algebra

Multivariable Polynomial Rings

Here is a truth that took me far too long to see: a polynomial in several variables is a description of a shape, and the ring of all such polynomials is the algebra of shapes.

The circle is not just a picture. It is the equation $x^2 + y^2 - 1 = 0$ . And once shapes are equations, you can add them, multiply them and factor them.

The core idea, in plain words

Start with a field of numbers $k$ (think of the real numbers $\mathbb{R}$ or the rationals $\mathbb{Q}$ ). The ring $k[x, y]$ consists of every finite sum of terms like

3x^2y - 5xy^3 + 7y + 2

with coefficients from $k$ and whole-number powers of $x$ and $y$ . More generally $k[x_1, \dots, x_n]$ has $n$ variables.

You add them by collecting like terms and multiply them by the distributive law, using $x^a x^b = x^{a+b}$ and $xy = yx$ . That makes $k[x_1, \dots, x_n]$ a Commutative Ring. It has no zero divisors, so it is an integral domain. But almost nothing is a unit: only the non-zero constants have inverses, because $\tfrac{1}{x}$ is not a polynomial.

Each term $x^a y^b$ is a monomial, and its degree is $a + b$ . The big shift from one variable to many is this: in $k[x]$ the monomials line up neatly by degree, but in $k[x, y]$ there is no single obvious way to say which of $x^2$ and $xy$ comes “first”. That small question has enormous consequences.

An everyday analogy: recipes with several ingredients

A one-variable polynomial is like a recipe with one ingredient in various amounts. A multivariable polynomial is a real recipe: so much flour, so much sugar, so much flour-times-sugar mixed together. Adding polynomials is combining recipes. Multiplying them is like doing one recipe on top of another: every ingredient of the first meets every ingredient of the second. And just as different recipes can give the same cake, different-looking polynomials can cut out the same shape.

A worked example: one variable divides cleanly, two don’t

In $k[x]$ you can always do long division: divide $f$ by $g$ and get a quotient and a remainder of smaller degree, and the answer is unique. That is why every ideal in $k[x]$ is generated by a single polynomial (its gcd).

In $k[x, y]$ things get stranger. Consider the ideal

I = (x, y) = \{\, x f + y g \;:\; f, g \in k[x, y] \,\}

which is every polynomial with zero constant term. Can one polynomial generate it? Suppose $I = (h)$ . Then $h$ divides both $x$ and $y$ . The only common divisors of $x$ and $y$ are constants, but a constant generates the whole ring, including $1$ , which is not in $I$ . Contradiction. This ideal needs two generators.

So the cosy one-variable world, where every ideal is one polynomial, is gone. What saves us is Hilbert’s Basis Theorem: every ideal of $k[x_1, \dots, x_n]$ is generated by finitely many polynomials. Not one, but never infinitely many. Any system of polynomial equations, however huge, can be replaced by a finite one.

And there’s a modern twist. If you fix a consistent way to order monomials (say “compare total degree, break ties by the power of $x$ ”), you can do a version of long division, and a special generating set called a Gröbner basis makes remainders unique again. That’s how computer algebra systems like Sage solve systems of polynomial equations.

A picture in words: equations cutting out shapes

Draw the plane. The polynomial $x^2 + y^2 - 1$ vanishes exactly on the unit circle. The polynomial $y - x$ vanishes on the diagonal line. Now:

The ideal generated by both, $(x^2 + y^2 - 1,\ y - x)$ , vanishes only where both are zero: the two points where the line crosses the circle, $\pm(\tfrac{1}{\sqrt{2}}, \tfrac{1}{\sqrt{2}})$ . Bigger ideal, smaller shape.
The product $(x^2 + y^2 - 1)(y - x)$ vanishes where either is zero: the circle and the line together. Multiplying polynomials takes the union of shapes.

graph LR
  A["More equations (bigger ideal)"] --> B["Fewer points (smaller shape)"]
  C["Multiply polynomials"] --> D["Union of shapes"]
  E["Add equations"] --> F["Intersection of shapes"]

This dictionary between algebra and geometry is called algebraic geometry, and Hilbert’s Nullstellensatz is the theorem that makes the dictionary exact (over the complex numbers).

Common misconceptions

“It’s just $k[x]$ with more letters.” No. Long division, single generators for ideals, and unique remainders all break when you add a second variable.
“Irreducible polynomials of degree 2 or more can always be factored over $\mathbb{C}$ .” True in one variable, false in many. $x^2 + y^2 - 1$ is irreducible even over the complex numbers. The circle doesn’t split into lines.
“Different polynomials mean different shapes.” $x$ and $x^2$ vanish on exactly the same line. Polynomials carry more information than their shapes: $x^2$ knows the line “twice”.
" $xy$ and $yx$ are different." In this ring the variables commute. (There are non-commutative polynomial rings too, but that’s another page.)

How it connects

Down to Rings and Commutative Rings: this is the most important example of a commutative ring after the integers.
Sideways to Fields: the coefficients come from a field, and adjoining a root of a polynomial is how new fields are built.
Up to Modules and to geometry, Topology and beyond.
Out to Math on Computers, where Gröbner bases are how machines solve equations.

The truth revealed: every equation is a shape, and every shape you can write down is an ideal in disguise.