Sequences as Functions

A sequence is a function whose domain is a set of consecutive integers (usually the positive integers or natural numbers). The output for input $n$ is called the $n$ -th term, written $a_{n}$ instead of $a (n)$ . Reframing sequences as functions unlocks tools from algebra and calculus — graphs, growth rates, limits, and transformations. Foundational topic in precalculus, calculus, and discrete mathematics; formalized through the 17th–19th century development of functions and limits.

A sequence is an ordered list of numbers, but more precisely it is a function $a : N \to R$ (or a similar discrete domain) that assigns to each index $n$ a unique value $a_{n}$ . The "function" viewpoint matters because it lets us apply everything we know about functions — domains, ranges, monotonicity, boundedness, and ultimately limits — to lists of numbers. This is the bridge between static algebra and the dynamic ideas of calculus, where the limit of a sequence becomes the gateway to convergence, series, and continuity. A surprising fact: although a sequence has infinitely many terms, its entire behavior can often be captured by a single closed-form rule like $a_{n} = 2^{n}$ — and two completely different-looking rules can generate the same sequence.

The Core Insight

A sequence is not just a list — it is a rule of assignment from a discrete domain to a codomain. Writing $a_{n}$ instead of $a (n)$ is pure notation: the index $n$ is the input, and $a_{n}$ is the output. Because the domain is discrete (separated integers, not a continuous interval), the graph of a sequence is a set of isolated points, never a connected curve. Everything about functions — composition, inverses, transformations, growth comparison, and limiting behavior — transfers directly to sequences, with the single restriction that we only ever evaluate at integers.

Formal Statement

Sequence as a Function

A sequence of real numbers is a function

a : N \to R, n \mapsto a_{n}

⟳This is simply the definition of a function whose domain is restricted to

N = {1, 2, 3, \dots}

(some authors start at

0

Here:

$n$ is the index (or position), an element of the domain $N$ .
$a_{n}$ is the $n$ -th term, the function's output, an element of the codomain $R$ .
The whole sequence is denoted $(a_{n})_{n = 1}^{\infty}$ or $(a_{n})$ .
The range ${a_{n} : n \in N}$ is a set; the sequence is an ordered object, so order and repetition matter even though they do not in the range.

A sequence may be specified in three equivalent ways:

Explicit (closed form): a_{n} = f (n)

⟳Direct functional rule — substitute the index to compute any term in one step.

Recursive: a_{n + 1} = g (a_{n}, a_{n - 1}, \dots), a_{1} = c

⟳Define later terms in terms of earlier ones plus one or more initial conditions (base cases).

Uniqueness of Terms

If a sequence is a well-defined function, then each index $n$ produces exactly one term $a_{n}$ . Two sequences $(a_{n})$ and $(b_{n})$ are equal if and only if $a_{n} = b_{n}$ for every $n \in N$ .

Monotonicity and Boundedness

Treating the sequence as a function lets us classify it:

Increasing: a_{n + 1} > a_{n} \forall n; Decreasing: a_{n + 1} < a_{n} \forall n

⟳Same definition as monotonic functions, applied only at integer inputs by comparing consecutive terms.

Bounded: \exists M > 0 such that ∣ a_{n} ∣ \leq M \forall n \in N

⟳A single constant

M

caps the absolute value of every term — the analogue of a bounded function.

Why It Holds / Derivation

Why the Function Framework Is Valid

We justify that the list-and-function views coincide.

A list defines a function. Given an ordered list $a_{1}, a_{2}, a_{3}, \dots$ , define $a (n)$ = the $n$ -th entry. Since each position has one entry, $a$ is single-valued — a genuine function on $N$ .

A function defines a list. Given $a : N \to R$ , write the outputs in index order: $a (1), a (2), a (3), \dots$ . This is an ordered list.

The two operations are inverse. Going list $\to$ function $\to$ list returns the original list, and similarly for function $\to$ list $\to$ function. Hence the descriptions are interchangeable.

