Multiply $3x^3 + 2x + 1$ by $4x^2 + 5x + 1$, with integer coefficients, and the result is
$$12x^5 + 15x^4 + 11x^3 + 14x^2 + 7x + 1,$$
of degree $3 + 2 = 5$, as school algebra predicts. Now do the same multiplication with coefficients in $\mathbb{Z}_{12}$, reducing every coefficient modulo 12. The leading coefficient $3 \cdot 4 = 12$ becomes 0, and the product is
$$3x^4 + 11x^3 + 2x^2 + 7x + 1,$$
of degree 4. Worse, in the same setting $(3x)(4x^2 + 8) = 12x^3 + 24x = 0$: two nonzero polynomials have product zero. Both effects come from the coefficients. $\mathbb{Z}_{12}$ has zero divisors, and polynomials over it inherit them. So which rules of school algebra survive when the coefficients come from an arbitrary ring, and what decides it?
A polynomial ring $R[x]$ is the set of polynomials with coefficients in a ring $R$, added and multiplied by the familiar rules. The definition of a ring listed polynomial rings as an example without proof; here the proof is given.
First we say what $x$ is, by defining a polynomial as its list of coefficients, and prove that the lists form a ring. Next we find that the degree of a product behaves as expected exactly when the coefficients form an integral domain, and that decides which polynomials have inverses. Finally we substitute numbers for $x$, and see why a polynomial is not the same thing as its function.
This is the first article of Polynomial Arithmetic, the unit of the MTH 532 notes that corresponds to Chapter 4 of the course text. It covers the first half of §4.1 (Hungerford, 2014, Theorems 4.1 to 4.5).
What is $x$?
A polynomial is usually written $a_0 + a_1 x + a_2 x^2 + \cdots + a_n x^n$, and the question of what $x$ is tends to be skipped. It is not a number. Two polynomials are equal when their coefficients agree, not when they agree at every number, and the last section of this article shows that the two notions differ. Nor is it an element of the coefficient ring. The clean answer starts from the observation that a polynomial is determined by its list of coefficients. We take the list itself as the definition, and the letter $x$ becomes a name for one particular list.
Throughout, $R$ is a ring in the sense of the ring axioms; it need not be commutative or have an identity unless that is stated.
Definition. A polynomial with coefficients in $R$ is a sequence $(a_0, a_1, a_2, \ldots)$ of elements of $R$ in which only finitely many terms are nonzero. Two polynomials are equal when they are equal term by term. The set of all of them is written $R[x]$. For $f = (a_0, a_1, \ldots)$ and $g = (b_0, b_1, \ldots)$, define
$$f + g = (a_0 + b_0, \; a_1 + b_1, \; a_2 + b_2, \; \ldots), \qquad fg = (c_0, c_1, c_2, \ldots), \quad c_k = \sum_{i + j = k} a_i b_j = a_0 b_k + a_1 b_{k-1} + \cdots + a_k b_0.$$
The term $a_k$ is the coefficient of $x^k$, for reasons that are about to appear. Addition adds like coefficients. The product rule is what distributing and collecting powers would give: $x^i \cdot x^j$ contributes to $x^{i+j}$, so the coefficient of $x^k$ gathers every product $a_i b_j$ with $i + j = k$. The rule is called the convolution of the two sequences.
To see that the formulas are the familiar ones, take $1 - 3x + x^2$ and $2 + x + 4x^2$ in $\mathbb{Z}[x]$, that is, the sequences $(1, -3, 1, 0, \ldots)$ and $(2, 1, 4, 0, \ldots)$. The convolution gives
$$\begin{aligned} c_0 &= (1)(2) = 2, \\ c_1 &= (1)(1) + (-3)(2) = -5, \\ c_2 &= (1)(4) + (-3)(1) + (1)(2) = 3, \\ c_3 &= (-3)(4) + (1)(1) = -11, \\ c_4 &= (1)(4) = 4, \end{aligned}$$
and every later $c_k$ is 0. So the product is $4x^4 - 11x^3 + 3x^2 - 5x + 2$, which is what expanding the brackets gives.
