1 Cauchy sequences: general definition in metric and normed settings
A Cauchy sequence is a sequence whose terms eventually become arbitrarily close to each other according to a specified notion of distance. This captures the idea of “internal convergence” without requiring the existence of a limit within the ambient space.
1.1 Metric formulation
Let \((X,d)\) be a metric space and \((x_n)\) a sequence in \(X\). The sequence is called Cauchy if for every \(\varepsilon>0\) there exists \(N\) such that for all \(n,m\ge N\), \[ d(x_n,x_m)<\varepsilon. \] Intuitively, the tail of the sequence becomes tight: all sufficiently late terms lie within a small neighborhood of each other.
1.2 Normed-space formulation
| If \(X\) is a normed vector space with norm \(\|\cdot\|\), the metric can be taken as \(d(x,y)=\|x-y\|\). Then \((x_n)\) is Cauchy if for every \(\varepsilon>0\) there exists \(N\) such that for all \(n,m\ge N\), |
|---|
\[
| \|x_n-x_m\|<\varepsilon. |
|---|
\] In analysis, this version is especially convenient for function spaces, where norms encode size or regularity.
1.3 Equivalent characterizations via differences \(f_n − f_m\)
In function spaces, Cauchy-ness is typically expressed directly in terms of increments \(f_n-f_m\). One says \((f_n)\) is Cauchy in a normed space if \[
| \|f_n-f_m\|\to 0 \quad \text{as } n,m\to\infty, |
|---|
\] meaning that for large indices the difference functions become small in the norm. In metric language, this is the same as requiring the metric distance between any two late terms to be small.
1.4 Relation to convergence and uniqueness of limits
Every convergent sequence is Cauchy. If \(x_n\to x\) in \(X\), then for large \(n,m\), both \(x_n\) and \(x_m\) are close to \(x\), forcing \(d(x_n,x_m)\) to be small.
In metric spaces, limits are unique when they exist. Consequently, in complete settings (where every Cauchy sequence converges), the limit of a Cauchy sequence is determined by the sequence rather than by auxiliary choices.
2 Function spaces and their induced “distance”
A function space becomes a metric or normed space by choosing how to measure the “distance” between two functions. Different choices can change which sequences are Cauchy and what convergence means.
2.1 L^p spaces and the L^p metric/norm
For \(1\le p<\infty\), the \(L^p\)-norm on a suitable measure space is \[
| \|f\|_{L^p}=\left(\int | f | ^p\right)^{1/p}. |
|---|
\]
| A sequence \((f_n)\) is Cauchy in \(L^p\) if \(\|f_n-f_m\|_{L^p}\to 0\). |
|---|
2.1.1 Pointwise representatives vs equivalence classes
In \(L^p\) spaces, functions are often treated as equivalence classes: two functions that differ only on a set of measure zero represent the same element. This means statements about convergence or Cauchy-ness should be understood in the norm sense, not as a claim about matching pointwise values everywhere.
2.2 Supremum norm and spaces of bounded functions
| In spaces where the sup norm is used—commonly denoted \(\|f\|_\infty=\sup | f(x) | \)—closeness is measured uniformly over the domain. A sequence is Cauchy in the sup norm when |
|---|
\[
| \sup_x | f_n(x)-f_m(x) | \to 0. |
|---|
\] This is typically stronger than \(L^p\)-type control because it controls the maximum deviation pointwise (in an appropriate sense).
2.2.1 Uniform (Cauchy) sequences
Uniform Cauchy-ness corresponds to the idea that the entire family becomes uniformly close: beyond some index, every function in the tail differs from every other function by less than \(\varepsilon\) at all points. When the underlying space is complete in that norm (for instance, bounded functions on a compact domain with an appropriate codomain), such sequences converge uniformly.
2.3 Sobolev spaces and weak norms
| Sobolev spaces measure not only function values but also derivatives in a generalized (weak) sense. A typical Sobolev norm combines \(L^p\) control of a function and its (weak) derivatives up to some order. For example, the \(H^1\) norm on a domain involves both \(\|f\|_{L^2}\) and \(\|\nabla f\|_{L^2}\). |
|---|
2.3.1 Gradient-based control and examples
Because Sobolev norms incorporate gradient information, a sequence can be controlled in a way that rules out wild oscillations that would still be small pointwise. As a consequence, Sobolev Cauchy sequences often admit limits with improved regularity properties compared with merely \(L^p\)-bounded sequences.
