ε–δ definition - Ciallo Wiki

1 Introduction to ε–δ Language

1.1 Motivation from informal “closeness” ideas

In real analysis, many intuitive arguments rely on statements like “make the input close to the target, and the output becomes close as well.” While persuasive informally, such reasoning can hide where the dependence lies—how “close” the input must be to guarantee a specified “closeness” of outputs. The ε–δ method formalizes this relationship so that proofs can be checked mechanically: the tolerance for the output is chosen first, and the proof must produce an input tolerance that enforces it.

1.2 Logical structure: quantifiers and implication

The core pattern has the form: for every allowed error level ε, there exists a corresponding restriction δ so that whenever the input is within δ of a reference value, the function values differ from the target by less than ε. The logical content is often captured by a quantifier sequence such as “for all ε > 0, there exists δ > 0 such that … implies ….” This quantifier order matters because it encodes that the analyst must respond to an adversarial choice of ε by providing a suitable δ.

1.3 Notation conventions and typical pitfalls

Common conventions include using absolute value to measure distance on the real line (e.g.,	x − a	) and writing inequalities like	f(x) − L	< ε. Typical pitfalls involve mixing up ε and δ roles, choosing δ before ε, omitting necessary conditions such as x ≠ a in limit definitions, or letting δ depend on x (which is not permitted). Another frequent issue is failing to state the “whenever” clause properly, especially when both lower and upper bounds on	x − a	appear.

2 ε–δ Definition of Limits

2.1 Limit of a function at a point

2.1.1 Formal statement and quantifier order

Let f be defined on a punctured neighborhood of a (i.e., for all x near a except possibly at a). The statement that the limit of f(x) as x approaches a equals L is expressed as:

- For every ε > 0, there exists δ > 0 such that for all x, if 0 <	x − a	< δ, then	f(x) − L	< ε.

This formulation captures two essential ideas: (1) the conclusion must hold for every input sufficiently close to a but not equal to it, and (2) the closeness threshold in the input is allowed to depend on ε but not on x.

2.1.1.1 Relationship between 0 < |x − a| < δ and |f(x) − L| < ε

The condition 0 <	x − a	< δ excludes the case x = a, reflecting that a limit describes behavior near the point rather than necessarily at the point. The implication links this restricted proximity in the domain to a controlled error in the function values, quantified by ε. In proofs, one typically starts from 0 <	x − a	< δ, derives bounds on the expression governing	f(x) − L	, and then verifies that the resulting estimate is strictly less than ε.

2.1.2 Examples illustrating correct δ–ε selection

A standard strategy for correct selection of δ is to bound	f(x) − L	by an expression that involves	x − a	, then ensure that this bound is below ε by choosing δ accordingly. For instance, if one can show	f(x) − L	≤ C	x − a	for some constant C > 0, then it suffices to pick δ = ε/C (or any smaller positive choice). This illustrates a general proof pattern: translate the goal “	f(x) − L	< ε” into a condition on	x − a	, and then choose δ so that the premise 0 <	x − a	< δ guarantees it.

2.2 One-sided limits (brief extension)

One-sided limits restrict x to approach a from one side. For the right-hand limit, the condition becomes 0 < x − a < δ, and the conclusion remains	f(x) − L	< ε. Similarly, for the left-hand limit one uses 0 < a − x < δ. These variations are useful when a function behaves differently on either side of the reference point.

2.3 Limits in metric form (optional viewpoint)

The ε–δ framework can be reinterpreted using metric spaces, where “distance” is expressed by a metric d(·,·) rather than absolute value. In that setting, the statement “f(x) approaches L as x approaches a” becomes: for every ε > 0, there exists δ > 0 such that if d(x,a) < δ and x ≠ a, then d(f(x),L) < ε. Over the real numbers with the usual metric, this reduces to the standard absolute-value formulation.

2.4 Common limit patterns and proof strategies

Many limit proofs follow repeatable structures. Typical patterns include:

Algebraic simplification: rewriting f(x) − L into factors that contain (x − a), then bounding remaining terms.
Difference quotients: especially in limits involving expressions like (f(x) − f(a))/(x − a).
Squeeze arguments: bounding f(x) between two expressions whose limits are known to converge to the same L, then using the ε–δ bounds from those limits.
Continuity substitutions: when a limit involves a continuous function, one may reduce the problem to the limit of the inner expression, provided the needed continuity is established.