Transfer of structure. Because $a$ is a function, every function property defined pointwise — comparison of values, bounds, and limiting behavior — applies. In particular, $a_{n}$ may be substituted into any real function: $sin (a_{n})$ , $a_{n}^{2}$ , etc., are themselves sequences (compositions of functions). $■$

Leonhard Euler | 1707–1783

Euler systematized the modern notation and treatment of sequences and series, popularizing the use of functional notation $f (x)$ and treating infinite sequences with the freedom of functions. His work on the harmonic series, the constant $e$ as a limit of $(1 + 1/ n)^{n}$ , and infinite products embodies the function-centric view of sequences. Euler's manipulations — sometimes formal and bold — drove the 18th-century explosion of results that later mathematicians made rigorous through the theory of limits.

Building Intuition

Think of a sequence as a vending machine for numbers. You press a button labeled with an integer $n$ , and out comes exactly one number $a_{n}$ . You can never press button $2.5$ — the machine only accepts whole-number buttons. That restriction (a discrete domain) is the only difference between a sequence and an ordinary function $f (x)$ .

This is why the graph of a sequence looks like a dotted constellation rather than a smooth curve: you can only plot points above $n = 1, 2, 3, \dots$ , never in between. If you mentally "fill in the gaps" with a continuous function $f (x)$ that agrees at the integers, you get a powerful tool: the curve's behavior as $x \to \infty$ tells you the sequence's behavior as $n \to \infty$ .

A sequence as isolated points: $a_n = 1 + 1/n$ versus its continuous companion $f(x) = 1 + 1/x$

Worked Examples

Explicit Rule and Term Computation

Given: A sequence defined by the explicit function

a_{n} = 3 n - 2

for

n \in N

. Find: (a) the first four terms, (b) the 50th term, (c) which index gives the value

148

. Steps:

(a) Evaluate the function at $n = 1, 2, 3, 4$ :

$a_{1} = 3 (1) - 2 = 1$
$a_{2} = 3 (2) - 2 = 4$
$a_{3} = 3 (3) - 2 = 7$
$a_{4} = 3 (4) - 2 = 10$

(b) Substitute $n = 50$ :

a_{50} = 3 (50) - 2 = 150 - 2 = 148

This is an arithmetic sequence (constant difference $3$ ); the function is linear in $n$ .

Answer:

a_{1}, \dots, a_{4} = 1, 4, 7, 10; a_{50} = 148; n = 50

Recursive Rule Converted and a Limit Read Off

Given: A recursive sequence

a_{n + 1} = \frac{1}{2} a_{n} + 3

, with

a_{1} = 2

. Find: (a) the first four terms, (b) the long-run value (limit) by treating it as a fixed point. Steps:

(a) Apply the recursion:

$a_{1} = 2$
$a_{2} = \frac{1}{2} (2) + 3 = 4$
$a_{3} = \frac{1}{2} (4) + 3 = 5$
$a_{4} = \frac{1}{2} (5) + 3 = 5.5$

The terms increase and appear to approach a ceiling.

(b) If the sequence converges to a limit $L$ , then for large $n$ both $a_{n + 1}$ and $a_{n}$ approach $L$ . Substitute:

L = \frac{1}{2} L + 3 \Rightarrow \frac{1}{2} L = 3 \Rightarrow L = 6

Check: terms $2, 4, 5, 5.5, 5.75, \dots$ climb toward $6$ , never exceeding it — consistent with a bounded, increasing sequence.

Answer:

a_{1}, \dots, a_{4} = 2, 4, 5, 5.5; n \to \infty lim a_{n} = 6

Common Misconceptions

A sequence is not a continuous curve. Even when you write

a_{n} = 1/ n

, you must not evaluate it at

n = 2.5

or draw a smooth line through the points. The domain is the integers only. The continuous companion

f (x) = 1/ x

is a separate function used as an analytical aid — it is not the sequence itself. The range is not the sequence. The sequence

1, 0, 1, 0, 1, 0, \dots

has range

{0, 1}

— just two numbers — but it is an infinite, non-constant sequence. Order and repetition are essential to the sequence even though they vanish in the range set. Claim (false): "If a closed-form formula gives the first few terms correctly, it is the only sequence fitting them." Refutation: Consider the partial list

2, 4, 8, \dots

. The natural guess is

a_{n} = 2^{n}

, giving

16, 32, \dots

. But the formula

a_{n} = 2^{n} + (n - 1) (n - 2) (n - 3)

also yields

2, 4, 8

for

n = 1, 2, 3

, then continues as

16 + 6 = 22, \dots

— a completely different sequence. A finite list of terms never uniquely determines a sequence; the defining function (rule) must be specified.