The usual notation
Each element $a$ of $R$ is identified with the constant polynomial $(a, 0, 0, \ldots)$. The definitions give
$$(a, 0, 0, \ldots) + (b, 0, 0, \ldots) = (a + b, 0, 0, \ldots), \qquad (a, 0, 0, \ldots)(b, 0, 0, \ldots) = (ab, 0, 0, \ldots),$$
since the only possibly nonzero product in the convolution is $a_0 b_0$. So adding and multiplying constants in $R[x]$ is the same as adding and multiplying in $R$, and the identification makes $R$ a subring of $R[x]$.
Suppose now that $R$ has an identity 1. Let $x$ be the sequence $(0, 1, 0, 0, \ldots)$. Multiplying a polynomial by $x$ shifts its coefficients one place to the right: if $f = (a_0, a_1, \ldots)$, the coefficient of $x f$ in position $k \ge 1$ is $\sum_{i+j=k} x_i a_j = 1 \cdot a_{k-1}$, because the only nonzero term of $x$ is $x_1 = 1$, and the coefficient in position 0 is $x_0 a_0 = 0$. By induction, $x^k$ is the sequence with 1 in position $k$ and 0 elsewhere, and for $a \in R$ the product $a x^k$ is the sequence with $a$ in position $k$ and 0 elsewhere. A sequence whose terms vanish after position $n$ is therefore the sum
$$(a_0, a_1, \ldots, a_n, 0, \ldots) = a_0 + a_1 x + a_2 x^2 + \cdots + a_n x^n,$$
and since sequences are equal only when they agree term by term, the coefficients in such an expression are uniquely determined. This recovers the usual notation, together with the rule that two polynomials are equal exactly when their coefficients are. For example, $2x^3 - x + 5$ in $\mathbb{Z}[x]$ is the sequence $(5, -1, 0, 2, 0, 0, \ldots)$. Writing the variable $x$ as a list removes any mystery about it, at the price of a little notation that is never needed again.
When $R$ has no identity, the sequence $(0, 1, 0, \ldots)$ is not available, because 1 is not. Even so, the expression $a_0 + a_1 x + \cdots + a_n x^n$ is still used as a name for the sequence $(a_0, \ldots, a_n, 0, \ldots)$. The course text takes a different route to the same ring: it asserts that every ring $R$ sits inside a larger ring containing an element $x$ that commutes with $R$ and satisfies no relations, and gives the construction in an appendix (Hungerford, 2014, Theorem 4.1). Judson (2026, §17.1) uses the formal expressions directly, for commutative rings with identity.
Polynomials form a ring
Do the definitions obey the ring axioms? Addition works one coefficient at a time, so the real content is that convolution is associative and distributive. The idea: both bracketings of $fgh$ give, as the coefficient of $x^k$, the sum of $a_s b_t c_j$ over all $s + t + j = k$.
Theorem. Let $R$ be a ring. Then $R[x]$, with the addition and multiplication above, is a ring. In addition:
- $R[x]$ is commutative if and only if $R$ is commutative;
- $R[x]$ has an identity if and only if $R$ does, and then the identity of $R$ is the identity of $R[x]$.
Proof. Write $f = (a_k)$, $g = (b_k)$ and $h = (c_k)$, with $a_k = 0$ for $k > m$ and $b_k = 0$ for $k > n$.
Closure. The sum has $a_k + b_k = 0$ for $k > \max(m, n)$. In the product, the coefficient $\sum_{i+j=k} a_i b_j$ with $k > m + n$ is a sum of terms in which either $i > m$ or $j > n$, since $i \le m$ and $j \le n$ would give $k \le m + n$, so every term is 0. Each coefficient of the product is a finite sum of elements of $R$, so it lies in $R$. Hence $f + g$ and $fg$ are polynomials.
Addition. Addition is performed coordinate by coordinate, so its associative and commutative laws follow from those of $R$ in each coordinate. The zero polynomial $(0, 0, \ldots)$ is an additive identity, and $(-a_0, -a_1, \ldots)$, which has only finitely many nonzero terms, is the negative of $f$.
