Vector values becoming NaN - c++

Question

I keep getting nan (not a number) as a result for the various floats and vector components throughout my program. I am almost 100% certain is has to do with collision because this was never an issue before I messed with the collision code.
Even the sum in the magnitude method is showing up as NaN when I debug it, which leads me to believe a vector is breaking before being passed into the function, but I cannot find which one it is.
Here are the methods that I believe are relevant to the problem
void Simplex::PhysicsInfo::Collision(PhysicsInfo info)
{
vector3 oldVel = velocity;
if (magnitude(oldVel) == 0.0f)
{
return;
}
vector3 nextVelDirect = glm::normalize(info.position - position);
//make all ball to ball collisions elastic
float angle = acosf(glm::dot(oldVel, nextVelDirect)
/ (magnitude(glm::normalize(oldVel)) * magnitude(nextVelDirect)));
angle = sinf(angle);
if (angle < 0)
angle *= -1;
float nextVecMag;
if (magnitude(info.velocity) == 0 && angle != 1)
{
//This next line is not correct, use if actual line isn't working and you absolutely need something
//info.velocity = 0.5f * oldVel.length * nextVelDirect;
//actual line
info.velocity = angle * magnitude(oldVel) * nextVelDirect;
vector3 nextVec = (magnitude(oldVel) * oldVel)
- (magnitude(info.velocity) * info.velocity);
nextVecMag = magnitude(nextVec);
if (nextVecMag < 0)
{
nextVecMag *= -1;
}
nextVecMag = sqrt(nextVecMag);
velocity = nextVecMag * glm::normalize(nextVec);
}
else if (magnitude(info.velocity) == 0)
{
info.velocity = oldVel;
velocity = vector3(0.0f);
}
if (isnan(velocity.x) || isnan(velocity.y) || isnan(velocity.z))
{
std::cout << "-" << std::endl;
}
}
PhysicsInfo::PhysicsInfo(float mss, vector3 pos, vector3 cent, vector3 limit)
{
velocity = vector3(0.1f);
acceleration = vector3(0.0f);
mass = mss;
position = pos;
center = cent;
limits = limit;
frictionMagnitude = 0.005f;
}
vector3 PhysicsInfo::normalize(const vector3 &v)
{
float sum = (v.x * v.x) + (v.y * v.y) + (v.z * v.z);
if (sum < 0)
{
sum *= -1;
}
float length_of_v = sqrt(sum);
return vector3(v.x / length_of_v, v.y / length_of_v, v.z / length_of_v);
}
float PhysicsInfo::magnitude(const vector3 &v)
{
float sum = (v.x * v.x) + (v.y * v.y) + (v.z * v.z);
if (sum < 0)
{
sum *= -1;
}
float length_of_v = sqrt(sum);
if (isnan(length_of_v))
{
throw ExceptionCollidedUnwind;
}
return length_of_v;
}
Sorry for the formatting. I am not used to posting here. Any help would be greatly appreciated.

Answer 1

I have not fixed the issue yet, but I have deciphered that the cause was the parameter to the acosf() function located the the Collision method. The dot method sometimes returns outside of the required range of [-1, 1], the required range of floats for the parameter to the acosf() method. This was caused because I did not normalize both vectors before getting the dot product.