3 ε–δ Definition of Continuity

3.1 Continuity at a point

3.1.1 Equivalent formulations (ε–δ vs. sequences, brief)

A function f is continuous at a if its value at a matches the limit value as x approaches a. In ε–δ form this is:

- For every ε > 0, there exists δ > 0 such that if	x − a	< δ, then	f(x) − f(a)	< ε.

An equivalent characterization uses sequences: f is continuous at a if for every sequence (x_n) converging to a, the sequence f(x_n) converges to f(a). In real analysis, these formulations are equivalent for functions defined on subsets of the real numbers with the usual metric.

3.1.2 Role of δ depending on both ε and the point

Continuity at a point requires a δ that may depend on both ε and the specific point a. The definition does not ask for a single δ that works uniformly for all points; rather, it allows δ to vary from point to point as long as the dependence is consistent with the chosen ε and the fixed a under consideration.

3.2 Continuity on sets and global statements

Continuity on a set means continuity at every point in that set. For example, f is continuous on an interval if it satisfies the ε–δ condition at each point of the interval. Many results in analysis rely on such “pointwise continuity” to ensure that limit operations can be performed locally, and then combined across the domain when additional compactness or uniformity hypotheses are present.

3.3 Discontinuities: what ε–δ fails to provide

If the ε–δ condition cannot be satisfied at a point, the function is discontinuous there. The failure is not merely “the inequality doesn’t hold for some δ,” but rather that there exists an output tolerance ε0 such that no matter how small a neighborhood δ is chosen, inputs within that neighborhood produce outputs that differ from f(a) by at least ε0. This quantifier reversal—producing an ε0 that defeats every δ—is the hallmark of discontinuity proofs.

4 ε–δ Definition of Derivatives

4.1 Derivative via the limit definition

4.1.1 Difference quotient in ε–δ form

The derivative of f at a, denoted f′(a), is defined by the limit of the difference quotient. In ε–δ language, f′(a)=M means:

- For every ε > 0, there exists δ > 0 such that if 0 <	h	< δ, then

\left	\frac{f(a+h)-f(a)}{h} - M\right	< \varepsilon.

Here h represents the increment from a. The condition 0 <	h	< δ mirrors the “punctured neighborhood” concept: the quotient is undefined at h = 0, and the derivative concerns the behavior for small nonzero increments.

4.1.2 Interpreting small h and controlled error

In practice, one reads the derivative condition as: when h is sufficiently small, the secant slope between a and a+h approximates M with error less than ε. The proof usually involves rewriting the difference between the quotient and M into an expression that can be bounded using algebraic identities, known limit facts, or established continuity of related functions.

4.2 Using ε–δ to prove differentiability

To prove differentiability at a, one may start from a candidate value M and then show the ε–δ inequality for the difference quotient. This often entails:

Simplifying the expression

\[ \frac{f(a+h)-f(a)}{h} - M \] into a form where factors of h can be controlled.

Bounding the remainder terms by quantities that tend to 0 as h → 0.
Choosing δ as a function of ε to force the bound below ε.

The ε–δ derivative definition is stricter than informal reasoning about tangent lines because it requires a uniform error control for all sufficiently small increments.

4.3 Connection to tangent-line approximation

When the derivative exists, the function admits a best linear approximation near the point: for small h, the increment f(a+h) − f(a) is close to Mh. The tangent line y = f(a) + M(x − a) captures the first-order behavior, and the ε–δ derivative condition ensures that the linear approximation error can be made arbitrarily small by taking x sufficiently close to a.

5 Techniques for Constructing δ from ε

5.1 Algebraic manipulation and bounding

Many constructions of δ begin with algebraic transformations that isolate the expression to be bounded. For limits and continuity proofs, one typically rewrites	f(x) − L	(or the derivative quotient error) into a product or quotient involving	x − a	. Then the inequality is transformed into something like “	x − a	< something involving ε,” which directly suggests a choice for δ.

