Proving Leading-Term End Behavior

Why the Leading Term Eventually Controls

A polynomial may contain many terms, and several of them can have a substantial effect when the input is moderate. But as the absolute value of the input grows, the highest power grows faster than every lower power. This suggests that the leading term determines a polynomial’s eventual sign and growth.

The idea can be seen by factoring the leading term out of the polynomial. The remaining factor is \(1\) plus terms involving negative powers of the input. Those terms can be bounded and shown to become small. The result is a proof, not just a pattern observed from a few values.

Factoring Out the Leading Term

For \(x\ne 0\), factor the leading term from \(P(x)\):

Each term after \(1\) contains a negative power of \(x\). The challenge is to show that all of them together remain small, not merely that each one becomes small on its own.

A Bound for All the Lower-Degree Terms

For each lower-degree coefficient, divide by the leading coefficient and define the nonnegative number

When \(|x|\ge 1\), each exponent \(n-i\) in the lower-degree terms is at least \(1\). Consequently, \(|x|^{n-i}\ge |x|\), so

This is the key estimate: the total relative contribution of the lower-degree terms is at most \(S/|x|\). In particular, if \(|x|>\max(1,2S)\), that contribution has absolute value less than \(1/2\).

Why the Bound Proves the Theorem

Write the factored polynomial as \(P(x)=a_nx^n(1+E(x))\), where \(E(x)\) is the sum of the lower-degree terms inside the parentheses. The estimate above gives \(|E(x)|\le S/|x|\) for \(|x|\ge 1\).

Choose \(R=\max(1,2S)\). If \(|x|>R\), then \(|E(x)|<1/2\), and therefore

The factor \(1+E(x)\) is positive. Thus \(P(x)=a_nx^n(1+E(x))\) has the same sign as \(a_nx^n\). The same bounds give a comparison of magnitudes:

These inequalities prove that the lower-degree terms cannot eventually reverse the sign or overwhelm the leading term. They also justify the growth comparison. More precisely, for \(|x|\ge1\),

Given any \(\varepsilon>0\), taking \(|x|>\max(1,S/\varepsilon)\) makes the right-hand side less than \(\varepsilon\). This proves that \(P(x)/(a_nx^n)\) approaches \(1\), as claimed.

Worked Examples

A Common Mistake: Ignoring the Direction of the Input

A frequent error is to say that an odd power is positive at both ends because its absolute value grows. But the sign of an odd power depends on the sign of its input. For example, the incorrect calculation \((-7)^3=343\) would lead to the wrong sign for \(-x^3+2x+1\) when \(x\) is negative.

The correction is to keep magnitude and sign separate: \((-7)^3=-343\), so \(-(-7)^3=343\). The leading term is positive at the negative end in this example, and the lower-degree terms do not change that eventual sign. For an even degree, \(x^n\) is positive at both ends; for an odd degree, it is negative at the negative end and positive at the positive end. The leading coefficient must also be included when determining the sign.

Why the Proof Matters

Looking at a polynomial’s terms gives a useful prediction, but a few large inputs cannot establish that the prediction holds for every sufficiently large input. The bound provides a threshold: once \(|x|\) exceeds it, the sum of all lower-degree terms is guaranteed to be too small relative to the leading term to change the sign. The same argument shows that the polynomial’s magnitude is comparable to the magnitude of its leading term.

This is the precise basis for polynomial end behavior. It also helps distinguish a statement about eventual behavior from a claim that a leading term controls the value everywhere. At moderate inputs, the lower-degree terms can still have a substantial effect.