1 Definition
Lower semicontinuity describes a function that does not jump downward abruptly. At a point, the function value is bounded above by nearby values in a precise limiting sense. This idea is used when exact continuity is too strong, but one still wants stable behavior under approximation.
1.1 Informal intuition
A lower semicontinuous function may rise or stay level near a point, but it does not suddenly drop below the values suggested by nearby points. In graph form, one can think of the function as being approachable from below by neighboring values. This makes it useful in settings where minima and limiting processes matter.
1.2 Pointwise definition
Let \(f\colon X \to \mathbb{R}\cup\{+\infty\}\) be a function on a topological space \(X\). The function is lower semicontinuous at a point \(x\in X\) if for every real number \(a<f(x)\), there exists a neighborhood \(U\) of \(x\) such that \(f(y)>a\) for all \(y\in U\). Equivalently, the value at \(x\) is at most the limit inferior of the values nearby.
1.3 Sequential characterization
In a metric space, lower semicontinuity can be expressed using sequences. A function \(f\) is lower semicontinuous at \(x\) if for every sequence \(x_n\to x\), one has \[ f(x)\le \liminf_{n\to\infty} f(x_n). \] This form is often convenient in analysis, since it directly compares the value at the limit point with the limiting behavior of the sequence.
1.4 Topological characterization
On a topological space, lower semicontinuity can be described without sequences. The essential idea is that the sets where the function exceeds a given threshold behave like open sets. This viewpoint connects lower semicontinuity to the structure of level sets and preimages of intervals.
2 Equivalent formulations
Lower semicontinuity has several equivalent descriptions, each useful in a different branch of mathematics. These formulations are especially important because they translate an analytic condition into a topological or geometric one.
2.1 In terms of open sets
A function \(f\colon X\to \mathbb{R}\cup\{+\infty\}\) is lower semicontinuous if, for every real number \(a\), the set \[ \{x\in X : f(x)>a\} \] is open. These sets are called superlevel sets. This criterion is often the easiest to verify in practice.
2.2 In terms of closed sets
An equivalent statement is that for every real number \(a\), the set \[ \{x\in X : f(x)\le a\} \] is closed. These are sublevel sets. This formulation is widely used in optimization, where closed sublevel sets help control minimizing sequences.
2.3 In terms of epigraphs
The epigraph of a function packages all points lying on or above its graph. Lower semicontinuity is equivalent to a closedness property of this set.
2.3.1 Epigraph definition
The epigraph of \(f\) is the set \[ \operatorname{epi}(f)=\{(x,t)\in X\times \mathbb{R}: t\ge f(x)\}. \] A function is lower semicontinuous precisely when its epigraph is closed in the product topology, under standard hypotheses on the domain.
2.3.2 Geometric interpretation
Geometrically, a closed epigraph means that the region above the graph has no missing boundary points. This captures the idea that the graph may be approximated from above without creating sudden downward gaps. The epigraph viewpoint is central in convex analysis and variational problems.
3 Basic properties
Lower semicontinuity is stable under several common operations. These properties explain why the concept appears so often in analysis and optimization.
3.1 Relation to continuity
Every continuous function is both lower semicontinuous and upper semicontinuous. The converse is not true: a function can fail to be continuous while still avoiding downward jumps. Thus lower semicontinuity is a weaker, one-sided form of continuity.
3.2 Relation to upper semicontinuity
Upper semicontinuity is the mirror image of lower semicontinuity. A function is upper semicontinuous if it does not jump upward abruptly. Together, the two properties imply continuity when they hold at the same point.
3.3 Stability under addition and scalar multiplication
The sum of two lower semicontinuous functions is lower semicontinuous, provided the expression is well defined in the extended real-valued setting. Multiplication by a nonnegative scalar also preserves lower semicontinuity. These closure properties make the class of lower semicontinuous functions flexible under algebraic manipulation.
3.4 Stability under pointwise limits
Pointwise limits of lower semicontinuous functions need not be lower semicontinuous in general. However, under suitable monotonicity assumptions, such as increasing pointwise limits, lower semicontinuity is preserved. This is useful when constructing functions as limits of simpler approximations.
4 Examples and nonexamples
Examples help distinguish lower semicontinuity from ordinary continuity and from upper semicontinuity. Nonexamples show how a function can fail the condition by dropping too sharply.
4.1 Continuous functions
Any continuous function is lower semicontinuous. For instance, polynomials, trigonometric functions, and exponentials all satisfy the property. Their values vary smoothly, so no downward discontinuity can occur.
4.2 Step functions
A step function can be lower semicontinuous if its jumps occur upward. If it drops downward at a jump point, it fails to be lower semicontinuous there. This makes step functions a simple testing ground for the definition.
