" << endl;
			cin >> a;
			cout << "b c?" << endl;
			cin >> bc;
			cout << "c c?" << endl;
			cin >> cc;
			cout << " " << endl;
			cout << " " << endl;
			cout << "-----------------------------------------" << endl;
			cout << " " << endl;
			cout << " " << endl;

	
			
			for(n = quadmin; n <= quadmax; n++) 
			{
				//keeps track of how many primes result for this quadratic
				int primes = 0;	
				//keeps track of the total iterations for this quadratic
				int total = 0;	
				
				//START of test for increasing values of i
				for(unsigned short i=0;i<iterations;i++)
				{
		


					short flag = 0;

					__int64 num = ((a*i*i) + (((2*a*n)+bc)*i) + ((a*n*n)+(bc*n)+cc));
		
					
					if(num>1)
					{
						if((num%2)==0)	//checks against 2
						{		

							flag = 1;
						}
		
						else
						{

							// checks odds from 3 up to sqrt(n)
							for(long m = 3;m<=((long)(sqrt(num)));m+=2)	
							{	
				
								if((num%m)==0)
								{ 
					
									flag = 1;
									break;
			
								}
	
							}

						}
		
		
						if(flag==0)
						{

							primes = primes + 1;

						}

						total++;
					}

					else
					{
						long double r = num;
						cout << r << " was tossed out for being <= 1" << endl;
						cout << " " << endl;
					}
		
				}

									
				cout << a << "x^2 + " << ((2*a*n)+bc) << "x + " << ((a*n*n)+(bc*n)+cc) << endl;
				cout << primes << " out of " << total << " were prime." << endl;
				cout << " " << endl;

			}

and here's the output (including the numbers I used)

Code:

--------------------------------------------


How many quadratics? (from n=min to n=max)
min?
60000
max?
70000
How many iterations? (from 0 to 65,535)
81
a?
1
b c?
-1
c c?
41


-----------------------------------------


-6.95027e+008 was tossed out for being <= 1

-6.94907e+008 was tossed out for being <= 1

-6.94787e+008 was tossed out for being <= 1

-6.94667e+008 was tossed out for being <= 1

-6.94547e+008 was tossed out for being <= 1

-6.94427e+008 was tossed out for being <= 1

-6.94307e+008 was tossed out for being <= 1

-6.94187e+008 was tossed out for being <= 1

-6.94067e+008 was tossed out for being <= 1

-6.93947e+008 was tossed out for being <= 1

What are the types of n,quadmin and quadmax ?

Because if they are short, as in your previous code, then you've wrapped your positive integers into negative

Oops, sorry I didn't include that part originally.

They are all longs.

long n;
long quadmin;
long quadmax;

I'd change those other variables to be long as well

For instance
> (a*i*i)
This is evaluated as a short expression which suggests rounding and truncation is going on.

The expression only gets converted to __int64 when the assignment is made, and by then it's too late

Ok, I switched all shorts to long (except for the flag)... and it still doesn't work! The same result as before.

I even replaced:

__int64 num = ((a*i*i) + (((2*a*n)+bc)*i) + (anum + bnum + cnum))

with:


__int64 = (a*i*i);
__int64 = (((2*a*n)+bc)*i);
__int64 = ((a*n*n)+(bc*n)+cc);
__int64 num = (anum + bnum + cnum);



And it still didn't work. It keeps saying the current num is a negative, which is impossible. In my example, I go up to 80 in the iterations, and 70000 in the n... So, num is:

1*80*80 + (2*1*70000 - 1)*80 + ((1*70000*70000)+(-1*70000)+41)

which is 4,911,136,361

The range for __int64 is –9,223,372,036,854,775,808 to 9,223,372,036,854,775,807

If num was a "long", it would be expected to screw up due to overflowing numbers, but its much larger than a "long".. I don't get it!

You've still got the same problem
1*70000*70000 does not fit into an unsigned long result

Try
( (__int64 )a*n*n )
which should force this part of the expression to be evaluated using __int64 arithmetic

Thanks! That did it! I had to basically put (__int64) in front of everything.

The final version was:

__int64 num = (((__int64)a*i*i) + (((__int64)((__int64)2*a*n)+bc)*i) + (((__int64)a*n*n)+((__int64)bc*n)+cc));