Associativity of multiplication. The coefficient of $x^k$ in $(fg)h$ is
$$\sum_{l + j = k} \Big( \sum_{s + t = l} a_s b_t \Big) c_j = \sum_{s + t + j = k} (a_s b_t) c_j,$$
using the distributive law in $R$ to multiply out each inner sum, and then collecting the terms into one sum over all triples of nonnegative indices with $s + t + j = k$. In the same way, the coefficient of $x^k$ in $f(gh)$ is
$$\sum_{s + l = k} a_s \Big( \sum_{t + j = l} b_t c_j \Big) = \sum_{s + t + j = k} a_s (b_t c_j).$$
The associative law of $R$ makes the two sums equal term by term.
Distributive laws. The coefficient of $x^k$ in $f(g + h)$ is $\sum_{i+j=k} a_i (b_j + c_j) = \sum_{i+j=k} a_i b_j + \sum_{i+j=k} a_i c_j$, which is the coefficient of $x^k$ in $fg + fh$. The other distributive law is the same with the factors in the other order.
So $R[x]$ is a ring.
(1) If $R$ is commutative, then $\sum_{i+j=k} a_i b_j = \sum_{j+i=k} b_j a_i$, so $fg = gf$. Conversely, if $R[x]$ is commutative, then in particular constant polynomials commute, and multiplying constants is multiplying in $R$, so $R$ is commutative.
(2) If $R$ has identity 1, the constant polynomial $1 = (1, 0, 0, \ldots)$ satisfies $1 \cdot f = f$, since the coefficient of $x^k$ in $1 \cdot f$ is $1 \cdot a_k$, the only term with a nonzero first factor; similarly $f \cdot 1 = f$. Conversely, suppose $e = (e_0, e_1, \ldots)$ is an identity of $R[x]$. For any $a \in R$, the constant term of $e \cdot a$ is $e_0 a$ and that of $a \cdot e$ is $a e_0$; since $e \cdot a = a = a \cdot e$, both equal $a$. So $e_0$ is an identity of $R$. $\blacksquare$
The script for this article checks the ring laws by brute force in one small case: in $\mathbb{Z}_3[x]$ there are 27 polynomials of degree at most 2, and multiplication is associative, distributive and commutative on all $27^3 = 19{,}683$ triples of them. The proof, not the check, is what covers polynomials of every degree. Judson (2026, Theorem 17.1.3) proves the ring axioms for a commutative ring with identity.
So $\mathbb{Z}_{12}[x]$ is a perfectly good ring. Why, then, did the degree drop in the opening example?
Degree
Definition. Let $f = a_0 + a_1 x + \cdots + a_n x^n$ be a nonzero polynomial with $a_n \neq 0$. The integer $n$ is the degree of $f$, written $\deg f$, and $a_n$ is its leading coefficient. When $R$ has an identity, $f$ is monic if its leading coefficient is 1. The zero polynomial has no degree.
The nonzero constants are exactly the polynomials of degree 0. Leaving the zero polynomial without a degree avoids a special case in every formula below; some books assign it degree $-\infty$ instead, with the conventions $-\infty + n = -\infty$ and $-\infty < n$, which lets the formulas hold without exception (Judson, 2026, §17.1, takes this route).
The answer is that only one product of coefficients can reach the top position of a product, the product of the two leading coefficients. If that one product is 0, the top term vanishes.
Proposition. Let $f$ and $g$ be nonzero polynomials in $R[x]$, of degrees $m$ and $n$, with leading coefficients $a_m$ and $b_n$.
- Either $f + g = 0$ or $\deg(f + g) \le \max(m, n)$.
- The coefficient of $x^{m+n}$ in $fg$ is $a_m b_n$, and every higher coefficient is 0. Hence either $fg = 0$ or $\deg(fg) \le m + n$, with equality exactly when $a_m b_n \neq 0$.
Proof. (1) Every coefficient of $f + g$ beyond position $\max(m, n)$ is $0 + 0$. (2) The closure argument above showed that the coefficients of $fg$ beyond position $m + n$ vanish. In position $m + n$, the sum $\sum_{i+j = m+n} a_i b_j$ has a term with $i > m$ or $j > n$, which is 0, except for the single term $i = m$, $j = n$. So the coefficient is $a_m b_n$, and the degree is $m + n$ exactly when it is nonzero. $\blacksquare$
In (1), the inequality is an equality whenever $m \neq n$, since then the leading coefficient of the longer polynomial survives. It can be strict only when the degrees are equal and the leading coefficients add to 0. In (2), the product of the leading coefficients decides everything, and that is where the coefficient ring matters. If the coefficient ring has no zero divisors, that product is never 0.