4.3 Indicator functions
Indicator functions of closed sets are lower semicontinuous when they take the value \(1\) on the set and \(0\) outside it. By contrast, indicator functions of open sets are generally upper semicontinuous rather than lower semicontinuous. These examples illustrate the connection between semicontinuity and set topology.
4.4 Functions with downward jumps
A function that takes a high value at a point and smaller nearby values is not lower semicontinuous at that point. A typical example is a function with an isolated spike. Since the spike is above the nearby behavior, it violates the one-sided limit condition.
5 Lower semicontinuity in optimization
Lower semicontinuity is fundamental in optimization because it supports existence results for minimization problems. It also interacts naturally with compactness and approximation.
5.1 Existence of minimizers
If a lower semicontinuous function is defined on a compact set, then it attains its minimum. This is a direct consequence of the fact that sublevel sets are closed and compactness prevents minimizing sequences from escaping. Such results are standard in the direct method of the calculus of variations.
5.2 Lower semicontinuous envelopes
Given a function that is not lower semicontinuous, one may form its lower semicontinuous envelope, the greatest lower semicontinuous function lying below it. This construction produces a regularized version that is better suited to minimization and limit arguments. It is often interpreted as a “completed” or “relaxed” form of the original function.
5.3 Coercivity and compactness
Coercivity, or growth to infinity at large distance, is frequently combined with lower semicontinuity to ensure that minimizers exist. The coercive condition helps make sublevel sets bounded, while lower semicontinuity makes them closed. Together, these properties can yield compactness in finite-dimensional settings.
6 Lower semicontinuity in measure and functional spaces
In spaces of functions and measures, lower semicontinuity often appears in weaker topologies. These settings are essential in modern analysis because convergence may not be pointwise or uniform.
6.1 Weak lower semicontinuity
A functional on a Banach space is weakly lower semicontinuous if it remains lower semicontinuous with respect to weak convergence. This property is especially important because minimizing sequences in infinite-dimensional spaces often converge only weakly. Convex functions frequently enjoy weak lower semicontinuity under appropriate assumptions.
6.2 Lower semicontinuity of integral functionals
Integral functionals of the form \[ u \mapsto \int_\Omega F(x,u(x),\nabla u(x))\,dx \] often arise in variational analysis. Lower semicontinuity of such functionals is a key criterion for the existence of minimizers. Conditions such as convexity in the gradient variable commonly play a central role.
6.3 Applications in variational methods
Variational methods study critical points and minimizers of functionals by considering limiting sequences. Lower semicontinuity ensures that the limit of an approximating minimizing sequence does not have larger energy than expected. This makes it possible to pass from approximate solutions to actual solutions.
7 Variants and generalizations
The notion of lower semicontinuity extends beyond ordinary real-valued functions. It adapts to broader contexts in topology, order theory, and functional analysis.
7.1 Extended real-valued functions
Allowing values in \(\mathbb{R}\cup\{+\infty\}\) is standard in convex analysis and optimization. The value \(+\infty\) can represent infeasibility or an excluded region. In this setting, lower semicontinuity remains well behaved and is particularly useful for encoding constraints.
7.2 Lower semicontinuity on metric spaces
On metric spaces, the sequence-based definition is often the most practical. Many theorems become easier to state and prove using \(\liminf\) along convergent sequences. This is one reason the concept is ubiquitous in analysis on metric spaces.
7.3 Lower semicontinuity in ordered structures
Lower semicontinuity also appears in ordered and lattice-based settings, where the notion of “approaching from below” has an intrinsic order-theoretic meaning. In such contexts, the definition may be adapted using order convergence or analogous closure properties. These generalizations broaden the concept beyond classical topology.
</INTERNAL_LINK_CANDIDATES> Topological space (the ambient space on which semicontinuity is defined) Limit inferior (the limiting lower bound used in the sequential criterion) Upper semicontinuity (the one-sided counterpart that forbids upward jumps) Continuity (the condition equivalent to having both semicontinuities) Superlevel set (the set where a function exceeds a fixed threshold) Sublevel set (the set where a function is at most a fixed threshold) Epigraph (the region on or above a function’s graph) Optimization (the study of minimizing or maximizing functions) Minimizer (a point where a function attains its minimum) Lower semicontinuous envelope (the greatest lower semicontinuous minorant) Coercivity (growth condition helping ensure compactness of sublevel sets) Compactness (a finiteness property used in existence proofs) Weak lower semicontinuity (lower semicontinuity with respect to weak convergence) Banach space (a complete normed vector space) Variational method (a technique using minimization of functionals) Integral functional (a functional defined by integrating an integrand) Convex analysis (the study of convex functions and related structures) Indicator function (a function taking values 0 and 1 on a set) Metric space (a space with a distance function) Order theory (the study of ordered structures and monotone relations) </INTERNAL_LINK_CANDIDATES>