You see x^{-1} in an expression and something in your brain shrugs: that is still x, still a letter raised to a power, still algebra — surely it is a polynomial, or at least a polynomial-like thing? You see \sqrt{x} and your brain shrugs the same way. Short symbol, only x, sits in the same chapters as x^2 + 1. How different can it be?
Different enough that every major theorem about polynomials stops working. A polynomial has a surgical definition: every exponent of the variable must be a non-negative integer — 0, 1, 2, 3, \dots and nothing else. x^{-1} fails because -1 is negative. \sqrt{x} = x^{1/2} fails because 1/2 is fractional. Knowing where that line is matters because polynomial theorems do not apply to expressions that have sneaked across it.
The rule, precisely
For a \cdot x^k to count as a polynomial term, k must be a non-negative integer: 0, 1, 2, 3, \dots. No other k works. The coefficient a can be anything real — positive, negative, fractional, even irrational like \sqrt{2} or \pi — but the exponent is policed.
What fails and what those exponents produce instead:
- Negative integers (k = -1, -2, \dots): rational functions like 1/x, 1/x^2. The variable sits in a denominator.
- Fractional exponents (k = 1/2, 3/4, \dots): radicals or rational-exponent expressions like \sqrt{x}, x^{5/3}. The variable sits under a root.
- Irrational exponents (k = \pi, \sqrt{2}, \dots): transcendental-style expressions like x^\pi, with no finite algebraic description in general.
A polynomial must pass the rule on every term. One bad term poisons the whole expression.
Why x^{-1} fails
Rewrite it: x^{-1} = 1/x. There is an x in the denominator, and polynomials — by definition — cannot have the variable in a denominator. Every polynomial term is a coefficient times a non-negative integer power of x; "divided by x" is not any such power.
The class of expressions that you get by allowing the variable in denominators is called rational functions — ratios of polynomials, p(x)/q(x). So 1/x is the ratio of the polynomial 1 and the polynomial x. Every polynomial is a rational function (divide by 1), but not every rational function is a polynomial. 1/x is the simplest counter-example.
Why \sqrt{x} fails
Rewrite it: \sqrt{x} = x^{1/2}. The exponent 1/2 is a fraction. That alone closes the case.
There is a deeper reason the rule is the rule. Almost every polynomial theorem assumes integer exponents so that counting works. "A degree-n polynomial has at most n real roots" relies on n being a whole number you can count. "The derivative of a degree-n polynomial has degree n-1" relies on n-1 still being a non-negative integer. Let fractional exponents in and you cannot answer "what is the degree of \sqrt{x}?" in any way that keeps the theorems working.
Expressions with fractional exponents are called radicals or, more broadly, algebraic functions.
So what ARE x^{-1} and \sqrt{x}?
Both are algebraic in the loose sense — built from arithmetic and roots — but each sits in its own named class.
- x^{-1} is a rational function: a ratio of polynomials. Rational functions pick up a whole theory of their own (partial fractions, asymptotes, removable singularities) that polynomials do not need.
- \sqrt{x} is an algebraic function: a function that satisfies a polynomial equation. Here y^2 = x, so y = \sqrt{x} solves a polynomial in y with polynomial coefficients in x.
Both belong to "algebra" in the broad sense. Neither is a polynomial in the strict sense.
The hierarchy
Four nested circles, each strictly bigger than the one before.
- Polynomials. Non-negative integer exponents. 3x^2 + 5, x^{100}, -7.
- Rational functions. Ratios of polynomials. Implicitly allow negative exponents. 1/x, (x^2 + 1)/(x - 3).
- Algebraic functions. Solutions of polynomial equations in y with polynomial coefficients in x. Implicitly allow fractional exponents. \sqrt{x}, x^{2/3}.
- Transcendental functions. \sin x, \cos x, e^x, \ln x, x^\pi.
