Resolving Cancellations in Series-Based Limits

Why Cancellation Requires More Terms

When a function is represented by a power series, its first few terms often determine its behavior near the center of the series. But if an expression subtracts two functions with the same leading terms, those terms cancel. The behavior of the difference is then controlled by a later coefficient.

This matters especially in limits. Direct substitution into a quotient may give \(0/0\), while a series comparison shows how quickly the numerator and denominator approach zero relative to one another. The essential task is to keep expanding until the first nonzero coefficient remains.

Key idea: Combine power series coefficient by coefficient. After cancellation, identify the first nonzero coefficient; its power and value determine the leading behavior of the expression near the center.

Finding the First Surviving Term

Suppose two power series converge for \(x\) near \(0\): \(f(x)=\sum_{k=0}^{\infty}a_kx^k\) and \(g(x)=\sum_{k=0}^{\infty}b_kx^k\). Their difference is represented near \(0\) by subtracting coefficients of like powers:

f(x)-g(x)=\sum_{k=0}^{\infty}(a_k-b_k)x^k.

If \(a_0=b_0\), the constant terms cancel. If \(a_1=b_1\) as well, the linear terms cancel too. Continue in this way until the first index \(m\) for which \(a_m-b_m\ne 0\). Then the difference begins with \((a_m-b_m)x^m\).

The same procedure applies to any finite linear combination of series. For instance, if \(F(x)=2f(x)-3g(x)+h(x)\), the coefficient of \(x^k\) in \(F\) is \(2a_k-3b_k+c_k\), where \(c_k\) is the coefficient of \(x^k\) in \(h\). Cancellation is determined by the combined coefficient, not by the apparent size of any one function.

Definition: If a power series for \(F\) has coefficients equal to zero for powers below \(x^m\), and its coefficient of \(x^m\) is nonzero, then \(F\) has leading term \(c_mx^m\), where \(c_m\ne 0\). Equivalently, \(F(x)=x^m(c_m+\text{terms that approach }0\text{ as }x\to0)\).

When using a known expansion, retain enough terms to reach the first nonzero coefficient of the expression being evaluated. A term-by-term subtraction that stops before that point does not determine the limit.

Why the First Surviving Term Determines the Limit

The principle can be stated and justified directly for convergent power series. Suppose \(F(x)=\sum_{k=0}^{\infty}c_kx^k\) converges for \(|x|<R\), where \(R>0\), and suppose \(c_0,\ldots,c_{m-1}\) are zero while \(c_m\ne0\). Factoring out \(x^m\) gives a new power series that is continuous near \(0\).

Theorem: Under these conditions, \(F(x)/x^m\to c_m\) as \(x\to0\), with \(x\ne0\). More generally, if \(F\) and \(G\) have leading terms \(a_mx^m\) and \(b_nx^n\), with \(a_m,b_n\ne0\), then their quotient is controlled by the difference between \(m\) and \(n\).

Factor the first surviving power.
Since the coefficients below \(m\) are zero, write \(F(x)=x^m(c_m+c_{m+1}x+c_{m+2}x^2+\cdots)\).

Use continuity of the remaining series.
The series in parentheses converges near \(0\), and its value approaches \(c_m\) as \(x\to0\).

Take the limit.
For \(x\ne0\), \(F(x)/x^m=c_m+c_{m+1}x+c_{m+2}x^2+\cdots\), so its limit is \(c_m\).

For a quotient, factor the leading powers in both series. If \(F(x)=x^m(A+u(x))\) and \(G(x)=x^n(B+v(x))\), where \(A\ne0\), \(B\ne0\), and \(u(x),v(x)\to0\), then

\frac{F(x)}{G(x)} = x^{m-n}\frac{A+u(x)}{B+v(x)}.

The second factor approaches \(A/B\). If \(m=n\), the quotient approaches \(A/B\). If \(m>n\), it approaches \(0\). If \(m<n\), its magnitude grows without bound; whether a two-sided limit is \(+\infty\), \(-\infty\), or fails to exist depends on the power \(x^{m-n}\) and the sign of \(A/B\).

