Shouldn't this be small but nonzero?

[daviddoria]

double P = 3.5/(.1 * sqrt(2.0*3.14159)) * exp(-pow(5.0,2) / (2.0*pow(.1,2)));
	cout << P << endl;
	if (P==0)
		cout << "P is zero!" << endl;

Is this underflowing so it is getting rounded to zero? The problem is that I am taking the log of this, so if it is zero it breaks everything because it is -inf instead of a normal double. Is there a way to prevent this?

Thanks,

Dave

[VernonDozier]

double P = 3.5/(.1 * sqrt(2.0*3.14159)) * exp(-pow(5.0,2) / (2.0*pow(.1,2)));
	cout << P << endl;
	if (P==0)
		cout << "P is zero!" << endl;

Is this underflowing so it is getting rounded to zero? The problem is that I am taking the log of this, so if it is zero it breaks everything because it is -inf instead of a normal double. Is there a way to prevent this?

Thanks,

Dave

You are raising e to the -1125th power. That's below 10 to the -308th power, which is the bare minimum of what a double can hold before rounding to 0. You may need to use an arbitrary precision data type instead.

[daviddoria]

Is there one of those in c++?

[NeoKyrgyz]

You may want to implement your own Double Class (which will hold data as a string internally)

[Rashakil Fol]

You may want to implement your own Double Class (which will hold data as a string internally)

Or how about a better representation.