2.4 Spaces of continuous functions with various norms
When the space consists of continuous functions, common norms include the sup norm and norms derived from measures or derivatives (in specialized settings). The Cauchy property depends on the selected norm: sup-norm Cauchy-ness enforces uniform behavior, while weaker norms may allow discrepancies on sets of small measure or at the level of pointwise representatives.
2.5 Inner-product spaces and the Hilbert-space viewpoint
| If \(X\) is an inner-product space with inner product \(\langle\cdot,\cdot\rangle\), the induced norm is \(\|x\|=\sqrt{\langle x,x\rangle}\). Then Cauchy-ness can be studied via inner-product identities and energy estimates. |
|---|
2.5.1 Induced norm from the inner product
In Hilbert spaces, many arguments rely on orthogonality and projections. Cauchy sequences interact naturally with these tools because the norm depends quadratically on the inner product.
3 Completeness and convergence of Cauchy sequences
Completeness is the structural property that connects Cauchy behavior to the existence of a limit within the same space. In analysis, it is the bridge that turns “approximation without a limit” into “actual solution.”
3.1 Banach spaces: every Cauchy sequence converges
A Banach space is a complete normed vector space. By definition, every Cauchy sequence in the norm converges to an element of the space. This is a cornerstone for proving existence of limits and constructing solutions via limits of approximations.
3.2 Hilbert spaces: Cauchy ⇔ convergence under completeness
Hilbert spaces are complete inner-product spaces, hence also Banach spaces. Therefore, in a Hilbert space, Cauchy sequences are equivalent to convergent sequences. The additional structure of an inner product supports stronger theorems, such as those involving projections and orthogonal decompositions.
3.3 How completion works for function spaces
When a function space is incomplete under a chosen norm, one can often enlarge it to a completion, adding “missing” limit objects. In \(L^p\)-type settings this completion is typically achieved by identifying limits defined via Cauchy behavior in the metric and constructing an appropriate space where these limits exist. This motivates why analysis often begins with a dense subspace (e.g., smooth functions) and then completes it in the desired norm.
3.4 Counterexamples in non-complete settings
If the ambient space is not complete, a Cauchy sequence may converge only in a larger space but not within the original one. In function settings, this may manifest as limits that fail to belong to the original class of functions. A classic pattern is that the approximating functions stay within the space while the would-be limit exits it due to insufficient regularity or missing integrability.
4 Practical criteria for verifying the Cauchy property
| Rather than checking the definition directly, analysts use inequalities, compactness-like ideas, and structural estimates to show that \(\|f_n-f_m\|\) becomes small for large indices. |
|---|
4.1 Estimating norms of increments
| A typical workflow is to bound \(\|f_n-f_m\|\) using known estimates about each term and how the differences behave. In many applications, \(f_n-f_m\) simplifies due to telescoping sums or stability properties of the underlying construction. |
|---|
4.2 Using inequalities (triangle inequality, Hölder, Minkowski)
Core tools include:
- Triangle inequality to relate differences to intermediate quantities.
- Hölder’s inequality to estimate integrals of products and convert between norms.
- Minkowski’s inequality to control norms of sums in \(L^p\) settings.
| These inequalities allow one to bound \(\|f_n-f_m\|\) without computing it explicitly. |
|---|
4.3 Compactness-style arguments yielding Cauchy sequences
Compactness arguments can sometimes upgrade boundedness and additional structure into convergence properties. While compactness often appears in extracting convergent subsequences, similar ideas can be used to show that entire sequences behave Cauchy in certain norms, especially when combined with uniqueness of limits.
4.4 Dense subspaces and approximation strategies
A common strategy is to approximate in a dense subspace where computations are easier (e.g., smooth functions) and then show that the approximations form a Cauchy sequence in the target norm. Completion then guarantees that the limit exists in the completed space even if it is not represented by a smooth element.
