The Central Idea
Suppose a sequence has terms
\[ a_1,a_2,a_3,\ldots \]An infinite series is written
\[ \sum_{n=1}^{\infty}a_n =a_1+a_2+a_3+\cdots. \]We do not add infinitely many terms in one completed arithmetic step. Instead, we study the finite sums
\[ S_N=a_1+a_2+\cdots+a_N \]and ask whether the sequence of partial sums \((S_N)\) approaches a finite limit.
From a Sequence to a Series
A sequence lists terms. A series accumulates them.
If
\[ a_n=\frac{1}{2^n}, \]then the sequence is
\[ \frac12,\frac14,\frac18,\frac1{16},\ldots \]while the corresponding series is
\[ \frac12+\frac14+\frac18+\frac1{16}+\cdots. \]The terms and their accumulated sums are related, but they are different mathematical objects.
Partial Sums
The first partial sum contains one term:
\[ S_1=a_1. \]The second contains two terms:
\[ S_2=a_1+a_2. \]In general,
\[ \boxed{S_N=\sum_{n=1}^{N}a_n}. \]Every partial sum is an ordinary finite sum. The infinite behavior appears only when we ask what happens as \(N\to\infty\).
The Definition of Series Convergence
The series
\[ \sum_{n=1}^{\infty}a_n \]converges to \(S\) if its sequence of partial sums converges to \(S\):
\[ \boxed{\lim_{N\to\infty}S_N=S}. \]When this happens, we write
\[ \boxed{\sum_{n=1}^{\infty}a_n=S}. \]What the Equality Really Means
Writing
\[ \sum_{n=1}^{\infty}a_n=S \]is shorthand for a limit statement. It means that finite partial sums can be made arbitrarily close to \(S\) by including sufficiently many terms.
The value \(S\) is therefore the limit of an accumulation process.
Example: A Series That Converges
Consider
\[ \frac12+\frac14+\frac18+\frac1{16}+\cdots. \]The first few partial sums are
\[ \frac12,\quad \frac34,\quad \frac78,\quad \frac{15}{16},\ldots \]These values approach \(1\), so the series converges to \(1\):
\[ \boxed{\frac12+\frac14+\frac18+\cdots=1}. \]The next lesson will develop the geometric-series formula that makes this calculation systematic.
Example: A Series That Diverges by Growth
Consider
\[ 1+1+1+1+\cdots. \]Its partial sums are
\[ 1,2,3,4,\ldots \]Since
\[ S_N=N\to\infty, \]the series does not approach a finite sum. It diverges.
Example: A Series That Diverges by Oscillation
Consider
\[ 1-1+1-1+1-1+\cdots. \]Its partial sums are
\[ 1,0,1,0,1,0,\ldots \]The partial sums do not approach one number, so the series diverges in the usual sense.
Convergence Is About Partial Sums
This point is fundamental:
\[ \boxed{\text{A series converges exactly when its partial sums converge.}} \]The terms \(a_n\) matter because they determine how the partial sums change, but the convergence question belongs to the sequence \((S_N)\).
The Increment Relationship
Consecutive partial sums differ by one term:
\[ S_N-S_{N-1}=a_N. \]Equivalently,
\[ \boxed{a_N=S_N-S_{N-1}}. \]A term of the series is the increment that moves one partial sum to the next.
A Necessary Condition for Convergence
Suppose
\[ \sum_{n=1}^{\infty}a_n \]converges. Then \(S_N\to S\) and also \(S_{N-1}\to S\). Since
\[ a_N=S_N-S_{N-1}, \]it follows that
\[ \boxed{\lim_{n\to\infty}a_n=0}. \]The Divergence Test for Terms
The necessary condition gives a fast divergence test.
If
\[ \lim_{n\to\infty}a_n\ne0 \]or if the limit of \(a_n\) does not exist, then
\[ \boxed{\sum_{n=1}^{\infty}a_n\text{ diverges}.} \]A convergent series cannot keep adding increments that remain noticeably different from zero.
