How To Find Upper And Lower Bounds Of Series

Introduction: How to Find Upper and Lower Bounds of Series

When working with infinite series, one of the most practical skills is the ability to determine upper and lower bounds for the sum. Which means these bounds give you a reliable range within which the true value of the series must lie, which is essential for convergence testing, error estimation, and numerical approximations. In this article we’ll walk through the theory, step‑by‑step procedures, and real‑world examples that show exactly how to find those bounds for a wide variety of series Simple, but easy to overlook..

Understanding Upper and Lower Bounds

Definition of Bounds

An upper bound of a series is a number U such that the partial sums Sₙ never exceed U (i.e., Sₙ ≤ U for all n). Conversely, a lower bound L satisfies Sₙ ≥ L for every partial sum. When both exist, they enclose the series’ limit, often written as

[ L \le \sum_{n=1}^{\infty} a_n \le U . ]

If L = U, the bound is tight and you have found the exact sum.

Why Bounds Matter in Series

• Convergence testing – Establishing that a series is bounded above (and below) can prove convergence via the Monotone Convergence Theorem.
• Error control – In numerical methods, the difference between a partial sum and the true sum is bounded by the gap between upper and lower estimates.
• Comparison techniques – Bounds allow you to compare unfamiliar series with well‑known ones (e.g., p‑series or geometric series) without computing the exact sum.

Key Methods for Determining Bounds

The Comparison Test

If you have two series (\sum a_n) and (\sum b_n) with (0 \le a_n \le b_n) for all n, then:

• If (\sum b_n) converges, so does (\sum a_n).
• If (\sum a_n) diverges, then (\sum b_n) must also diverge.

This test is powerful for establishing upper bounds (using a convergent larger series) or lower bounds (using a divergent smaller series).

The Integral Test

For a positive, decreasing function f such that aₙ = f(n), the series (\sum a_n) and the integral (\int_{1}^{\infty} f(x),dx) share convergence behavior. Beyond that, the integral provides explicit bounds:

[ \int_{1}^{N+1} f(x),dx \le \sum_{n=1}^{N} a_n \le a_1 + \int_{1}^{N} f(x),dx . ]

Taking the limit as N → ∞ yields both an upper and lower bound for the infinite series And that's really what it comes down to..

The Ratio Test

While primarily a convergence test, the ratio test can also give asymptotic bounds. If (\lim_{n\to\infty} \frac{a_{n+1}}{a_n} = L < 1), the series converges, and you can often bound the sum by a geometric series with ratio r just above L. Conversely, if L > 1, the series diverges, giving a lower bound of infinity.

The Alternating Series Test

For an alternating series (\sum (-1)^{n-1} b_n) with decreasing bₙ → 0, the remainder after N terms satisfies

[ |R_N| \le b_{N+1}. ]

Thus b_{N+1} serves as an upper bound on the error, while the partial sum itself provides a lower bound (or vice‑versa, depending on the sign pattern) Took long enough..

Step‑by‑Step Procedure to Find Bounds

Step 1: Identify the Type of Series

Determine whether the series is:

• Positive term (e.g., p‑series, geometric)
• Alternating (e.g., (\sum (-1)^{n-1}/\sqrt{n}))
• Mixed sign (requires absolute/conditional convergence analysis)

Knowing the type guides the choice of bounding technique And that's really what it comes down to..

Step 2: Choose an Appropriate Test

Match the series to a test that yields usable bounds:

Step 3: Apply the Test to Obtain Estimates

• Integral test: Compute (\int_{1}^{\infty} f(x),dx) and (\int_{1}^{N+1} f(x),dx) for a large N.
• Comparison test: Find a convergent series bₙ with aₙ ≤ bₙ (upper bound) or a divergent series cₙ with aₙ ≥ cₙ (lower bound).
• Ratio test: Determine L and select a geometric series with ratio r > L to bound the sum.
• Alternating series: Use the first omitted term b_{N+1} as the error bound.

Step 4: Verify and Refine the Bounds

• Check monotonicity: Ensure partial sums are indeed increasing or decreasing as assumed.
• Tightness: If the upper and lower bounds are close, you may have a near‑exact estimate.
• Iterate: For more precision, increase N in the integral or comparison bounds and recompute.

Practical Examples

Example 1: p‑Series (\displaystyle \sum_{n=1}^{\infty} \frac{1}{n^{2}})

The series is positive and decreasing. Apply the integral test with *f(x) =

…(f(x)=\frac{1}{x^{2}}). Because (f) is continuous, positive and decreasing on ([1,\infty)), the integral test gives

[ \int_{1}^{N+1}\frac{dx}{x^{2}};\le;\sum_{n=1}^{N}\frac{1}{n^{2}};\le;1+\int_{1}^{N}\frac{dx}{x^{2}} . ]

Evaluating the integrals, (\int_{1}^{M}\frac{dx}{x^{2}}=1-\frac{1}{M}). Hence

[ 1-\frac{1}{N+1};\le;S_{N};\le;2-\frac{1}{N}, \qquad S_{N}:=\sum_{n=1}^{N}\frac{1}{n^{2}} . ]

Letting (N\to\infty) yields the classic bounds for the infinite sum

[ 1;\le;\sum_{n=1}^{\infty}\frac{1}{n^{2}};\le;2 . ]

A tighter estimate is obtained by choosing a moderate (N). For (N=10),

[ 0.909;<;S_{10};<;1.900, ]

and adding the tail bound (\int_{10}^{\infty}x^{-2}dx=0.1) gives

[ 1.009;<;\sum_{n=1}^{\infty}\frac{1}{n^{2}};<;2.000 . ]

In fact the exact value is (\pi^{2}/6\approx1.6449), which lies comfortably inside these intervals No workaround needed..

Example 2: Alternating Harmonic Series

Consider (\displaystyle\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}). The terms (b_{n}=1/n) decrease to zero, so the alternating series test applies. After summing the first (N) terms, the remainder satisfies

[ |R_{N}|\le b_{N+1}=\frac{1}{N+1}. ]

Thus the partial sum (S_{N}) brackets the true sum (S):

[ S_{N}-\frac{1}{N+1};\le;S;\le;S_{N}+\frac{1}{N+1}. ]

For (N=5), (S_{5}=1-\frac12+\frac13-\frac14+\frac15\approx0.7833); the error bound is (1/6\approx0.1667), giving

[ 0.6166;<;S;<;0.9500 . ]

Increasing (N) shrinks the interval symmetrically around (S), which is known to be (\ln 2\approx0.6931) That's the part that actually makes a difference..

Example 3: Ratio‑Test Bound for a Factorial Series

Examine (\displaystyle\sum_{n=0}^{\infty}\frac{n!}{(2n)!}). Compute

[