3 The “original sin” of computer arithmetic
3.1 The acceptance of incorrect answers
Binary floats can express very large and very small numbers, with a wide range of precisions. But they cannot express all real numbers. They can only represent rational numbers where the denominator is some power of 2. Some readers might be taken aback by this statement, since they have long ago gotten used to having floats represent square roots, π, decimals, and so on, and were even taught that floats were designed to represent real numbers. What is the problem?
The “original sin” of computer arithmetic is committed when a user accepts the following proposition: ...
Get The End of Error now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