Example: Immediate Divergence
Consider
\[ \sum_{n=1}^{\infty}\frac{n+1}{n}. \]Its terms satisfy
\[ \frac{n+1}{n}=1+\frac1n\to1. \]Since the terms do not approach zero, the series diverges immediately.
Terms Approaching Zero Is Not Enough
The implication works only one way.
\[ \sum a_n\text{ converges} \quad\Longrightarrow\quad a_n\to0. \]But
\[ a_n\to0 \]does not by itself imply that \(\sum a_n\) converges.
The Harmonic Series as the Standard Warning
The harmonic series is
\[ \sum_{n=1}^{\infty}\frac1n =1+\frac12+\frac13+\frac14+\cdots. \]Its terms approach zero:
\[ \frac1n\to0. \]Nevertheless, the harmonic series diverges. Later convergence-test lessons will prove this carefully.
Why Tiny Terms Can Still Accumulate Without Bound
A term can be small without being small enough, soon enough, to produce a finite total accumulation.
Infinite-series analysis therefore asks two questions:
- Do the individual terms approach zero?
- Do the accumulated partial sums approach a finite number?
The first is necessary. The second decides convergence.
Reading a Series from a Table of Partial Sums
Suppose a series has the following partial sums.
| \(N\) | \(S_N\) |
|---|---|
| 1 | 0.5000 |
| 2 | 0.7500 |
| 3 | 0.8750 |
| 4 | 0.9375 |
| 8 | 0.9961 |
The table strongly suggests convergence to \(1\), because the partial sums are getting closer to \(1\).
Graphing Partial Sums
The graph of a series is often best understood as the graph of its partial sums:
\[ (1,S_1),(2,S_2),(3,S_3),\ldots \]If these points settle toward a horizontal level \(S\), the series converges to \(S\). If they grow, fall, or keep oscillating without settling, the series diverges.
Indexing Can Start Somewhere Else
A series may begin at any convenient integer index:
\[ \sum_{n=0}^{\infty}a_n, \qquad \sum_{n=3}^{\infty}a_n, \qquad \sum_{k=10}^{\infty}b_k. \]The letter used for the index is a dummy variable. What matters is the term formula and the starting point.
Changing the Index Letter Does Not Change the Series
For example,
\[ \sum_{n=1}^{\infty}\frac1{2^n} \]and
\[ \sum_{k=1}^{\infty}\frac1{2^k} \]represent the same series. The index symbol only labels the terms.
Finite Sums Versus Infinite Series
A finite sum has a last term:
\[ \sum_{n=1}^{100}a_n. \]An infinite series has no final index:
\[ \sum_{n=1}^{\infty}a_n. \]The finite sum is evaluated by arithmetic. The infinite series is defined through a limit of finite sums.
Finite Changes Do Not Determine Convergence
Adding, removing, or changing finitely many terms can change the numerical value of a convergent series, but it cannot change whether the tail converges or diverges.
Convergence is an eventual property.
Why Finite Changes Do Not Affect Convergence
Suppose two series differ only in their first \(m\) terms. After that point, their partial sums differ by one fixed constant \(C\).
\[ T_N=S_N+C \]for all sufficiently large \(N\). Therefore one partial-sum sequence has a finite limit exactly when the other does.
Tails of a Series
A tail is the part of a series after a chosen finite point. For example,
\[ \sum_{n=1}^{\infty}a_n =a_1+a_2+\cdots+a_m+\sum_{n=m+1}^{\infty}a_n. \]The original series converges exactly when any one of its tails converges.
The Tail Remainder
If a series converges to \(S\), then after computing the first \(N\) terms we have the partial sum \(S_N\). The omitted tail is
\[ R_N=S-S_N. \]This quantity is called the remainder or tail error after \(N\) terms.