Theorem. Let $R$ be an integral domain. If $f$ and $g$ are nonzero polynomials in $R[x]$, then $fg \neq 0$ and
$$\deg(fg) = \deg f + \deg g.$$
Consequently $R[x]$ is an integral domain. Conversely, if $R[x]$ is an integral domain, so is $R$.
Proof. The leading coefficients $a_m$ and $b_n$ are nonzero, and an integral domain has no zero divisors, so $a_m b_n \neq 0$. By the proposition, $fg$ is nonzero of degree $m + n$. An integral domain is a commutative ring with identity $1 \neq 0$ and no zero divisors; by the previous theorem $R[x]$ is commutative with identity $1 \neq 0$, and the product of nonzero polynomials has just been shown to be nonzero. For the converse, $R$ is a subring of $R[x]$ containing its identity, and a product of nonzero constants is nonzero because it is a product of nonzero elements of $R[x]$. $\blacksquare$
This is Hungerford's Theorem 4.2 and Corollary 4.3, and Judson's Proposition 17.1.4. The opening example shows both halves of the picture. In $\mathbb{Z}[x]$, an integral domain, the degrees add: $3 + 2 = 5$. In $\mathbb{Z}_{12}[x]$ they do not, because the leading coefficients 3 and 4 are zero divisors in $\mathbb{Z}_{12}$, and the nonzero polynomials $3x$ and $4x^2 + 8$ are zero divisors in $\mathbb{Z}_{12}[x]$.
Check your understanding. $\mathbb{Z}_{12}$ is not an integral domain. Can the product of $5x^2 + 1$ with some nonzero polynomial in $\mathbb{Z}_{12}[x]$ have degree less than expected?
It cannot. By the proposition, only the product $5c$ matters, where $c \neq 0$ is the leading coefficient of the other factor. Since $5 \cdot 5 = 25 = 1$ in $\mathbb{Z}_{12}$, $5c = 0$ would give $c = 5 \cdot 5c = 0$. For instance $(5x^2 + 1)(3x + 2) = 3x^3 + 10x^2 + 3x + 2$ keeps degree 3. A leading coefficient 4 would behave differently, since $4 \cdot 3 = 0$. What matters is the leading coefficient, not the whole ring, and the next article builds its division algorithm on this observation.
The formula is the polynomial counterpart of a fact about integers that the laws of divisibility used repeatedly: if $b = ak$ with $b \ne 0$, then $\lvert a \rvert \le \lvert b \rvert$. For polynomials over an integral domain, if $g = fh$ with $g \neq 0$, then $\deg f \le \deg f + \deg h = \deg g$. Degree plays the part of absolute value, and the next two articles lean on it at every step.
Units
The units of a ring are its elements with multiplicative inverses. In $\mathbb{Z}$ they are $\pm 1$, and in a field they are all the nonzero elements. Polynomial rings over integral domains behave like $\mathbb{Z}$: almost nothing is invertible. The reason is the degree formula. Multiplying never lowers the degree, and the identity 1 has degree 0, so a polynomial of positive degree can never multiply to 1.
Theorem. Let $R$ be an integral domain. A polynomial $f \in R[x]$ is a unit of $R[x]$ if and only if $f$ is a constant polynomial that is a unit of $R$.
Proof. If $f$ is a constant $u$ that is a unit of $R$, with $uv = 1$ in $R$, then the constant polynomial $v$ is its inverse in $R[x]$. Conversely, suppose $fg = 1$ for some $g \in R[x]$. Neither $f$ nor $g$ is zero, since $1 \neq 0$. By the degree formula, $\deg f + \deg g = \deg 1 = 0$. Degrees are nonnegative, so both are 0: $f$ and $g$ are nonzero constants $u$ and $v$ with $uv = 1$, and $u$ is a unit of $R$. $\blacksquare$
Corollaries.
- The units of $\mathbb{Z}[x]$ are $1$ and $-1$.
- If $F$ is a field, the units of $F[x]$ are the nonzero constants.
