Continuous Mathematics HT23, Floating point numbers


Flashcards

What is the most common form of a computer representation of real numbers, in terms of $p$, $m$ and $e$?

\[\pm \left( 1 + \frac{m}{2^p} \right)\cdot 2^e\]

In the computer representation of real numbers as

\[\pm \left( 1 + \frac{m}{2^p} \right)\cdot 2^e\]

How many bits are used for the binary representation of $m$?

\[p\]
a == b

How should you modify this code when a and b are floats?

abs(a - b) < small

@Define machine epsilon.

The quantity $\epsilon$ such that every real number $x \in \mathbb R$ can be stored as $\tilde x$ with relative error no more than $\epsilon$.