A polynomial is a rational function; a rational function is algebraic; an algebraic function is not transcendental but transcendentals are the outermost ring. x^{-1} and \sqrt{x} live in rings 2 and 3 — close to polynomials, not inside.
Why polynomial theorems fail for these other classes
Being precise about what is a polynomial lets you know which theorems apply.
- "A polynomial of degree n has at most n real roots." What is the degree of \sqrt{x}? There is no coherent answer — 1/2 is not the kind of number that fits the theorem. It simply does not apply.
- "Polynomials are continuous everywhere." But 1/x is famously not continuous at x = 0 — it blows up. Assume 1/x is a polynomial and you will also assume it is continuous everywhere, and you will be wrong.
- "Polynomials expand by the binomial theorem for integer n." (x+y)^3 expands cleanly because 3 is a non-negative integer. (x+y)^{1/2} needs an infinite series and convergence conditions — not a finite polynomial expansion.
Each theorem has "non-negative integer exponents" baked into its statement. Step outside and the theorem breaks silently — you will apply it, get an answer, and be wrong.
Common misclassifications — these are NOT polynomials
Students often suspect these are polynomials because they look short or ordinary. They are not.
- 1/x: rational function. Negative exponent when written as x^{-1}.
- 2x + 1/x: rational function. The 2x term is fine; the 1/x term poisons the whole expression. One bad term is enough.
- \sqrt{x} + 3: algebraic function. The 3 is fine; \sqrt{x} breaks the rule.
- x^{2.5}: the exponent 2.5 is neither a non-negative integer nor anything nicer. Algebraic (it equals x^{5/2}).
- x^\pi: transcendental in general. \pi is irrational, not even rational, let alone a non-negative integer.
Common correct classifications — these ARE polynomials
Students sometimes doubt these, but each is a legitimate polynomial.
- x^0 = 1: yes. Exponent 0 is a non-negative integer. Degree 0.
- (x+1)(x-3): yes. When expanded: x^2 - 2x - 3. All integer exponents.
- 5: yes. Write it as 5x^0. Constant polynomial.
- -x + 7: yes. Exponents 1 and 0. Linear polynomial.
- x^{100}: yes. 100 is a non-negative integer. Very large degree, still valid.
Large degree does not disqualify an expression. Fractional or negative exponents do.
What makes polynomials special
The restriction to non-negative integer exponents is what gives polynomials all their good behaviour in one package. They are the simplest class that has:
- Closure under +, -, and \times — add, subtract, or multiply polynomials and you get a polynomial.
- A well-defined degree — one integer summarising how large the expression gets for large x.
- A fixed bound on roots — at most n real roots for degree n.
- Continuity and differentiability everywhere — no blow-ups, no corners.
- Direct formulas for manipulation — binomial theorem, synthetic division, factor theorem.
Relax the rules and you lose guarantees. Allow division (rational functions) and you gain \div but lose the fixed root bound and continuity everywhere. Allow roots (algebraic functions) and you gain \sqrt{\phantom{x}} but lose closure under polynomial expansion.
Recognition quick drill
Classify each:
- x^3 + x^2 + 1 — polynomial. Exponents 3, 2, 0.
- 2/x — rational function, not polynomial. Equals 2x^{-1}, negative exponent.
- 5\sqrt{x} — algebraic, not polynomial. Equals 5x^{1/2}, fractional exponent.
- (x^2 + 1)/(x - 2) — rational function, not polynomial. Denominator contains x.
- x^2 \cdot x^{-2} = 1 — technically simplifies to the constant 1, which is a polynomial of degree 0. As written, though, the pre-simplification form suggests a rational function.
- e^x — transcendental, not polynomial. The variable is in the exponent.
If you can do this drill without hesitation, you have internalised the boundary.
The closing rule
If the exponent on x in any term is not a non-negative integer, the expression is not a polynomial. No matter how algebraic it feels, how short the symbol looks, or how natural it seems. The rule is strict because the theorems built on top of it are strict — step outside the definition and you step outside the theorems.