Rational Integrands and Dominant Powers

Why Dominant Powers Decide Convergence

A rational function is a quotient of polynomials. Its formula may contain many terms, but near infinity the highest powers dominate. Near a finite denominator root, the factors that vanish there determine the strength of the singularity. These observations reduce many improper-integral questions to the familiar integrals of powers.

The order of the analysis matters. First cancel any common polynomial factors, then identify the behavior at each endpoint or singularity. A convergence conclusion about one part of an interval does not settle what happens at another.

Definition: A rational function has the form \(f(x)=P(x)/Q(x)\), where \(P\) and \(Q\) are polynomials and \(Q\) is not the zero polynomial. A point where the displayed denominator vanishes is a possible singularity; after cancellation, it is a genuine pole only if a denominator factor remains.

The Degree Test at Infinity

Let \(P\) and \(Q\) be nonzero polynomials with degrees \(n\) and \(m\), and leading coefficients \(a_n\) and \(b_m\). As \(x\) tends to positive infinity, their leading terms give

\frac{P(x)}{Q(x)} \sim \frac{a_n}{b_m}x^{n-m}.

Here \(f(x)\sim g(x)\) means that \(f(x)/g(x)\) tends to \(1\). The nonzero constant \(a_n/b_m\) does not change whether the corresponding power has a convergent integral. Thus the key exponent is the degree difference \(n-m\).

Theorem: For a nonzero rational function \(P/Q\), its improper integral over a tail \([A,\infty)\), with \(A\) chosen beyond any real poles, converges if and only if \(n-m<-1\), or equivalently \(m\ge n+2\). If \(n-m\ge -1\), the tail diverges. The same criterion applies at negative infinity, using the magnitude of \(x\).

This statement concerns an ordinary improper integral, defined using separate one-sided limits where needed. A rational tail has an eventually constant sign because its leading term is nonzero. Consequently, a divergent power-like tail cannot become convergent through sign cancellation. A Cauchy principal value, which can pair contributions from opposite sides, is a different notion.

Why the Degree Test Works

The leading-term calculation supplies an asymptotic comparison, and the power-integral threshold supplies the conclusion. The reasoning can be made explicit as follows.

Compare leading terms.
Dividing numerator and denominator by their highest powers gives \(f(x)/((a_n/b_m)x^{n-m})\to 1\) as \(x\to\infty\).

Use a positive comparison.
For sufficiently large \(x\), the magnitude of \(f(x)\) is bounded above and below by positive constant multiples of \(x^{n-m}\).

Apply the power threshold.
The integral \(\int_A^\infty x^d\,dx\) converges exactly when \(d<-1\). Taking \(d=n-m\) yields the criterion.

For completeness, if \(d\ne -1\), then

\int_A^R x^d\,dx = \frac{R^{d+1}-A^{d+1}}{d+1}.

As \(R\to\infty\), this has a finite limit exactly when \(d+1<0\). If \(d=-1\), the integral is \(\ln R-\ln A\), which diverges. These calculations establish the power threshold used in the degree test.

Worked Example: A Convergent Rational Tail

Determine whether \(\int_1^\infty \frac{2x+1}{x^3+x}\,dx\) converges.

The numerator has degree \(1\), and the denominator has degree \(3\). The predicted dominant power is \(x^{1-3}=x^{-2}\). Check the comparison by calculating its ratio:

\frac{\dfrac{2x+1}{x^3+x}}{x^{-2}} = \frac{(2x+1)x^2}{x^3+x} = \frac{2x^3+x^2}{x^3+x} \longrightarrow 2.

The limit is positive and finite, so limit comparison with \(x^{-2}\) applies. The comparison integral is

\int_1^\infty x^{-2}\,dx = \lim_{R\to\infty}\left[-\frac{1}{x}\right]_1^R = \lim_{R\to\infty}\left(1-\frac{1}{R}\right) = 1.

It converges, so the rational tail converges. The check is consistent with the degree condition \(1-3=-2<-1\), and the ratio limit \(2\) confirms the leading coefficient comparison.

Worked Example: A Divergent Rational Tail

Determine whether \(\int_1^\infty \frac{x^2+1}{x^2+3}\,dx\) converges.

Both polynomials have degree \(2\), so the dominant power is \(x^0=1\). Directly,

\frac{x^2+1}{x^2+3}\longrightarrow 1.

For a lower-bound check, when \(x\ge 1\),

\frac{x^2+1}{x^2+3}\ge \frac12 \quad\Longleftrightarrow\quad 2x^2+2\ge x^2+3 \quad\Longleftrightarrow\quad x^2\ge1.

Therefore, for every \(R>1\),

\int_1^R\frac{x^2+1}{x^2+3}\,dx \ge \frac12\int_1^R 1\,dx = \frac{R-1}{2}.

The lower bound tends to infinity, so the integral diverges. This agrees with the degree difference \(2-2=0\), which is not less than \(-1\).

Finite Singularities: Count the Surviving Factors

Suppose \(a\) is a real root of the denominator. Let its multiplicity in the numerator be \(r\), and its multiplicity in the denominator be \(s\); take \(r=0\) if the numerator does not vanish at \(a\). Factoring out those powers gives, near \(a\),

f(x)=(x-a)^{r-s}h(x), \qquad h(a)\ne 0,

where \(h\) is continuous near \(a\). The net exponent \(r-s\), not the original denominator multiplicity by itself, determines the local behavior. If \(r\ge s\), cancellation removes the denominator zero and the function remains bounded near \(a\). If \(r<s\), a pole remains.

