1 Definition and intuition

Continuity at a point formalizes the idea that a function does not misbehave exactly where it is evaluated. A function is continuous at an input value if its output near that input settles toward the output value the function assigns at the input, with no sudden jump, gap, or unbounded spike occurring at that location.

1.1 Continuity at a point via the ε–δ definition

Let \(f\) be defined on a domain containing \(a\). The function is continuous at \(a\) if for every tolerance \(\varepsilon>0\), there exists a distance \(\delta>0\) such that whenever \(x\) satisfies \(x-a<\delta\), the outputs satisfy \(f(x)-f(a)<\varepsilon\). The condition requires \(x\) to be taken close to \(a\), while \(f(x)\) remains close to \(f(a)\).

This definition captures the “input closeness implies output closeness” principle in a precise, quantitative way.

1.2 Equivalent limit-based definition

A function \(f\) is continuous at \(a\) exactly when \[ \lim_{x\to a} f(x) = f(a), \] provided \(f(a)\) is defined. Here the limit describes what the function approaches as the input tends to \(a\), while continuity requires that the approach matches the actual value at \(a\).

1.3 Geometric and graphical interpretation

Graphically, continuity at \(a\) means the curve of \(y=f(x)\) does not “break” at the vertical line \(x=a\). More precisely, as the graph is traced toward \(x=a\) from both sides, the \(y\)-values approach a single number, and that number is the value drawn at \(x=a\). Isolated points or endpoints can be consistent with continuity, provided the limiting behavior matches the point’s height.

1.4 One-sided continuity (left and right)

Sometimes continuity is analyzed from only one side. The left-hand continuity at \(a\) requires \(\lim_{x\to a^-} f(x)=f(a)\). The right-hand continuity requires \(\lim_{x\to a^+} f(x)=f(a)\). If both one-sided conditions hold, the function is continuous at \(a\) in the usual two-sided sense.

One-sided notions are especially useful when \(a\) is a boundary point of the domain, where only one direction of approach is available.

2 One-sided limits and common pitfalls

Continuity is a statement about behavior near a point. Many failures of continuity can be classified by how the one-sided limits behave or whether the function becomes unbounded.

2.1 Removable discontinuities (“holes”)

A removable discontinuity occurs when the limit exists but differs from the function value: \(\lim_{x\to a} f(x)\) exists, but \(\lim_{x\to a} f(x)\neq f(a)\). In a graph this resembles a “hole” or an incorrectly placed point; redefining \(f(a)\) to equal the limit would produce continuity.

2.2 Jump discontinuities

A jump discontinuity arises when the left and right limits exist but are not equal: \[ \lim_{x\to a^-} f(x)\neq \lim_{x\to a^+} f(x). \] Then the function approaches different values from either side, forcing an abrupt change at \(a\).

2.3 Infinite discontinuities and vertical asymptotes

An infinite discontinuity happens when the limit diverges in magnitude, such as \(\lim_{x\to a} f(x)=\pm\infty\). This corresponds to a vertical asymptote at \(x=a\), where values of \(f(x)\) grow without bound as \(x\) approaches the point.

2.4 Oscillatory behavior near a point

In oscillatory discontinuities, the function fails to settle to a single limit because it keeps swinging between values as \(x\to a\). Even if the function remains bounded, the limit may fail to exist when the limiting superior and inferior do not coincide.

These cases often defeat intuition based solely on “boundedness,” showing that continuity requires convergence to one specific value.

2.5 Discontinuity caused by disagreement between f(a) and lim x→a f(x)

Even when the limit exists cleanly, continuity can still fail if the function’s assigned value at the point is incorrect. The mismatch \(f(a)\neq \lim_{x\to a} f(x)\) leads to discontinuity. This emphasizes that continuity is not only about nearby behavior; it also requires compatibility with the function value at the point.

3 Local behavior and characterizations

Continuity can be studied using building blocks: formulas, compositions, and operations. Several standard characterizations describe when continuity survives algebraic and structural transformations.

3.1 Continuity of polynomials and rational functions (conditions)

Polynomials are continuous at every real number because they are built from continuous operations such as addition and multiplication of \(x\). Rational functions \(p(x)/q(x)\) are continuous wherever \(q(x)\neq 0\), since division by a nonzero continuous denominator preserves continuity. At zeros of the denominator, the function may be undefined or discontinuous.

