CODEWITHSUNDEEP
c++cycle
Max Frai
Max Frai

Reputation: 64356

Right circular loop

I have to cycle from 0 to 5 with step 0.5. But the cycle should be reversed when the value is 5.

bool forward = true;
int counter = 0;
float val = 5.0f;

// And somewhere in global app cycle:
if (val <= 0.0 && forward) forward = false;
if (forward) val -= .5f;

if (val >= 5.0 && !forward) forward = true;
if (!forward) val += .5f;

But the result is sometimes negative numbers and, I think, that's a bit ugly code.

0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
2.98023e-08
-0.02
2.98023e-08
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.2
0.18

Upvotes: 1

Views: 680

Answers (4)

Oliver Charlesworth
Oliver Charlesworth

Reputation: 272687

The problem you are experiencing is that 0.2 cannot be exactly represented in floating-point. So when you write val -= 0.2, what you're really doing is val -= 0.20000000298023224;. Therefore, you accumulate errors.

The solution is to not use floating-point to control your loop. Use an int instead, and multiply by a constant to get val. e.g.:

int inc = -1;
int i = 10;

while (whatever)
{
    if (i == 0) inc = +1;
    if (i == 10) inc = -1;

    i += inc;

    float val = i * 0.02;
}

Upvotes: 4

Scott Langham
Scott Langham

Reputation: 60411

Instead of specifying the distance to move each time, could the number of divisions be specified? This way, your value won't ever drift due to precision/rounding errors EG:

void Step(float from, float to, int divisions)
{
    float range = to - from;
    float perDivision = range / divisions;

    int i = 0;
    for(;;)
    {
        for(; i < divisions; ++i)
        {
           UseCurrentValue(from + (i * perDivision));
        }
        for(; i > 0; --i)
        {
           UseCurrentValue(from + (i * perDivision));
        }
    }
}

void UseCurrentValue(float val)
{
   // Do something with val here.
}

Upvotes: 0

user173973
user173973

Reputation:

|val - 5.0| < eps && |val - 0 | < eps where | | in your case fabsf() i guess

also keep in mind that there is some loss in precision whenever two almost equal (floating point representation) values are subtracted.

Upvotes: 0

&#214;&#246; Tiib
&#214;&#246; Tiib

Reputation: 11011

Use integers for such cycles from 0 to 10 with step 1 or from 0 to 50 with step 5. Float may be 0.00000012 and so it is not less or equal than 0.0. You can calculate the float from your cycle variable when needed.

Upvotes: 2

Related Questions