In this series of posts, we explore the mathematical foundations of polynomials over a field. These objects are at the heart of several results in computer science: secret sharing, Multi Party Computation, Complexity, and Zero Knowledge protocols.

All this wonder and more can be traced back to a very useful fact about polynomials over a field:

Theorem: any non-trivial polynomial over a field of degree at most $d$ has at most $d$ roots

Let’s unpack this statement. Let $K$ be a field and let $p_0,…,p_m \in K$ be coefficients. A polynomial over $K$ is an element of $K[X]$. Here is an example of one:

\[P=p_0+p_1 X + p_2 X^2 + p_3 X^3 +...+ p_{m-1} X^{m-1} + p_m X^m\]

Recall that a field supports both multiplication and division (i.e., every non-zero element has a unique multiplicative inverse). An important property of fields is that the additive and multiplicative inverses are unique. Note that the set of polynomials $K[X]$ is a ring, so it supports multiplication, but not every element has a multiplicative inverse (more on division in $K[X]$ later).

A polynomial is non-trivial if some coefficient of it is non-zero. Then, we define the degree of $P$, denoted $deg(P)$, to be the maximal $i$ such that $p_i \neq 0$. Observe that $deg(P+Q)\leq \max{deg(P),deg(Q)}$. It is natural to define the degree of the trivial polynomial to be $- \infty$. This way $deg(P Q) = \deg(P) + \deg(Q)$ always holds.

We say that $a \in K$ is a root of $P \in K[X]$ if $P(a)=0$ and say that $P$ has at most $d$ roots if there are at most $d$ elements in $K$ that are a root of $P$.

For example, consider the polynomial $P=2X-4$. It is clearly a polynomial of degree one and we all know that, over the rational field $K=\mathbb{Q}$, it has just one root (at $2$). Now, consider that same polynomial over the finite field $K=\mathbb{Z}_7$, which are just the integers modulo 7: i.e., $\{0,1,2,\dots,6\}$. Then, a quick check shows that $2$ is still the only root of $P$. In other words, the equation $2X=4 \pmod 7$ has exactly one solution: $2$.

Note that if instead of a field $K=\mathbb{Z}_7$ we chose the ring $K=\mathbb{Z}_{12}$, then the equation $2X=4 \pmod {12}$ would have 2 (!) solutions: $2$ (because $2\times 2 - 4 = 0$) and $8$ (because $2\times 8 - 4 = 16 - 4 = 12$, which is equal to 0 modulo 12). In other words, $P$ has more roots (two) than its degree (one)!

Proof

To prove the theorem, we prove an important claim:

Claim: If $deg(P)\geq 1$ and $P(a)=0$ then there exists polynomial $Q$ such that $P=(X-a)Q$ and $deg(Q)<deg(P)$.

The proof is by induction on $d$.

For the base case $d=1$, $P$’s coefficients are $p_0,p_1$ and $p_1\neq 0$. Let $a= -(p_0) (p_1)^{-1}$ (this is well defined since $K$ is a field) and let $Q=p_1$. Note that $Q$ is a non-trivial degree zero polynomial. It is easy to check that indeed $(X-a) Q = p_1 X + p_0 = P$.

For $d>1$, define a new polynomial $P’ = P - p_d X^{d-1} (X-a)$, note that $p_d$ is the largest coefficient of $P$. Lets make a few observations:

  1. The degree of $P’$ is smaller than $d$. This is because the $d$th coefficient of $p_d X^{d-1} (X-a)$ equals $p_d$, so it will cancel out.
  2. $P’$ has the property that $P’(a)=0$. This is because $P(a)=0$ and because $p_d X^{d-1} (X-a)$ also has a root at $a$.

Hence, we can apply the induction hypothesis on $P’$ to obtain that there exists $Q’$ such that $P’=(X-a)Q’$ and $deg(Q’) < d-1$.

Since:

\[P =P '+ p_d X^{d-1} (X-a)\]

Then, we can substitute $P’=(X-a)Q’$ and get:

\[P= (X-a)Q' + (X-a) p_d X^{d-1}\]

Hence, we have proved that:

\[P = (X-a)Q\ \text{with}\ Q=Q'+ p_d X^{d-1}\]

Since $deg(Q) \leq \max { deg(Q’), deg(p_d X^{d-1})}$ then $Q$ has degree at most $d-1$. This completes the proof of the claim.

Proof of the Theorem

The proof is by induction on $d$. For $d=0$, since $P$ is non-trivial, we have $p_0 \neq 0$ and hence $P$ has no roots (as it should be).

For $d=1$, we use the fact that $K$ is a field. The unique root of $P=p_0+p_1 X$ (for $p_1 \neq 0$) is the unique element $-p_0/p_1 = -(p_0) (p_1)^{-1}$. This follows from the uniqueness of the inverse for both addition and multiplication in a field (if $K$ were just a ring and not a field (e.g., $K=\mathbb{Z}_{12}$), then the inverse of $p_1$ may not exist or, more worrisome, may not be unique!).

For $d\geq 2$, we use an induction step. There are two cases. If $P$ has no roots, then we are done. Otherwise, let $a \in K$ be such that $P(a)=0$. Using the claim above, there exists a polynomial $Q$ of degree $<d$ such that $P=(X-a) Q$. Since $deg(Q)<d$ we use the induction hypothesis on $Q$. So $P$ can have at most $d-1$ roots from $Q$ and at most one more root (at $a$) from the degree one polynomial $(X-a)$.

Discussion

In the next posts, we will use this very useful fact about roots of polynomials over finite fields. First, we will use it as the foundation for secret sharing and then as the foundation for Zero Knowledge Proofs.

A significantly more general result about polynomials over a field views them as a special case of a unique factorization domain. This view exposes deep connections between the natural numbers, polynomials over a field, and the fundamental theorem of Arithmetic.

Acknowledgment. Thanks to Alin for helpful feedback on this post.

Please leave comments on Twitter.