5 Mode of convergence induced by the function space
“Convergence” is not a single notion; it depends on the norm or metric that defines the function space. This section clarifies how different induced modes of convergence relate to Cauchy properties.
5.1 Strong (norm) convergence versus weaker notions
Norm convergence implies Cauchy-ness, and in complete spaces implies convergence to a specific element. Weaker notions (such as convergence almost everywhere or convergence in measure) may not ensure Cauchy-ness under strong norms like \(L^p\), because they ignore certain global quantitative features.
5.2 Almost everywhere convergence vs L^p convergence
Almost everywhere convergence concerns pointwise behavior on the domain, typically outside a null set. By contrast, \(L^p\) convergence measures averaged \(p\)-power deviations. A sequence might converge almost everywhere without converging in \(L^p\), and conversely \(L^p\) convergence often yields subsequences that converge almost everywhere (under standard hypotheses).
5.3 Uniform convergence and equicontinuity links
Uniform convergence corresponds to convergence in the sup norm. When functions are continuous, equicontinuity can help relate compactness to uniform convergence via classic principles (such as those stemming from Arzelà–Ascoli). In such settings, Cauchy sequences in the sup norm naturally connect to uniform control.
5.4 Weak convergence for sequences that are bounded but not Cauchy
In Hilbert spaces, bounded sequences may fail to be Cauchy while still having weakly convergent subsequences. Weak convergence is weaker than norm convergence: it tests against inner products with fixed elements rather than measuring the full norm of differences. Thus, weak convergence does not generally imply that a sequence is Cauchy in the strong norm.
6 Standard examples across common function spaces
Examples clarify how Cauchy behavior depends on the chosen norm and what kinds of limits are admitted by each space.
6.1 Cauchy sequences in C([a,b]) under the sup norm
| For continuous functions on a compact interval, Cauchy-ness in the sup norm corresponds to uniform convergence of the sequence and to existence of a continuous limit. If \((f_n)\) is Cauchy in \(\|\cdot\|_\infty\), then for every \(\varepsilon>0\) the tail functions are within \(\varepsilon\) uniformly, forcing a unique uniform limit that remains continuous. |
|---|
6.2 Cauchy sequences in L^p that converge in norm
| In \(L^p\) spaces with \(1\le p<\infty\), completeness ensures that any \(L^p\)-Cauchy sequence converges in \(L^p\)-norm to an element of the space. The limit is defined up to almost everywhere equality and reflects the averaged decay of \( | f_n-f_m | ^p\). |
|---|
6.3 Sequences of smooth functions approximating Sobolev limits
A standard pattern in Sobolev analysis is that smooth functions can approximate rougher Sobolev functions in the Sobolev norm. If a sequence of smooth approximations is shown to be Cauchy in a Sobolev norm, completeness implies convergence to a Sobolev limit. This underlies many “approximate by smooth” constructions used in partial differential equations.
6.4 Sequences that are Cauchy in a weaker norm but not in a stronger one
Different norms can disagree sharply on whether a sequence is Cauchy. For instance, a sequence may be Cauchy in an \(L^p\) norm yet fail to be Cauchy in the sup norm because pointwise oscillations persist even if their average magnitude decreases. Such examples illustrate that strengthening the norm can destroy the Cauchy property.
7 Consequences and applications
Cauchy sequences support existence proofs and stability statements throughout analysis, particularly when solutions are obtained as limits of approximating objects.
7.1 Existence of limits for analytic constructions
Many constructions define objects as limits of sequences. When those sequences can be shown to be Cauchy in an appropriate function space, completeness provides the existence of the limit without having to guess the explicit form of the endpoint function.
7.2 Building solutions via approximation schemes
In differential equations and variational problems, one often approximates a desired solution by a sequence of simpler functions. Establishing that the approximations form a Cauchy sequence in a chosen norm then yields a candidate limit that can be shown to satisfy the intended equation or minimizing property.
7.3 Stability under limits for linear operators
If a linear operator is bounded (continuous) between normed spaces, it preserves Cauchy sequences in the sense that applying the operator to a Cauchy sequence yields a Cauchy sequence in the target norm. In complete spaces, this allows one to pass to the limit and obtain continuity of the operator with respect to convergence.