Worked Examples

Worked Example: A Logarithm After Two Cancellations

Evaluate \(\displaystyle\lim_{x\to0}\frac{\ln(1+x)-x+\frac{x^2}{2}}{x^3}\).

For \(|x|<1\), the logarithm series begins \(\ln(1+x)=x-\frac{x^2}{2}+\frac{x^3}{3}-\frac{x^4}{4}+O(x^5)\). Substitute this into the numerator and group like powers:

\begin{aligned} \ln(1+x)-x+\frac{x^2}{2} &=\left(x-\frac{x^2}{2}+\frac{x^3}{3}-\frac{x^4}{4}+O(x^5)\right)-x+\frac{x^2}{2}\\ &=x-x-\frac{x^2}{2}+\frac{x^2}{2}+\frac{x^3}{3}-\frac{x^4}{4}+O(x^5)\\ &=\frac{x^3}{3}-\frac{x^4}{4}+O(x^5). \end{aligned}

The constant term is zero, and the \(x\) and \(x^2\) terms cancel. Divide by \(x^3\), for \(x\ne0\):

\frac{\ln(1+x)-x+\frac{x^2}{2}}{x^3} = \frac{1}{3}-\frac{x}{4}+O(x^2) \longrightarrow \frac{1}{3}.

As a check, the numerator divided by \(x^3\) is explicitly \(1/3-x/4+O(x^2)\); every term after \(1/3\) tends to zero. The limit is \(\boxed{1/3}\).

Worked Example: Continuing Until the Fifth-Power Term

Evaluate \(\displaystyle\lim_{x\to0}\frac{\arctan x-x+\frac{x^3}{3}}{x^5}\).

The arctangent series near zero is \(\arctan x=x-\frac{x^3}{3}+\frac{x^5}{5}-\frac{x^7}{7}+O(x^9)\). Substitute and combine each power:

\begin{aligned} \arctan x-x+\frac{x^3}{3} &=\left(x-\frac{x^3}{3}+\frac{x^5}{5}-\frac{x^7}{7}+O(x^9)\right)-x+\frac{x^3}{3}\\ &=x-x-\frac{x^3}{3}+\frac{x^3}{3}+\frac{x^5}{5}-\frac{x^7}{7}+O(x^9)\\ &=\frac{x^5}{5}-\frac{x^7}{7}+O(x^9). \end{aligned}

Now divide by \(x^5\):

\frac{\arctan x-x+\frac{x^3}{3}}{x^5} = \frac{1}{5}-\frac{x^2}{7}+O(x^4) \longrightarrow \frac{1}{5}.

The \(x\) and \(x^3\) terms cancel, so the \(x^5\) term is the first surviving term. The displayed quotient also verifies that the terms remaining after \(1/5\) vanish. The limit is \(\boxed{1/5}\).

Worked Example: Comparing Two Series with the Same Leading Order

Evaluate \(\displaystyle\lim_{x\to0}\frac{\ln(1+x)-x+\frac{x^2}{2}}{\arctan x-x}\).

From the logarithm expansion, the numerator is \(\frac{x^3}{3}-\frac{x^4}{4}+O(x^5)\). From the arctangent expansion, the denominator is \(-\frac{x^3}{3}+\frac{x^5}{5}+O(x^7)\). Both start at order \(x^3\), but their leading coefficients have opposite signs. Factor \(x^3\) from both:

\begin{aligned} \frac{\ln(1+x)-x+\frac{x^2}{2}}{\arctan x-x} &= \frac{x^3\left(\frac{1}{3}-\frac{x}{4}+O(x^2)\right)} {x^3\left(-\frac{1}{3}+\frac{x^2}{5}+O(x^4)\right)}\\ &= \frac{\frac{1}{3}-\frac{x}{4}+O(x^2)} {-\frac{1}{3}+\frac{x^2}{5}+O(x^4)}. \end{aligned}

The denominator of this last quotient approaches \(-1/3\), which is nonzero. Therefore, the limit is the ratio of the limiting coefficients:

\lim_{x\to0}\frac{\ln(1+x)-x+\frac{x^2}{2}}{\arctan x-x} = \frac{1/3}{-1/3} =-1.

