⇤  <   ?  >   ⇥

⇤  <   ?  >   ⇥
Forum . RSS . Digg . Facebook . Reddit . Twitter . Stumbleupon

The mouseoever text is wrong here... both ints and floats are 32-bit data types, so there is the same number of each. >That's just a standard in programming languages; there are a lot more numbers that have a unique definition in the type of data a float is ultimately supposed to represent. Nope... There are only negative infinity + positive infinity + 1(0)integers. There are an infinite number of floating point numbers between each integer. i.e. between 0 and 1 there numbers from 0.000...1 to 0.999...9 so the number of floating point numbers is infinity to the power of infinity. Infinity doesn't really work like that. In any case, the set of integers is countably infinite while the set of reals are uncountably infinite. There are >more< reals even in the interval [0, 1] than there are integers. You can pretty much make that interval as small as you like; it'll still contain more elements as long as you're not looking at an empty set such as [0, 0]. See Cantor's diagonal argument for more details; it's not the only proof that this is the case, but it's probably the easiest to understand. slightly fewer floats than ints... (wasted NaNs) slightly fewer doubles than long ints... (wasted NaNs) however, many more long doubles than long long ints! (80-128 bits compared with 64)