Convergence Means the Tail Becomes Negligible
If
\[ S_N\to S, \]then
\[ R_N=S-S_N\to0. \]So a convergent series can be approximated by a sufficiently long finite partial sum with arbitrarily small error.
The Cauchy Viewpoint for Series
A series converges when its partial sums eventually become arbitrarily close to one another.
If \(m>n\), then
\[ S_m-S_n=a_{n+1}+a_{n+2}+\cdots+a_m. \]Thus convergence means that sufficiently far-out finite blocks of terms have arbitrarily small total effect.
The Cauchy Criterion
For real or complex series, convergence is equivalent to the following statement: for every \(\varepsilon>0\), there exists \(N\) such that whenever \(m>n\ge N\),
\[ \left|a_{n+1}+a_{n+2}+\cdots+a_m\right|<\varepsilon. \]This criterion describes convergence entirely in terms of tails, without requiring us to know the final sum in advance.
Positive-Term Series
If every term satisfies
\[ a_n\ge0, \]then the partial sums are increasing:
\[ S_{N+1}=S_N+a_{N+1}\ge S_N. \]Therefore a positive-term series converges exactly when its partial sums are bounded above. Later lessons develop practical tests for determining that boundedness.
Why Positive-Term Series Are Structurally Simpler
With nonnegative terms, partial sums cannot cancel or move backward. They either approach a finite ceiling or increase without bound.
This is why comparison, integral, ratio, and root tests are especially natural for positive-term series.
Signed Series
When positive and negative terms are mixed, partial sums can move in both directions. Cancellation can become important.
The alternating pattern
\[ a_1-a_2+a_3-a_4+\cdots \]is one important example. A later lesson will develop the alternating-series test systematically.
Cancellation Can Create Convergence
A series with signed terms may converge even when the corresponding series of absolute values behaves differently.
This distinction leads to the later concepts of absolute convergence and conditional convergence.
Basic Algebra of Convergent Series
Suppose
\[ \sum_{n=1}^{\infty}a_n=A \]and
\[ \sum_{n=1}^{\infty}b_n=B. \]Then finite linear combinations preserve convergence.
Constant Multiples
For any constant \(c\),
\[ \boxed{\sum_{n=1}^{\infty}c a_n =c\sum_{n=1}^{\infty}a_n =cA}. \]This follows because the partial sums satisfy
\[ \sum_{n=1}^{N}ca_n=cS_N. \]Sums and Differences of Convergent Series
If both component series converge, then
\[ \boxed{ \sum_{n=1}^{\infty}(a_n+b_n)=A+B } \]and
\[ \boxed{ \sum_{n=1}^{\infty}(a_n-b_n)=A-B. } \]These are direct applications of sequence limit laws to partial sums.
Linearity of Infinite Summation
Combining the previous rules gives
\[ \boxed{ \sum_{n=1}^{\infty}(ca_n+db_n)=cA+dB } \]whenever both original series converge.
Do Not Subtract Divergent Series Carelessly
Linearity is safest when the component series are known to converge.
An expression such as
\[ \sum(a_n-b_n) \]can converge even when \(\sum a_n\) and \(\sum b_n\) individually diverge. In that situation, treating the result as “\(\infty-\infty\)” is not valid.
Example: Cancellation Before Summation
Consider
\[ \sum_{n=1}^{\infty}\left(\frac1n-\frac1{n+1}\right). \]The two separate harmonic-type series diverge, so we should not subtract their supposed infinite values. Instead, keep the terms paired and examine the partial sums.
The next lesson will show that this series telescopes.
Reindexing a Series
Sometimes a series is easier to compare or combine after shifting its index. For example,
\[ \sum_{n=0}^{\infty}\frac1{2^{n+1}} \]lists the same terms as
\[ \sum_{k=1}^{\infty}\frac1{2^k}. \]Reindexing changes the labels, not the actual ordered terms.
A Safe Reindexing Procedure
- Choose the new index variable.
- Write the substitution relating old and new indices.