- If $F$ is a field, $F[x]$ is an integral domain but not a field. In particular $x$ has no inverse.
The third corollary follows from the second, because $x$ has degree 1 and so is not a constant. It can also be seen directly: whatever $g$ is, the product $xg$ has constant term 0 and so cannot equal 1. This is Hungerford's Corollary 4.5. The ring $\mathbb{Q}[x]$ is therefore much more like $\mathbb{Z}$ than like $\mathbb{Q}$: an integral domain whose only units are a small, well-understood set. That resemblance is the theme of the unit.
When the coefficients have zero divisors
What if the coefficients are not an integral domain? The degree argument breaks, and so does the conclusion. Return to $\mathbb{Z}_{12}$, where the opening example lost a degree. There
$$(5 + 6x)^2 = 25 + 60x + 36x^2 = 1,$$
because $25 = 1$ while 60 and 36 are 0 in $\mathbb{Z}_{12}$. So $5 + 6x$ is a unit of degree 1, its own inverse, and the degrees of a polynomial and its inverse add to 2, not 0. Two facts about the coefficients combine here. The constant term 5 is a unit of $\mathbb{Z}_{12}$, and the coefficient 6 has $6^2 = 0$, so the term $6x$ is "small" in a sense that has no counterpart in an integral domain. Over $\mathbb{Z}$ the same square is $25 + 60x + 36x^2$, of degree 2 as the theorem predicts, and by the theorem $5 + 6x$ has no inverse in $\mathbb{Z}[x]$.
Polynomials and power series
A second contrast removes the finiteness condition instead of the integral domain. A formal power series over $R$ is a sequence $(a_0, a_1, \ldots)$ with no finiteness condition, added and multiplied by the same rules; each coefficient of a product is still a finite sum. The formal power series form a ring $R[[x]]$ containing $R[x]$ (the proof above carries over word for word, without the closure step). In $\mathbb{Z}[[x]]$, the polynomial $1 - 2x$ has the inverse
$$1 + 2x + 4x^2 + 8x^3 + \cdots = \sum_{k \ge 0} 2^k x^k,$$
since multiplying by $1 - 2x$ cancels every term after the first: the coefficient of $x^k$ for $k \ge 1$ is $2^k - 2 \cdot 2^{k-1} = 0$. Any polynomial truncation of the series falls short. The script computes $(1 - 2x)(1 + 2x + \cdots + 2^7 x^7) = 1 - 256x^8$, which is not 1, and no polynomial does better, because $1 - 2x$ is not a unit of $\mathbb{Z}[x]$. Power series have more units because they are allowed to go on forever. Polynomials are not, and the degree formula is the bookkeeping that enforces it. The geometric-series identity $1/(1 - r) = 1 + r + r^2 + \cdots$ holds here as an identity of formal series, with no question of convergence.
Check your understanding. $1 - 2x$ has an inverse in $\mathbb{Z}[[x]]$. Does $2 - x$?
It does not, and the obstacle is the constant term rather than the degree. The constant term of a product of series is the product of the constant terms, so an inverse $b_0 + b_1 x + \cdots$ would need $2 b_0 = 1$ in $\mathbb{Z}$. Over $\mathbb{Q}$, $2 - x$ has the inverse $\tfrac{1}{2} + \tfrac{1}{4}x + \tfrac{1}{8}x^2 + \cdots$, whose truncation at $x^7$ leaves $1 - \tfrac{1}{256}x^8$ (script).
Evaluation
So far $x$ has been a formal symbol. When can we substitute a number for it and still compute as usual?
Let $R$ be commutative. For $f = a_0 + a_1 x + \cdots + a_n x^n$ in $R[x]$ and $c \in R$, the value of $f$ at $c$ is the element
$$f(c) = a_0 + a_1 c + a_2 c^2 + \cdots + a_n c^n$$
of $R$, obtained by replacing the indeterminate by $c$ and computing in $R$. Each polynomial thus defines a function $R \to R$, $c \mapsto f(c)$, called the polynomial function of $f$. Evaluation is useful because it respects the ring operations. The idea of the proof: multiplying out $f(c)\,g(c)$ and collecting by powers of $c$ gives $(fg)(c)$, provided $c^i$ can be moved past $b_j$.