For a check, the factored numerator and denominator have limiting bracket values \(1/3\) and \(-1/3\), respectively; their ratio is \(-1\). The limit is \(\boxed{-1}\).

Worked Example: A Smaller Numerator Order Than Denominator Order

Evaluate the behavior of \(\displaystyle\frac{\arctan x-x}{x^4}\) as \(x\to0\).

The series gives \(\arctan x-x=-\frac{x^3}{3}+\frac{x^5}{5}+O(x^7)\). Divide by \(x^4\) and simplify:

\frac{\arctan x-x}{x^4} = -\frac{1}{3x}+\frac{x}{5}+O(x^3).

Because the leading term is \(-1/(3x)\), the quotient is unbounded. Its direction differs on the two sides: as \(x\to0^+\), \(-1/(3x)\to-\infty\); as \(x\to0^-\), \(-1/(3x)\to+\infty\). The terms \(x/5+O(x^3)\) tend to zero and do not change these conclusions. Thus the two-sided limit does not exist, and the one-sided limits have opposite infinite signs.

As a check, the numerator has leading order \(x^3\), while the denominator has order \(x^4\). Their quotient therefore contains one uncancelled factor \(1/x\), with negative coefficient, consistent with the one-sided behavior just found.

A Common Mistake: Stopping Before the First Surviving Term

Consider \(\displaystyle\lim_{x\to0}\frac{\ln(1+x)-x}{x^2}\). A mistaken calculation might say: “The linear terms \(x-x\) cancel, so the numerator is zero to the relevant order; therefore the limit is zero.” That conclusion is wrong because cancellation of the linear terms does not mean the whole numerator vanishes to higher than second order.

Continue the expansion instead:

\begin{aligned} \ln(1+x)-x &=\left(x-\frac{x^2}{2}+\frac{x^3}{3}+O(x^4)\right)-x\\ &=-\frac{x^2}{2}+\frac{x^3}{3}+O(x^4), \end{aligned}

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

The correction is to keep expanding after the first cancellation. The first surviving term is \(-x^2/2\), not zero, and it determines the limit.

A Reliable Procedure

For limits involving cancellations in series, proceed systematically. First write each series about the limit point, then combine coefficients of matching powers. Continue until the first nonzero coefficient appears in each relevant expression. Finally, factor out the corresponding powers and compare what remains.

Use a common center.
For a limit at \(x=0\), express all functions as power series in \(x\) valid in a neighborhood of zero.

Combine like powers carefully.
Write the contributions to each power explicitly; do not assume that cancellation of one term settles the expression.

Stop at the first nonzero combined coefficient.
For a quotient, find this term in both numerator and denominator.

Factor and evaluate.
Compare the powers and take the limit of the remaining factors, checking that the denominator's remaining factor has a nonzero limit.

If the limit is centered at \(x=a\), the same reasoning uses powers of \(x-a\) instead of powers of \(x\). The crucial requirements are that the series represent the functions near the center and that the denominator's first surviving coefficient is nonzero when taking a quotient.

Why This Method Is Useful

Series-based cancellation turns an apparent \(0/0\) form into a comparison of explicit leading terms. It also makes the role of assumptions visible: the expansions must be valid near the limit point, and a quotient comparison requires a nonzero leading denominator coefficient. Checking these conditions prevents invalid division and helps distinguish a finite limit from unbounded behavior.

The main discipline is simple but important: cancellation tells you which terms to discard, while the first surviving coefficient tells you what remains. Keep enough terms to find that coefficient, and then verify the resulting limit from the factored expression.

Check Your Understanding

For each question, identify the first surviving term before deciding the limit.

Use the logarithm series to evaluate \(\displaystyle\lim_{x\to0}\frac{\ln(1+x)-x}{x^2}\).
In a difference of two power series, what must be true of the coefficients below \(x^m\) if the first surviving term is \(c_mx^m\)?
If a numerator has leading term \(Ax^4\) and a denominator has leading term \(Bx^4\), where \(A\) and \(B\) are nonzero, what is the limit of their quotient?
If the numerator starts at order \(x^5\) and the denominator at order \(x^3\), what is the limit of their quotient?
Why is it not enough to observe that the first terms cancel in a proposed limit?