- Transform the lower bound.
- Transform the term formula.
- Check the first few terms before and after the shift.
Matching the actual terms is the best verification.
Example: Shifting an Index
Start with
\[ \sum_{n=1}^{\infty}\frac1{2^n}. \]Let \(k=n-1\), so \(n=k+1\). When \(n=1\), \(k=0\). Therefore,
\[ \boxed{ \sum_{n=1}^{\infty}\frac1{2^n} =\sum_{k=0}^{\infty}\frac1{2^{k+1}}. } \]Grouping Terms: Use Care
Grouping consecutive terms of a convergent series into finite blocks does not change the sum. For example,
\[ a_1+a_2+a_3+a_4+\cdots \]can be grouped as
\[ (a_1+a_2)+(a_3+a_4)+\cdots. \]But changing the order of terms is a separate issue and can be delicate for conditionally convergent series.
Rearrangement Is Not Always Harmless
For finite sums, changing the order never changes the result. Infinite series require more care because different partial-sum paths can matter.
Absolutely convergent series may be rearranged safely. Conditionally convergent series can behave differently under rearrangement. That topic belongs to the later lesson on absolute and conditional convergence.
Series as Accumulation
A series can model a total quantity assembled from many increments:
- successive payments or costs;
- repeated corrections in an algorithm;
- distance traveled in shrinking stages;
- energy distributed across modes;
- probability allocated across countably many outcomes;
- terms in a function approximation.
Convergence determines whether the accumulated total remains finite.
Series and Area: A Preview
Some series can be compared with areas under curves. This creates a bridge between discrete sums and continuous integrals.
Later, the integral test will use this connection to study positive-term series such as
\[ \sum_{n=1}^{\infty}\frac1{n^p}. \]Series and Approximation
A convergent infinite series gives a finite target value, but actual computation uses only finitely many terms.
The practical problem becomes
\[ \boxed{\text{How large must }N\text{ be so that }|S-S_N|\text{ is small?}} \]Error and remainder estimates later make this question quantitative.
Fast and Slow Convergence
Two series can both converge while requiring very different numbers of terms for accurate approximation.
A rapidly shrinking geometric pattern may converge quickly, while a series whose terms decay only algebraically can converge much more slowly.
Benchmark Family: Geometric Series
A geometric series has the form
\[ a+ar+ar^2+ar^3+\cdots. \]Its convergence depends on the common ratio \(r\). The next lesson develops the exact criterion and sum formula.
Benchmark Family: Telescoping Series
A telescoping series is arranged so that many terms cancel inside its partial sums. A typical pattern is
\[ \sum_{n=1}^{\infty}(b_n-b_{n+1}). \]The next lesson will show how partial sums expose the cancellation directly.
Benchmark Family: \(p\)-Series
A \(p\)-series has the form
\[ \sum_{n=1}^{\infty}\frac1{n^p}. \]Its convergence depends on the exponent \(p\). This family becomes a standard reference when learning positive-term convergence tests.
Benchmark Family: Alternating Series
A typical alternating series has the form
\[ \sum_{n=1}^{\infty}(-1)^{n-1}b_n, \qquad b_n\ge0. \]Alternating signs create cancellation. A later lesson gives conditions under which that cancellation guarantees convergence.
Power Series: A Preview
A power series is an infinite series whose terms depend on a variable:
\[ \sum_{n=0}^{\infty}c_n(x-a)^n. \]Instead of asking only whether one numerical series converges, we ask for which values of \(x\) the resulting series converges.
Common Error: Treating \(\infty\) as the Last Index
In
\[ \sum_{n=1}^{\infty}a_n, \]there is no term \(a_{\infty}\). The symbol \(\infty\) describes an unbounded limiting process for the upper index.
Common Error: Checking Only That \(a_n\to0\)
The condition
\[ a_n\to0 \]is necessary but not sufficient for convergence. The harmonic series is the standard counterexample.
