Assertion error

Question

I have written(shown below) a method getCelsius to convert from Fahrenheit to Celsius but when I run the program it shows me an assertion error.

Code:

Method in class Temperature:

public double getCelsius() {

        double c = 0;
        c =  ((this.temp_value - 32) * (5 / 9));
        return c;
    }

and here is the method call in class TestTemperature, which gives me an assertion error:

Temperature t = new Temperature(212.0, "F");

assert t.getCelsius() == 100.0;  // < error in this line

Help!

possible duplicate of [strange output in comparision of float with float literal](http://stackoverflow.com/questions/1839422/strange-output-in-comparision-of-float-with-float-literal) — Ignacio Vazquez-Abrams, Apr 24 '11 at 03:53

score 7 · Answer 1 · answered Apr 24 '11 at 03:53

There are a couple problems: firstly, the 5 / 9 part is using integer division, so it will return 0 all the time. You need to change 5 / 9 to 5f / 9 or something similar.

Secondly (212f - 32) * (5f / 9) is not exactly 100.0: it's hard to do floating point comparison because not all values that you can write in Java are exactly representable in IEEE754 floating point numbers. So you should compare the two numbers like this: assert Math.abs(t.getCelsius() - expected) < 0.000001 or some other desired maximum difference.

score 0 · Answer 2 · answered Apr 24 '11 at 03:52

You need to take into account precision/representation. i.e. perform your test like so:

assert Abs(t.getCelsius() - 100.0) < epsilon;

where epsilon is your tolerance (say, 0.000001)

You should very rarely use exact equality comparison when dealing with floating point quantities; use a tolerance instead.

Also, write your conversion factor to use floating point arithmetic:

    c =  ((this.temp_value - 32) * (5.0 / 9));

irreputable · Answer 3 · 2011-04-24T04:52:33.607

well, if you want it that way

// will return "exact" value if input is 212.0 or 32.0
double f2c(double f)
{
    return  (f-32.0)/(212.0-32.0) * 100.0;
}

assert f2c(212.0)==100.0 ;  // true!
assert f2c( 32.0)==  0.0 ;  // true!

more generally, if we have two end points (x1,y1) and (x2,y2), this linear interpolation

y = (x-x2)/(x1-x2) * y1 + (x-x1)/(x2-x1) * y2

will evaluate "exactly" to y1 at x=x1, and y2 at x=x2. Valid for double/float/int/short/byte

Assertion error

3 Answers3