Theorem. Let $R$ be a commutative ring, $f, g \in R[x]$ and $c \in R$. Then
$$(f + g)(c) = f(c) + g(c) \qquad \text{and} \qquad (fg)(c) = f(c)\, g(c).$$
The constant polynomial $a$ has value $a$ at every $c$.
Proof. Write $f = \sum a_i x^i$ and $g = \sum b_j x^j$. The first identity is $\sum (a_k + b_k) c^k = \sum a_k c^k + \sum b_k c^k$, which is the distributive law in $R$. For the second, the coefficient of $x^k$ in $fg$ is $\sum_{i+j=k} a_i b_j$, so
$$(fg)(c) = \sum_k \Big( \sum_{i+j=k} a_i b_j \Big) c^k = \sum_{i, j} a_i b_j c^{i+j} = \sum_{i,j} (a_i c^i)(b_j c^j) = \Big( \sum_i a_i c^i \Big) \Big( \sum_j b_j c^j \Big) = f(c)\, g(c).$$
The third equality uses $c^{i+j} = c^i c^j$ and commutativity, to move $c^i$ past $b_j$; the others use the distributive law to expand and regroup finite sums. The statement about constants is immediate. $\blacksquare$
For example, with $f = x^2 - 3x + 1$ and $g = 2x + 5$ in $\mathbb{Z}[x]$ and $c = 4$: $f(4) = 5$ and $g(4) = 13$, while $(f + g)(4) = 18$ and $(fg)(4) = 65 = 5 \cdot 13$, computed by the script from the expanded product. In the language of a later unit, the theorem says that evaluation at $c$ is a ring homomorphism from $R[x]$ to $R$ (Judson, 2026, Theorem 17.1.5). It is what makes the remainder and factor theorems of Irreducible Polynomials and Roots work.
Commutativity is used, in the one step that moves $c^i$ past $b_j$, and without it the theorem fails. In $M_2(\mathbb{R})[x]$, the polynomials over the ring of $2 \times 2$ real matrices, the indeterminate commutes with every coefficient, so the constant polynomial $A$ times the polynomial $x$ is $Ax$. Take
$$A = \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}, \qquad B = \begin{pmatrix} 0 & 0 \\ 1 & 0 \end{pmatrix}.$$
Evaluating $Ax$ at $B$ (coefficients written on the left) gives $AB = \begin{pmatrix} 1 & 0 \\ 0 & 0 \end{pmatrix}$, while the product of the values of $x$ and $A$, in that order, is $BA = \begin{pmatrix} 0 & 0 \\ 0 & 1 \end{pmatrix}$. The two differ, so substituting a matrix for $x$ does not respect products. For this reason the theory of polynomial functions, and everything built on roots, is developed over commutative rings.
A polynomial is not its function
The first section claimed that a polynomial and its function are different things. Here is the evidence: over a finite ring, different polynomials can define the same function. Counting shows that this must happen. There are only $5^5 = 3125$ functions from $\mathbb{Z}_5$ to itself, but infinitely many polynomials in $\mathbb{Z}_5[x]$, one for each finite list of coefficients, so infinitely many polynomials share each function that occurs.
For a pair that can be seen, take $x^6$ and $x^2$ in $\mathbb{Z}_5[x]$. They are different polynomials, of degrees 6 and 2. Yet their values agree at every element of $\mathbb{Z}_5$:
| $c$ | 0 | 1 | 2 | 3 | 4 |
|---|---|---|---|---|---|
| $c^2$ in $\mathbb{Z}_5$ | 0 | 1 | 4 | 4 | 1 |
| $c^6$ in $\mathbb{Z}_5$ | 0 | 1 | 4 | 4 | 1 |
The reason is Fermat's little theorem: $c^4 = 1$ for every nonzero $c$ in $\mathbb{Z}_5$, so $c^6 = c^4 c^2 = c^2$, and at $c = 0$ both values are 0. So $x^6$ and $x^2$ define the same function on $\mathbb{Z}_5$, although they are different elements of $\mathbb{Z}_5[x]$, and their difference $x^6 - x^2$ is a nonzero polynomial whose function is 0.
