TOPICS
Search

Boolean Noise Operator


The Boolean noise operator averages a function over random coordinate reversals on {-1,1}^n. For -1<=rho<=1, let X have the discrete uniform distribution on this set, and put Y_i=X_iZ_i, where the signs Z_i in {-1,1} are independent of X and one another, with Pr(Z_i=1)=(1+rho)/2. The linear operator is

 T_rhou(x)=E[u(Y)|X=x],

where the right side is a conditional expectation (O'Donnell 2021, p. 55). Thus T_1u=u, T_0u=E[u(X)], and T_rho(x_i)=rhox_i. On a Boolean Fourier expansion, it acts diagonally,

 T_rhou(x)=sum_(S subset= {1,...,n})rho^(|S|)u^^(S)product_(i in S)x_i.

Its crossover probability as a binary symmetric channel is (1-rho)/2.


See also

Binary Symmetric Channel, Boolean Fourier Expansion, Conditional Expectation, Courtade-Kumar Conjecture, Hellinger Conjecture, Linear Operator, Noise

Explore with Wolfram|Alpha

References

O'Donnell, R. Analysis of Boolean Functions. 21 May 2021. https://arxiv.org/abs/2105.10386.

Cite this as:

Weisstein, Eric W. "Boolean Noise Operator." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/BooleanNoiseOperator.html

Subject classifications