5.2 Inequalities and “choose δ = …” tactics

After obtaining an upper bound such as	f(x) − L	≤ G(x,a) where G involves	x − a	, a common tactic is to enforce G < ε. If G has the form C	x − a	, the choice δ = ε/C is natural. If G has the form C	x − a	^k with k > 0, then δ = (ε/C)^(1/k) may be appropriate. The general principle is to select δ so that the worst-case bound under the premise is below ε.

5.3 Handling absolute values and triangle inequality

Absolute values are central to ε–δ proofs. The triangle inequality, \[

u+v	\le	u	+	v	,

\] and related bounds help separate complicated expressions into manageable pieces. Proofs often use these inequalities repeatedly to convert a complicated target into a sum of simpler terms that can each be forced below ε/2, ε/3, and so on, by adjusting δ.

5.4 Managing constants and scaling factors

Constants arise from algebraic operations such as expanding polynomials, factoring, or applying Lipschitz-like bounds. A typical approach is to track each constant carefully so that δ can be chosen in a way that accounts for all scaling effects. When multiple constants multiply, one often bounds the product by a single constant times the relevant power of	x − a	or	h	, ensuring the final estimate remains controlled.

5.5 Proof templates for standard functions

Standard templates exist for common functions:

Polynomials: rewrite differences to factor out (x − a) and bound the remaining polynomial term using a uniform bound near a.
Rational functions: control the numerator via algebra and ensure denominators stay away from 0 by choosing δ small enough.
Composition with continuous functions: use known continuity to bound inner differences, then transfer the bound through the outer function via an ε–δ relationship.

These templates reduce repeated work and emphasize the shared structure of ε–δ reasoning.

6 Proofs Using the ε–δ Definition

6.1 Direct proof workflows

A direct ε–δ proof usually follows a disciplined workflow:

Start with ε > 0 and identify the quantity that must be made smaller than ε.

2. Assume the premise (e.g., 0 <	x − a	< δ) and derive bounds on the relevant expression.

Choose δ so that the derived bound implies the desired inequality.
Conclude by verifying that the implication holds for all x satisfying the premise.

A common best practice is to explicitly state the chosen δ and show the final inequality step-by-step.

6.2 Contradiction and counterexample viewpoints

Instead of constructing δ, one can prove a limit or continuity statement by contradiction. For example, to show that a limit equals L, one negates the definition: if the limit were not L, then there would exist some ε0 > 0 such that for every δ > 0, one finds x within δ of a (but not equal) where	f(x) − L	≥ ε0. This often leads to the construction of a sequence approaching a while violating the intended conclusion, contradicting the assumed properties. This approach is especially effective when a direct δ selection is algebraically cumbersome.

6.3 Using previous results: limit laws and continuity principles

ε–δ proofs are often built on previously established lemmas:

Limit laws allow one to combine limits of sums, products, and quotients (with nonzero limiting denominators).
Continuity principles allow the replacement of limits through continuous functions: if g is continuous at L, then g(x) approaches g(L) whenever x approaches L.

Even when such laws are used, a rigorous proof typically still relies on ε–δ reasoning at some stage to ensure that every step preserves the quantifier structure.

6.4 Composition and substitution in ε–δ proofs

When f(x) is expressed as a composition, proofs frequently proceed in layers. One first ensures that the inner expression is within some δ1 of the relevant limit point, then uses continuity of the outer function to translate that closeness into an ε bound for the full expression. This method highlights that δ can depend on ε through intermediate tolerances, and that selecting δ1 correctly is the key to closing the argument.

7.1 Uniform continuity vs. pointwise continuity (ε–δ contrast)

Pointwise continuity at each point a allows δ to depend on both ε and a. Uniform continuity strengthens the requirement by requiring a single δ that works for all points in a set: for every ε > 0, there exists δ > 0 such that for any x and y in the set with	x − y	< δ, one has	f(x) − f(y)	< ε. This difference can be crucial on unbounded domains, where pointwise control may fail to be uniform.

7.2 Cauchy-type conditions and ε–δ parallels

Cauchy-type characterizations replace “convergence to a limit” with “internal closeness of inputs.” For example, a sequence converges if and only if it is Cauchy in the ε sense: for every ε > 0 there exists N such that for m,n ≥ N, the distance between terms is less than ε. In real analysis, these Cauchy perspectives often parallel the ε–δ logic by centering the idea of guaranteeing small differences under appropriate restrictions.