Common Error: Confusing Terms with Partial Sums
The term \(a_N\) is one increment. The partial sum \(S_N\) is the accumulated total through that term:
\[ S_N=\sum_{n=1}^{N}a_n. \]Series convergence is determined by \(S_N\), not directly by \(a_N\).
Common Error: Writing a Sum Before Proving Convergence
Before assigning a finite value \(S\) to an infinite series, establish that its partial sums actually converge.
Formal manipulations with divergent series can produce misleading finite expressions if convergence has not been justified.
Common Error: Using Finite-Sum Intuition Without Limits
Associativity and distributivity remain useful, but an infinite series is not simply a very long finite sum. Limit behavior must be preserved.
Whenever a manipulation changes grouping, order, or decomposition, ask what it does to the partial sums.
A Reliable Procedure for Analyzing an Infinite Series
- Identify the general term \(a_n\) and starting index.
- Check the necessary condition \(a_n\to0\).
- If that limit is nonzero or nonexistent, conclude divergence.
- If the terms do approach zero, do not conclude convergence yet.
- Form or study the partial sums \(S_N\).
- Look for recognizable structure such as geometric or telescoping form.
- Use an appropriate convergence method when needed.
- If the series converges, identify or approximate its sum.
- When approximating, estimate the tail error \(R_N=S-S_N\).
Example: Full Partial-Sum Workflow
Consider
\[ \sum_{n=1}^{\infty}\frac1{2^n}. \]The terms approach zero, so immediate divergence is ruled out.
The partial sums are
\[ S_N=\frac12+\frac14+\cdots+\frac1{2^N}. \]For this geometric pattern,
\[ S_N=1-\frac1{2^N}. \]Therefore,
\[ \lim_{N\to\infty}S_N =1-0 =1. \]Hence
\[ \boxed{\sum_{n=1}^{\infty}\frac1{2^n}=1}. \]Example: Full Divergence-Test Workflow
Consider
\[ \sum_{n=1}^{\infty}\frac{3n-1}{2n+5}. \]First inspect the terms:
\[ \lim_{n\to\infty}\frac{3n-1}{2n+5}=\frac32. \]Since the terms do not approach zero,
\[ \boxed{\sum_{n=1}^{\infty}\frac{3n-1}{2n+5}\text{ diverges}.} \]No further test is needed.
Example: The Necessary Condition Is Inconclusive
Consider
\[ \sum_{n=1}^{\infty}\frac1n. \]We find
\[ \lim_{n\to\infty}\frac1n=0. \]This result does not prove convergence. It only tells us that the simplest divergence test is inconclusive.
Another argument is required, and that argument ultimately shows the series diverges.
Example: Recovering Terms from Partial Sums
Suppose a series has partial sums
\[ S_N=3-\frac{2}{N+1}. \]Then the series converges because
\[ \lim_{N\to\infty}S_N=3. \]Its individual terms can be recovered from
\[ a_N=S_N-S_{N-1}. \]For \(N\ge2\),
\[ a_N= \left(3-\frac{2}{N+1}\right) -\left(3-\frac{2}{N}\right) =\frac{2}{N(N+1)}. \]Practice 1: Identify the Partial Sum
For
\[ \sum_{n=1}^{\infty}\frac1{3^n}, \]write the first four partial sums \(S_1,S_2,S_3,S_4\).
Practice 2: Immediate Divergence
Determine whether the divergence test settles the behavior of
\[ \sum_{n=1}^{\infty}\frac{2n+3}{n+4}. \]Practice 3: Necessary Condition Only
For
\[ \sum_{n=1}^{\infty}\frac1{n^2}, \]evaluate \(\lim_{n\to\infty}1/n^2\). What can that limit alone tell you about the series?
Practice 4: Partial-Sum Limit
A series has partial sums
\[ S_N=5-\frac{3}{N}. \]Determine whether the series converges and, if so, find its sum.