3.2 Continuity for composite functions

If \(f\) is continuous at \(a\) and \(g\) is continuous at \(f(a)\), then the composite \(g\circ f\) is continuous at \(a\). The idea is that inputs close to \(a\) produce outputs close to \(f(a)\), which then feed into \(g\) in a region where \(g\) behaves continuously.

Composite continuity is one of the main reasons many complicated expressions inherit continuity from simpler parts.

3.3 Continuity and algebraic operations (sum, product, quotient)

If \(f\) and \(g\) are continuous at \(a\), then \(f+g\) and \(f\cdot g\) are continuous at \(a\). For the quotient, if \(g(a)\neq 0\) and \(g\) is continuous at \(a\), then \(f/g\) is continuous at \(a\). These rules rely on the continuity of basic algebraic manipulations when denominators do not vanish.

3.4 Continuity of inverse functions (where defined)

For a strictly monotone continuous function \(f\) defined on an interval, its inverse (defined on the corresponding range) is continuous. More broadly, inverses behave continuously when they exist as functions and the original mapping does not fold the interval back on itself. This connects continuity to the stability of solving equations locally.

Continuity is not limited to real-valued functions of one real variable; it generalizes to neighborhoods, sequences, and broader metric and topological settings.

4.1 Continuity on a domain versus at a point

A function is continuous at a point if the required property holds at that point. It is continuous on a domain if it is continuous at every point in the domain. This distinction is important when the domain is restricted (for example, to an open interval versus a closed interval), since continuity can be discussed point-by-point based on where the function is defined.

4.2 Local continuity and neighborhood-based viewpoint

A neighborhood perspective states that continuity at \(a\) depends only on how \(f\) behaves near \(a\), not on distant points. In the ε–δ formulation, the “nearness” of \(x\) to \(a\) is measured by δ, while output closeness is measured by ε. This local viewpoint underlies many generalizations to higher dimensions and abstract spaces.

4.3 Continuity along sequences approaching the point

A common characterization uses sequences: \(f\) is continuous at \(a\) if for every sequence \((x_n)\) in the domain with \(x_n\to a\), the sequence \(f(x_n)\to f(a)\). This provides an alternative route to proving continuity, especially when ε–δ computations are cumbersome.

4.4 Continuity in metric spaces (generalization to topological ideas)

In a metric space setting, continuity at a point \(a\) means that whenever points \(x\) approach \(a\), their images \(f(x)\) approach \(f(a)\) in the target metric. The same concept can be expressed with sequences: if \(x_n\to a\), then \(f(x_n)\to f(a)\). This generalization highlights that continuity is fundamentally about the preservation of “closeness,” independent of coordinate systems.

5 Continuity of functions built from formulas

Many continuity checks in practice reduce to recognizing which standard functions or constructions are known to be continuous.

5.1 Power functions, exponentials, and logarithms

Power functions \(x^n\) with integer \(n\) are continuous everywhere. For rational exponents, continuity holds on domains where the expression is real-valued (such as \(x&gt;0\) for non-integer rational powers). Exponential functions \(e^x\) are continuous for all real \(x\). The logarithm \(\log x\) is continuous on its domain \(x&gt;0\), but cannot be defined at or across nonpositive values in the real setting.

These functions are frequently used because they arise naturally as solutions to differential equations and as models in applied contexts.

5.2 Trigonometric and inverse trigonometric functions

Trigonometric functions such as \(\sin x\) and \(\cos x\) are continuous everywhere. \(\tan x\) is continuous on intervals that exclude points where \(\cos x=0\). Inverse trigonometric functions (like \(\arcsin x\) and \(\arccos x\)) are continuous on their standard domains (for example, \([-1,1]\) for \(\arcsin\) in the real setting). These continuity domains often match exactly where the input-to-angle relationship is well-defined.

5.3 Piecewise-defined functions and endpoint checks

For piecewise functions, continuity must be verified separately on each region where a different formula applies. At internal breakpoints \(a\), continuity requires that the left-hand and right-hand limits agree and match the assigned value at \(a\). At endpoints of the domain, only the one-sided behavior consistent with domain approach is relevant.

This is a common source of errors: a piece may be continuous within its region while still failing at the junction.