7.3 Generalization to metric spaces (high-level overview)

In metric spaces, limits and continuity can be defined using the metric d rather than absolute value. The same ε–δ pattern reappears: “for every ε, there exists δ such that distance in the domain smaller than δ implies distance in the codomain smaller than ε.” This generalization extends the methodology beyond the real line and supports a unified treatment of convergence and continuity across varied mathematical settings.

8 Worked Examples (Representative)

8.1 Polynomial limits and continuity

Consider a polynomial p(x). Polynomials are continuous everywhere. An ε–δ proof can be organized by factoring differences: \[ p(x)-p(a) = (x-a)q(x,a), \] where q(x,a) is another polynomial expression. Near x = a, q(x,a) can be bounded by a constant M, yielding \[

p(x)-p(a)	\le M	x-a	.

Choosing δ = ε/M ensures	p(x) − p(a)	< ε whenever	x − a	< δ.

8.2 Rational functions with nonzero denominators

For a rational function r(x) = \frac{g(x)}{h(x)} where h(a) ≠ 0, continuity at a can be shown by controlling the denominator. The proof typically splits the error: \[

\left	\frac{g(x)}{h(x)} - \frac{g(a)}{h(a)}\right

into terms involving	g(x) − g(a)	and	h(x) − h(a)	, then uses continuity of g and h along with a bound that keeps h(x) away from 0. One ensures	h(x)	stays above a positive fraction of	h(a)	by choosing δ small enough, then uses the resulting denominator bound to convert numerator smallness into overall smallness.

8.3 Trigonometric function limit/continuity sketches

Trigonometric functions are continuous everywhere. In ε–δ proofs, one can use inequalities that relate differences in function values to differences in inputs. For example, the identity and estimates behind bounds like \[

\sin x - \sin y	\le	x-y

provide an immediate route: once the input closeness	x − y	< δ is set, the output closeness follows with ε = δ or with a comparable scaling. When direct inequalities are not as straightforward, one may use addition formulas and triangle inequality to derive an appropriate estimate.

8.4 Derivative of simple functions (outline-level)

For a basic example such as f(x) = x^2, the derivative at a is derived from the difference quotient: \[ \frac{(a+h)^2 - a^2}{h} = \frac{2ah + h^2}{h} = 2a + h. \]

The ε–δ step then targets	(2a + h) − 2a	=	h	< ε. Choosing δ = ε makes the proof immediate: if 0 <	h	< δ, then the quotient deviates from 2a by less than ε. More complicated derivatives follow the same blueprint but require more algebra to rewrite the quotient error into something manageable and then choose δ to dominate the remaining terms.

9 Conceptual Takeaways

9.1 Why quantifier order matters

The essential feature of ε–δ reasoning is that ε is chosen first and δ must be produced afterward. This ordering reflects a worst-case mindset: the definition must succeed no matter how tightly the result is demanded. Swapping quantifiers can turn a true statement into a false one, so careful attention to “for every ε” followed by “there exists δ” is fundamental.

9.2 How to read and write ε–δ proofs clearly

Reading ε–δ proofs involves tracking the implication: the premise defines what is assumed, and the conclusion gives what must be shown. Writing them clearly means stating ε at the beginning, selecting δ explicitly, and then showing how the assumption “	input − point	< δ” leads to the desired “	output − target	< ε.” Well-structured proofs reduce ambiguity about what depends on what.

9.3 Typical errors and how to avoid them

Common mistakes include using a δ that depends on the particular x being considered, selecting δ without confirming that all inequalities hold, or neglecting the strict inequality versus non-strict inequality distinctions. Another error is failing to account for the “punctured” condition (such as 0 <	x − a	) when proving limits. Avoiding these requires close attention to the definition being used and disciplined logical referencing of each step.

9.4 Practice checklist for ε–δ problem solving

A practical checklist includes:

Identify whether the task is a limit, continuity, or derivative statement.
Write the exact quantifier form required by that definition.
Begin with ε and isolate an expression that must be made smaller than ε.

- Use algebra and inequalities to bound that expression in terms of	x − a	or	h	.

Choose δ explicitly and verify the final inequality for all allowed inputs.
If construction is difficult, consider contradiction or sequence-based viewpoints.