A polynomial is identically zero if it evaluates to zero for all elements of its domain.
The degree of a term of a multivariate polynomial is the sum of the degrees of each variable in the term.
A polynomial is multilinear if it is linear in each of its terms when each of the other terms is held fixed, in other words, each variable has degree 1 or 0.
The total degree of the polynomial is the degree of the highest degree term.
For an m-variate multilinear polynomial, the total degree must be ≤m because no variable can have degree >1.
Schwartz-Zippel Lemma
Let F be any field and let g:Fm→F be a nonzero m-variate polynomial of total degree ≤d. Then on any finite set S⊆F
x←SmPr[g(x)=0]≤∣S∣d
For a subset S⊆F
The product set Sm=S×S×S… from which x is drawn contains ∣S∣m values.
The probability that a g(x)=0 for a random vector x∈Sm is ≤∣S∣d, so there are at most d⋅∣S∣m−1 values of x∈Sm where g(x)=0.
It follows that any two polynomials pa and pb of total degree ≤d can agree on at most ∣S∣d fraction of the points in Sm, or d⋅∣S∣m−1 values of x∈Sm
pa−pb cannot be a polynomial with more than d⋅Sm−1 roots
For a Finite Field Fp
For an m-variate polynomial g:Fpm→Fp of total degree ≤d, the probability that an m-dimensional vector x∈Fp will be a root of g is ≤d/p.
x∈FpmPr[g(x)=0]≤∣Fp∣d⇒x∈FpmPr[g(x)=0]≤pd
There are pm possible m-dimensional vectors in Fpm, so there are ≤d⋅pm−1 roots of g.
pm⋅pd=d⋅pm−1
Univariate Case
Reed-Solomon and univariate Lagrange interpolation for an n-dimensional vector both resulted in unique univariate polynomials of degree ≤n−1. Applying Schwartz-Zippel in the univariate case, we can verify their distance amplifying properties.
Let pa be the unique univariate extension of degree ≤n−1 for a vector a∈Fpn and pb be the the univariate extension of degree ≤n−1 for a different vector b∈Fpn. Then let q=pa=pb be a degree ≤n−1 polynomial. By Schwartz-Zippel:
x∈Fp1Pr[q(x)=0]≤pn−1
Thus, pa and pb have an pn−1 probability of agreeing at any point x∈Fp.