Theorem: Near a finite point \(a\), a rational function with local form \(f(x)=(x-a)^k h(x)\), where \(h\) is continuous and \(h(a)\ne0\), is absolutely integrable on a one-sided neighborhood of \(a\) exactly when \(k>-1\). For rational functions \(k\) is an integer. Thus a genuine pole, for which \(k\le -1\), makes the ordinary improper integral diverge on each side that is included in the interval.

To see the threshold, set \(t=|x-a|\). Since \(h\) stays bounded above and bounded away from zero sufficiently close to \(a\), the local magnitude is comparable to \(t^k\). For \(k\ne -1\),

\int_\varepsilon^\delta t^k\,dt = \frac{\delta^{k+1}-\varepsilon^{k+1}}{k+1}.

As \(\varepsilon\to0^+\), this has a finite limit exactly when \(k+1>0\). At \(k=-1\), the integral is \(\ln\delta-\ln\varepsilon\), which diverges. The argument applies separately on the left and right of an interior singularity. Opposite-side cancellation does not rescue the ordinary improper integral, which requires both one-sided integrals to converge.

Worked Example: A Double Pole Inside a Finite Interval

Classify \(\int_0^2 \frac{x+2}{(x-1)^2(x+3)}\,dx\).

The only denominator root in the interval is \(x=1\). The numerator is \(1+2=3\) there, and the factor \(x+3\) equals \(4\), so the local leading factor is \(\frac34(x-1)^{-2}\). Verify this using a ratio:

\frac{\dfrac{x+2}{(x-1)^2(x+3)}}{(x-1)^{-2}} = \frac{x+2}{x+3} \longrightarrow \frac34 \qquad (x\to1).

The comparison integral diverges on both sides. For example, on the right, with \(t=x-1\),

\int_{1+\varepsilon}^{2}\frac{dx}{(x-1)^2} = \int_\varepsilon^1 t^{-2}\,dt = \left[-\frac1t\right]_\varepsilon^1 = \frac1\varepsilon-1 \longrightarrow\infty.

The positive finite ratio limit \(3/4\) gives the same divergence for the rational integrand. Hence the improper integral over \([0,2]\) diverges. The check is that the surviving pole order is \(2\), giving exponent \(-2\le -1\).

Cancel Before Classifying

A factor displayed in the denominator does not necessarily create a genuine pole: the numerator may contain the same factor. Cancel common factors before applying either the degree test or the finite-pole test. If an original formula is undefined at a cancelled point, the resulting function has a removable hole there; a bounded discontinuity does not prevent integrability on a finite interval.

Worked Example: A Removable Hole and a Convergent Tail

Classify \(\int_0^\infty \frac{x-2}{(x-2)(x^2+1)}\,dx\), treating the original expression as undefined at \(x=2\).

For \(x\ne2\), cancel the common factor:

\frac{x-2}{(x-2)(x^2+1)} = \frac{1}{x^2+1}.

The cancelled point is bounded, so it does not create a divergent finite contribution. At infinity the reduced function has degree difference \(0-2=-2<-1\). Its antiderivative is \(\arctan x\), since \(\frac{d}{dx}\arctan x=\frac{1}{1+x^2}\). Splitting at the hole, the integral is

\int_0^2\frac{dx}{1+x^2} + \int_2^\infty\frac{dx}{1+x^2} = \arctan 2 + \left(\frac{\pi}{2}-\arctan 2\right) = \frac{\pi}{2}.

The two finite-point contributions agree with the continuous extension, and the tail limit is finite. Differentiating the antiderivative verifies the integrand, and the endpoint values confirm convergence.

A Reliable Classification Procedure

For a rational integral over a finite or infinite interval, use the following sequence. It prevents a convergent tail from concealing a divergent pole, or a removable factor from being mistaken for a pole.

Reduce the rational function.
Cancel common factors and identify any removable holes.

List every problematic location.
Check finite endpoints, real denominator roots inside the interval, and each infinite endpoint.

Analyze each finite root locally.
Compare the surviving local factor with \(|x-a|^k\); a pole has \(k\le-1\) and causes divergence.

Analyze each infinite tail by degrees.
If the reduced numerator and denominator have degrees \(n\) and \(m\), the tail converges exactly when \(m\ge n+2\).

Combine the local conclusions.
The improper integral converges only if every required one-sided limit is finite.

Common Pitfall: Using a Principal Value Instead

For \(1/x\) on \([-1,1]\), symmetric limits can cancel:

\int_{-1}^{-\varepsilon}\frac{dx}{x} + \int_{\varepsilon}^{1}\frac{dx}{x} = \ln\varepsilon-\ln 1+\ln 1-\ln\varepsilon = 0.

This calculation takes the two cutoffs to zero together. It gives a Cauchy principal value, not convergence of the ordinary improper integral. In the ordinary definition, the left and right limits are evaluated separately: each diverges. Rational functions with simple poles can exhibit this same misleading cancellation. Always check each side independently.

Why This Method Matters

Rational functions can look complicated while their convergence behavior depends on only a few local facts: the degree difference at infinity and the net multiplicity at each finite root. Factoring and cancellation expose those facts. The method also makes the scope of a conclusion clear: a complete classification requires checking every endpoint and every genuine pole, not merely the most visually prominent feature of the formula.

Check Your Understanding

For each question, identify the dominant power or surviving factor before deciding convergence.

What condition on the numerator and denominator degrees makes a rational tail over \([A,\infty)\) converge?
For \(\frac{x+1}{(x-2)^3}\), what is the local exponent at \(x=2\), and does the integral converge on a one-sided neighborhood?
Why must common numerator and denominator factors be cancelled before counting pole orders?
Does a finite Cauchy principal value establish convergence of an ordinary improper integral? Explain briefly.
A rational function has degree difference \(-1\) at infinity. Does its tail converge, and what power integral gives the comparison?