Difference between revisions of "Address"
m |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 11: | Line 11: | ||
== Properties == | == Properties == | ||
* The [[influence]] of one of the <math>k</math> address bits is 1/2 (changing the address means pointing to a new, independent, data bit). The influence of every one of the <math>2^k</math> data bits is <math>1/2^{k}</math> (that's the probability that the address points to that particular bit). | * The [[influence]] of one of the <math>k</math> address bits is 1/2 (changing the address means pointing to a new, independent, data bit). The influence of every one of the <math>2^k</math> data bits is <math>1/2^{k}</math> (that's the probability that the address points to that particular bit). | ||
| − | * The degree of <math>f</math> is <math>d = k+1 \approx \log n</math>: The [[Fourier representation]] is a sum of <math>2^{k}</math> indicator monomials of the form <math>y_{a}\prod_{i=1}^{k}\left(x_{i}-a_{i}\right)</math> | + | * The degree of <math>f</math> is <math>d = k+1 \approx \log n</math>: The [[Fourier representation]] is a sum of <math>2^{k}</math> indicator monomials of the form <math>y_{a}\prod_{i=1}^{k}\left(x_{i}-a_{i}\right)</math>. |
| − | |||
== References == | == References == | ||
<references/> | <references/> | ||
| − | [[Category:balanced function]] | + | [[Category:balanced function]] [[Category:Noise sensitive function]] |
Latest revision as of 14:59, 7 April 2021
Definition
Let [math]n = k + 2^k[/math]. A function [math]f:\{-1,1\}^n \to \{-1,1\}[/math] is called an address function if it returns the bit pointed to by the first [math]k[/math] bits:
[math]f(x_1, \ldots, x_k, y_1,\ldots,y_{2^k}) = y_{\tilde{x}}, [/math]
where [math]\tilde{x}[/math] is number whose binary representation of the vector [math](x_1,\ldots, x_k)[/math].
Properties
- The influence of one of the [math]k[/math] address bits is 1/2 (changing the address means pointing to a new, independent, data bit). The influence of every one of the [math]2^k[/math] data bits is [math]1/2^{k}[/math] (that's the probability that the address points to that particular bit).
- The degree of [math]f[/math] is [math]d = k+1 \approx \log n[/math]: The Fourier representation is a sum of [math]2^{k}[/math] indicator monomials of the form [math]y_{a}\prod_{i=1}^{k}\left(x_{i}-a_{i}\right)[/math].