Check your understanding. $x^6 - x^2$ vanishes at all five elements of $\mathbb{Z}_5$. Does $x^3 - x$?
It does not. Its values at $0, 1, 2, 3, 4$ are $0, 0, 1, 4, 0$; for example $2^3 - 2 = 6 = 1$ in $\mathbb{Z}_5$. In fact no nonzero cubic could vanish on all of $\mathbb{Z}_5$, since over a field it has at most three roots, as Irreducible Polynomials and Roots proves.
Pairs like $x^6$ and $x^2$ are the reason for insisting that $x$ is an indeterminate and that polynomials are equal only when their coefficients are. Over $\mathbb{Q}$ the same two polynomials are far apart: at 2 they take the values 64 and 4. Over an infinite field the two notions of equality agree: distinct polynomials always define distinct functions. The proof needs the fact that a nonzero polynomial of degree $n$ has at most $n$ roots, and it is given in Irreducible Polynomials and Roots. The polynomial $x^p - x$, which vanishes on $\mathbb{Z}_p$ by the same theorem of Fermat, returns there too, factored completely into linear factors.
Several indeterminates
Since $R[x]$ is a ring, it can serve as the coefficient ring of a new polynomial ring $R[x][y]$, whose elements are polynomials in $y$ with coefficients that are polynomials in $x$. Collecting terms writes each element as a finite sum of terms $a_{ij} x^i y^j$, and the ring is written $R[x, y]$. The theorems above apply at each stage: if $R$ is an integral domain, so is $R[x]$, and therefore so is $R[x][y]$, and the same holds for any number of indeterminates.
The arithmetic of the coefficients still shows through. In $\mathbb{Z}[x, y]$, $(x + y)^3 = x^3 + 3x^2 y + 3x y^2 + y^3$. In $\mathbb{Z}_3[x, y]$ the middle coefficients are $3 = 0$, and the identity becomes $(x + y)^3 = x^3 + y^3$. Rings in more than one indeterminate are not the subject of this unit; one difference from $F[x]$ appears in the unit on ideals, where Judson (2026, Example 17.3.11) shows that $F[x, y]$ has an ideal that is not generated by a single element.
Where it leads
- With coefficients in a field, polynomials can be divided with a remainder of smaller degree, and the theory of greatest common divisors carries over from the integers (Polynomial Division over a Field).
- Irreducible polynomials, which play the part of primes, and the connection between roots and linear factors are the subject of Irreducible Polynomials and Roots.
- The quotient rings $F[x]/(p(x))$ built from an irreducible $p(x)$ are new fields (Building Fields from Irreducible Polynomials), and evaluation at a number is the leading example of a ring homomorphism.
What we have learned
A polynomial over $R$ is its list of coefficients, and the lists form a ring $R[x]$. Everything about degree comes down to the product of the leading coefficients. Over an integral domain it is never 0, so degrees add, $R[x]$ is an integral domain, and only the constant units of $R$ are invertible. Over $\mathbb{Z}_{12}$ a degree drops from 5 to 4, and the polynomial $5 + 6x$ is its own inverse. Evaluation respects sums and products when the coefficients commute, yet a polynomial is more than its function, as $x^6$ and $x^2$ on $\mathbb{Z}_5$ show.
The degree formula makes $F[x]$ look like $\mathbb{Z}$, with degree in place of absolute value. Does the resemblance go as far as division with remainder, greatest common divisors and Bezout's identity? That is Polynomial Division over a Field.
References
- Hungerford, T. W. (2014). Abstract Algebra: An Introduction (3rd ed.), §4.1. Brooks/Cole, Cengage Learning. The course text: the existence of $R[x]$ (Theorem 4.1, constructed in Appendix G), the degree of a product (Theorem 4.2), $R[x]$ an integral domain (Corollary 4.3), the inequality for degrees (Corollary 4.4), and the units of $R[x]$ (Corollary 4.5). record · 44 pages
- Judson, T. W. (2026). Abstract Algebra: Theory and Applications (annual edition, August 2026), §17.1 (polynomial rings: Theorem 17.1.3, Proposition 17.1.4, Theorem 17.1.5 on evaluation) and Example 17.3.11. Free online at judsonbooks.org. record · 44 pages