Connections

Viewing sequences as functions ties together several neighbouring topics. Arithmetic Sequences are sequences whose generating function is linear in $n$ ( $a_{n} = a_{1} + (n - 1) d$ ), and Geometric Sequences are those whose function is exponential in $n$ ( $a_{n} = a_{1} r^{n - 1}$ ). The classification of a sequence as increasing, decreasing, or bounded is the discrete shadow of the same concepts for Functions and Their Properties. Pushing the index to infinity leads directly to the Limit of a Sequence, and summing the terms produces an Infinite Series. Recursive definitions connect to Mathematical Induction, the proof technique tailored to statements indexed by $N$ .

Real-World Applications

Compound interest (finance): The balance after $n$ periods is the geometric sequence $A_{n} = P (1 + r)^{n}$ — a function of the integer period count.
Population biology: Discrete generation models like the logistic map $x_{n + 1} = r x_{n} (1 - x_{n})$ are recursively defined sequences used to model animal populations year by year.
Computer science / algorithms: The number of operations an algorithm performs on input size $n$ is a sequence; Big-O analysis compares the growth rate of these sequences-as-functions.
Digital signal processing: A sampled audio or sensor signal is literally a sequence $x_{n} = x (n Δ t)$ — a function evaluated at discrete time steps.
Amortization schedules (loans): Monthly remaining balance follows a recursive sequence combining interest accrual and fixed payment, exactly like the worked fixed-point example.
Numerical methods: Newton's method generates a sequence $x_{n + 1} = x_{n} - f (x_{n}) / f^{'} (x_{n})$ that converges to a root; its limit behavior is read off as a function of the iteration index.
Fibonacci modeling (nature): Branching patterns, phyllotaxis, and rabbit-population puzzles use the recursive sequence $F_{n + 1} = F_{n} + F_{n - 1}$ .

Historical Development

c. 300 BCE — Euclid: Elements Book IX treats geometric progressions and proves the sum formula, an early systematic study of structured number lists.
c. 1202 — Fibonacci (Leonardo of Pisa): Liber Abaci introduces the rabbit problem, popularizing the recursively defined Fibonacci sequence in the West.
1655 — John Wallis: Arithmetica Infinitorum uses infinite sequences and interpolation, foreshadowing limits.
1668 — Nicholas Mercator / James Gregory: Develop infinite series and the sequences of partial sums, linking sequences to function expansions.
1734 — Leonhard Euler: Establishes the limit $e = lim (1 + 1/ n)^{n}$ and treats sequences with full functional notation.
1821 — Augustin-Louis Cauchy: Cours d'analyse gives the first rigorous $ε$ -style definition of the limit of a sequence, founding modern analysis.
1872 — Karl Weierstrass & Eduard Heine: Formalize convergence and the Bolzano–Weierstrass theorem; the function view of sequences becomes fully rigorous.
1888 — Richard Dedekind / Giuseppe Peano (1889): Axiomatize $N$ , providing the precise domain on which sequences-as-functions are defined.

Subtleties & Limits

Several conceptual subtleties survive at higher levels. Indexing conventions differ: some define $N$ to start at $0$ , others at $1$ , which shifts every closed-form formula — there is no universal convention. Existence of a closed form is not guaranteed: many recursively defined sequences (and famous ones from number theory, like the prime-counting sequence) have no known elementary explicit formula, and proving that none exists can be deeply hard. The Collatz sequence $a_{n + 1} = a_{n} /2$ (if even) or $3 a_{n} + 1$ (if odd) is perfectly well-defined as a function, yet whether it always reaches $1$ remains an unsolved problem. Finally, distinguishing a sequence (function on $N$ ) from a net or generalized sequence (function on an arbitrary directed set) matters in advanced topology, where ordinary sequences are no longer sufficient to detect continuity.