7.4 Continuity and boundedness of operators on function spaces
| Operator estimates frequently rely on inequalities that control \(\|Tf\|\) by \(\|f\|\). Such controls imply that Cauchy sequences in the domain map to Cauchy sequences in the range. This is a standard mechanism for transferring convergence behavior through functional analytic frameworks. |
|---|
8 Technical results often used with Cauchy sequences
Several recurring tools help manage Cauchy sequences, especially in proofs where direct convergence is hard to show.
8.1 Subsequence extraction from Cauchy sequences
Even though Cauchy sequences need not converge in incomplete spaces, their Cauchy property ensures control over tails. One can often extract subsequences with additional properties, for example making the increments summable or ensuring rapid convergence in the metric, which is useful for diagonal arguments.
8.2 Completeness arguments with diagonal subsequences
When dealing with families indexed by multiple parameters, diagonal subsequences help select a single subsequence that satisfies multiple constraints simultaneously. This technique is common when proving existence of limits via successive approximation steps.
8.3 Interchange of limits with integrals (where justified)
Passing limits inside integrals requires conditions. In \(L^p\) settings, one uses convergence theorems and dominated convergence-type arguments when available, or relies on norm convergence to justify the operation. The key is that integrals are continuous with respect to the chosen mode of convergence under appropriate bounds.
8.4 Lower semicontinuity of norms in relevant contexts
Lower semicontinuity statements often guarantee that certain functionals do not jump upward in the limit. While not every lower semicontinuity claim follows from norm convergence alone, many results in variational analysis and Sobolev spaces use it to pass to limits in energy-like quantities when Cauchy/weak convergence is present.
9 Variations and extensions
Cauchy sequences extend beyond classical functions into generalized objects and modified metrics that emphasize different aspects of behavior.
9.1 Cauchy sequences of distributions (generalized function spaces)
Distributions are continuous linear functionals on test-function spaces. One can define topologies on spaces of distributions so that generalized Cauchy sequences converge in that topology. The precise structure depends on the chosen framework, such as weak-* or other locally convex topologies.
9.2 Vector-valued function spaces
If functions take values in a vector space (e.g., \(\mathbb{R}^k\) or a Hilbert space), the norms typically combine the size in the target space with integration over the domain. Cauchy-ness is then defined using these combined norms, and the resulting convergence theory parallels the scalar case.
9.3 Weighted L^p spaces and modified norms
| Weighted norms replace \( | f | ^p\) with weights such as \(w(x) | f(x) | ^p\). This alters which regions of the domain dominate the metric, thereby changing whether a sequence is Cauchy. Such spaces are common when behavior at infinity or near singularities must be controlled. |
|---|
9.4 Cauchy sequences under different metrics inducing the same topology
Sometimes different metrics generate the same notion of convergence on a space. In those cases, the Cauchy property may be equivalent or at least comparable, though subtleties can arise if the metrics are not uniformly equivalent. Analysts often choose convenient metrics that align with the problem’s estimates.
10 Summary of key takeaways and common pitfalls
Cauchy sequences provide a robust internal criterion for convergence, but their meaning hinges on the surrounding structure of the function space.
10.1 “Cauchy implies convergence” depends on completeness
A sequence being Cauchy does not guarantee convergence in an incomplete space. The implication holds precisely when the space is complete under the selected norm or metric.
10.2 Norm choice changes the notion of Cauchy
Different norms can lead to different Cauchy sequences for the same underlying pointwise behavior. For example, uniform control corresponds to the sup norm, while \(L^p\) norms average discrepancies and can permit behaviors excluded by the sup norm.
10.3 Equivalence-class issues in L^p spaces
In \(L^p\) spaces, convergence concerns functions up to changes on measure-zero sets. It is therefore possible for representatives to differ pointwise while the \(L^p\) distance remains small.
10.4 Differences between pointwise and norm Cauchy behavior
| Pointwise Cauchy behavior does not automatically imply Cauchy-ness in a strong norm such as \(L^p\) or \(\|\cdot\|_\infty\). Conversely, norm Cauchy-ness can enforce strong global constraints that may imply pointwise properties only after additional subsequence or regularity arguments. |
|---|