Practice 5: Divergence by Oscillating Partial Sums
Suppose
\[ S_N=(-1)^N. \]Does the corresponding series converge?
Practice 6: Recover a Term
If
\[ S_N=2-\frac1N, \]find \(a_N=S_N-S_{N-1}\) for \(N\ge2\).
Practice 7: Finite Change
Suppose \(\sum_{n=1}^{\infty}a_n\) converges. If only the first three terms are replaced by different finite values, must the new series still converge? Explain.
Practice 8: Tail Error
A convergent series has sum \(S=7\), and its tenth partial sum is \(S_{10}=6.98\). Find the remainder \(R_{10}\).
Practice 9: Reindexing
Rewrite
\[ \sum_{n=2}^{\infty}\frac1{2^n} \]so that the new index begins at \(k=0\).
Practice 10: Explain the Error
A student observes that
\[ \lim_{n\to\infty}\frac1n=0 \]and concludes that
\[ \sum_{n=1}^{\infty}\frac1n \]must converge. Explain the mistake.
Practice Answers
Practice 1
\[ S_1=\frac13, \] \[ S_2=\frac13+\frac19=\frac49, \] \[ S_3=\frac{13}{27}, \qquad S_4=\frac{40}{81}. \]Practice 2
The terms satisfy
\[ \lim_{n\to\infty}\frac{2n+3}{n+4}=2\ne0. \]Therefore,
\[ \boxed{\text{the series diverges}.} \]Practice 3
\[ \lim_{n\to\infty}\frac1{n^2}=0. \]This is necessary for convergence, but the limit alone does not prove that the series converges. Another convergence argument is required.
Practice 4
\[ \lim_{N\to\infty}\left(5-\frac3N\right)=5. \]Therefore the series converges and
\[ \boxed{\sum a_n=5}. \]Practice 5
No. The partial sums alternate between \(-1\) and \(1\), so they do not have a limit. The series diverges.
Practice 6
\[ a_N=S_N-S_{N-1} =\left(2-\frac1N\right) -\left(2-\frac1{N-1}\right). \]Thus
\[ \boxed{a_N=\frac1{N(N-1)}} \]for \(N\ge2\).
Practice 7
Yes. Changing finitely many terms changes later partial sums by a fixed constant but does not change whether the tail converges.
Practice 8
\[ R_{10}=S-S_{10}=7-6.98=\boxed{0.02}. \]Practice 9
Let \(k=n-2\), so \(n=k+2\). Then
\[ \boxed{ \sum_{n=2}^{\infty}\frac1{2^n} =\sum_{k=0}^{\infty}\frac1{2^{k+2}}. } \]Practice 10
The fact that \(a_n\to0\) is only a necessary condition. It does not force the partial sums to converge. The harmonic series is itself the standard example: its terms approach zero, but its partial sums grow without bound.
What to Remember
An infinite series is defined through its partial sums:
\[ S_N=\sum_{n=1}^{N}a_n. \]The series converges to \(S\) when
\[ \boxed{\lim_{N\to\infty}S_N=S}. \]- A series is an accumulation of sequence terms.
- Convergence is determined by the sequence of partial sums.
- If a series converges, then its terms must approach zero.
- The condition \(a_n\to0\) alone does not prove convergence.
- Finite changes do not affect convergence or divergence.
- A series converges exactly when any of its tails converges.
- For a convergent series, the tail remainder \(R_N=S-S_N\) approaches zero.
- Positive-term series have increasing partial sums.
- Linearity works cleanly for series already known to converge.
- Geometric, telescoping, \(p\)-series, and alternating series provide major benchmark families.
Where We Go Next
The definition of convergence tells us what an infinite series means, but we still need efficient ways to recognize and evaluate important patterns.
The next lesson, Geometric and Telescoping Series, develops two of the most useful exact structures. Geometric series reduce repeated multiplication to a closed-form sum, while telescoping series use cancellation inside partial sums to reveal the limiting value.