5.4 Using known continuous building blocks

A practical strategy is to recognize a complicated expression as a combination of known continuous functions and operations. If every component is continuous at the point (and any denominators stay nonzero, if division occurs), then the overall expression is continuous by closure properties such as those for sums, products, quotients, and compositions.

This “library of continuous functions” approach reduces the need for repeated foundational proofs.

6 Tests and strategies for verification

Determining continuity often involves choosing a suitable method based on the expression’s form and the point being tested.

6.1 Direct ε–δ verification workflow

To prove continuity at \(a\) directly, one typically starts from \(f(x)-f(a)\) and uses algebraic manipulation to bound it by an expression involving \(x-a\). The goal is to find a δ that guarantees the bound is smaller than ε. For functions built from basic operations, ε–δ proofs can be streamlined by known inequalities and by choosing δ based on required constraints.

Although rigorous, this method may be time-consuming for complex formulas, motivating alternative tests.

6.2 Using limits and limit laws to conclude continuity

If \(\lim_{x\to a} f(x)\) can be computed and shown to equal \(f(a)\), then continuity follows immediately. Limit laws can simplify computations when \(f\) is formed from sums, products, quotients (with nonzero denominators), and compositions of known functions. This approach often uses fewer steps than the ε–δ method while staying fully formal.

6.3 Sequence test for continuity at a point

When the problem involves convergence naturally—such as in contexts using approximations or iterative processes—sequence criteria provide a clean pathway. Proving that for any sequence \(x_n\to a\), the images \(f(x_n)\to f(a)\) can be effective, particularly when direct limit computations are awkward.

Sequence-based reasoning also aligns well with generalizations to metric spaces.

6.4 Squeeze theorem as a continuity aid

If a function is hard to evaluate directly near \(a\), it may be bounded between two other functions whose limits at \(a\) agree. If \(g(x)\le f(x)\le h(x)\) near \(a\) and \(\lim_{x\to a} g(x)=\lim_{x\to a} h(x)=L\), then \(\lim_{x\to a} f(x)=L\). If additionally \(f(a)=L\), continuity at \(a\) results. This technique is especially useful for functions involving absolute values or trigonometric bounds.

6.5 Continuity under parameter changes (basic guidance)

In families of functions depending on a parameter, continuity in the input can persist under certain parameter constraints. A common guideline is to check continuity pointwise with the parameter treated as fixed, and then verify whether operations remain valid (for example, denominators stay away from zero, or domains do not change at the tested point). While full parameter-uniform statements require additional conditions, basic continuity often follows from the same closure properties used for single functions.

7 Continuity versus stronger notions

Continuity is a baseline regularity condition. Stronger properties impose additional structure on how a function behaves, typically strengthening the way changes in input are reflected in changes in output.

7.1 Uniform continuity: local vs global perspective

Uniform continuity strengthens ordinary continuity by requiring the same δ to work for all points in a set, rather than allowing δ to depend on the particular location. A function may be continuous at every point of an interval yet fail to be uniformly continuous across the whole interval. Uniform continuity is particularly relevant on certain unbounded domains, where “local control” does not automatically extend to “global control.”

7.2 Differentiability implies continuity (at a point)

If a function is differentiable at a point, it is necessarily continuous at that point. Differentiability requires the existence of a linear approximation with an error that becomes small compared with \((x-a)\), which forces the change in function values to vanish as \(x\to a\). Thus differentiability provides a stronger guarantee than continuity.

7.3 Lipschitz continuity and Hölder continuity (brief comparison)

Lipschitz continuity bounds the output change by a constant multiple of the input change: \[

f(x)-f(a)\le Lx-a.

\] This implies continuity and gives a strong form of control. Hölder continuity generalizes this by allowing a power smaller than 1: \[

f(x)-f(a)\le Cx-a^\alpha,\quad 0<\alpha<1.

\] Hölder continuity still ensures continuity but permits weaker regulation than Lipschitz. These notions are useful in analysis, approximation, and partial differential equations.

7.4 Continuity of derivatives versus continuity of the original function

The continuity of a derivative is not equivalent to continuity of the original function’s derivative’s existence. If \(f\) is differentiable, then \(f\) is continuous, but its derivative \(f&#039;\) may fail to be continuous. Conversely, a function might have continuous derivatives up to a certain order or might only have weaker regularity. This hierarchy clarifies that different levels of smoothness are independent requirements beyond mere continuity.