Floating Point Doubt

[pits]

Hi...Have this doubt about floating points in C

Look at the two programs below.

1. main()
{ float a = 0.7;
if(a < 0.7)
printf("Less");
else
printf("More");
}


2. main()
{ float a = 0.7, b=0.7;
if(a < b)
printf("Less");
else
printf("More");
}

Now, the first program prints Less and the second one prints More. And in both the programs if the value of a is printed, we get 0.700000.
The same program behaves differently when we use the datatype double instead of float.

Can anyone tell me why?
--
pits

[Ganon11]

Take a look at this article, which explains why this type of error occurs.

[pits]

Thanks! This link looks useful...

[weaknessforcats]

You have fallen victim to a) arithmetic conversions and b) automatic rounding.

When you have:

{ float a = 0.7, b=0.7;
if(a < b)

the calculation is done as a float (7 signiticant digits) whereas this code:

{ float a = 0.7;
if(a < 0.7)

converts a to double since 0.7 is a double. The calculation is done as a double (15 significant digits) so your answer may vary from the float version.

Plus, operators like <, >, ==, !=, etc generally cannot be used with floating point. If you Google for Floating Point Arithmetic you will be inudated with material.

Also, this code:

float a = 0.7;

should produce a compiler warning about truncation and possible loss of data as that double (0.7) is jammed into the float.

[JosAH]

What every computer scientist should know about floating-point arithmetic.

kind regards,

Jos