Calculus Of Variations

The calculus of variations asks a deceptively simple question: among all the ways of doing something, which way is best?

12 frameworks, from 1730 to today · Analysis · Mathematics

still a working view ended · each leads to its entry below

Overview

The calculus of variations asks a deceptively simple question: among all the ways of doing something, which way is best? Among all curves joining two points, which is shortest1? Among all shapes enclosing a given area, which has the least perimeter2? Among all paths a falling bead might take, which is fastest3? The objects being compared are not numbers but whole functions—entire curves, surfaces, or histories of a system—and the task is to find the one that makes some quantity (length, time, energy, cost) as small or as large as possible. This shift from optimizing numbers to optimizing functions is what makes the subject a branch of analysis, and it is why the field sits at the crossroads of geometry, mechanics, partial differential equations, and optimization. Its history is a long negotiation between formal rules that produce candidate answers and the harder question of whether those candidates are genuinely best.

The Brachistochrone and the Birth of a Method

The subject crystallized in 1696, when Johann Bernoulli publicly challenged the mathematicians of Europe: find the curve down which a bead sliding without friction under gravity travels from one point to another in the least time. The straight line is shortest but not fastest; a steeper initial drop builds speed sooner. Bernoulli already knew the answer—a cycloid, the curve traced by a point on a rolling circle—and he posed the problem partly to demonstrate the power of a new kind of reasoning.

The decisive contribution came from Euler, who in the 1730s and 1740s turned such ad hoc tricks into a general algorithm1. The idea is to perturb a candidate curve by a small "variation"—a slight wiggle that vanishes at the endpoints—and demand that the quantity being optimized not change to first order. Carrying this out yields a differential equation the optimal curve must satisfy. For a quantity written as an integral of a function $L$ depending on position, slope, and the independent variable, the condition is the Euler–Lagrange equation, a second-order ordinary differential equation14. Lagrange, in the 1750s and 1760s, recast the whole procedure in purely analytic terms, freeing it from geometric pictures and giving the field the name by which it is still known. The method's power was that it converted a search over infinitely many curves into the solution of a single differential equation.

Least Action and the Mechanics of Lagrange and Hamilton

The variational method found its most spectacular application in physics. Maupertuis and Euler had proposed that nature acts so as to minimize a quantity they called "action," and Lagrange made this precise: the motions of a mechanical system are exactly those that make a certain integral stationary5. Hamilton, in the 1830s, reformulated mechanics around two functions—the Lagrangian, in terms of positions and velocities, and the Hamiltonian, in terms of positions and momenta—linked by a transformation that bears his name6. The resulting Hamilton's equations are a symmetric system of first-order equations whose geometric structure (the symplectic form) has become a subject in its own right.

The payoff was not merely elegance. The variational formulation made it possible to derive conservation laws from symmetries—Noether's theorem, proved in 1918, states that every continuous symmetry of the action corresponds to a conserved quantity, so that invariance under time translation gives conservation of energy and invariance under rotation gives conservation of angular momentum7. This result became foundational for twentieth-century physics. It also revealed a limitation: a stationary point need not be a minimum. The principle of least action is more accurately a principle of stationary action, and the calculus of variations had to learn to distinguish minima, maxima, and saddle points.

Constraints, Isoperimetry, and the Multiplier

Many of the oldest variational problems carry side conditions. The isoperimetric problem—among all closed curves of given length, which encloses the greatest area?—was known in antiquity (the answer is the circle) and became a template for constrained problems2. Dido's problem, the legendary founding of Carthage by cutting a hide into strips to enclose land against the sea, is the same question in a different guise.

Lagrange introduced the device that handles such constraints: the multiplier8. To optimize one quantity subject to another being fixed, one optimizes a combined expression with an adjustable coefficient, the Lagrange multiplier, chosen so that the constraint holds. The same idea extends to constraints expressed as integrals, as differential equations, or as inequalities. In the twentieth century this machinery was generalized into the theory of optimal control, where one steers a system governed by differential equations while minimizing a cost, and where the multiplier reappears as the "costate" or adjoint variable. The Pontryagin maximum principle, developed in the Soviet Union in the 1950s, gives necessary conditions for optimal control problems in a form that directly generalizes the Euler–Lagrange equation. Optimal control became the language of trajectory planning, economics, and engineering, and it remains one of the field's most active interfaces with applications.

The Direct Method: Minimizing Without Solving Equations

For all its success, the classical approach had a structural weakness. It produced a differential equation that any minimizer must satisfy, but it did not prove that a minimizer exists. Worse, the equation might have solutions that are not minima, and the true minimizer might be non-smooth—a curve with a corner, a surface with a crease—where the equation fails to make sense.

The response, developed from the late nineteenth century onward and brought to maturity in the twentieth, is the direct method. Instead of solving a differential equation, one works directly with the quantity being minimized. The strategy has three steps: take a sequence of admissible functions whose values approach the infimum; extract a convergent subsequence using compactness; and show that the limit is admissible and actually attains the infimum. Hilbert, in his 1900 lecture on the twentieth century's problems, singled out this variational program as a central challenge, and it became a driving force in analysis1910.

The direct method forced mathematicians to enlarge the class of admissible functions. If one insists on smooth curves, the infimum may not be attained; if one allows limits of smooth curves, one enters the world of weak solutions—functions with derivatives only in an averaged, integral sense. Sobolev spaces, introduced by Sergei Sobolev in the 1930s, provide exactly the right setting: spaces of functions whose derivatives exist as integrable objects, complete enough that minimizing sequences converge. The modern calculus of variations is largely conducted in these spaces, and the phrase "weak solution" signals that the equation is satisfied only after integration against test functions.

Lower Semicontinuity and the Convex Turn

The direct method has a subtle requirement. A minimizing sequence converges to a limit, but the quantity being minimized need not behave continuously under that convergence: the limit can have a larger value than the limit of the values. What is needed is lower semicontinuity—the property that the limit's value is no greater than the limiting value of the sequence. Without it, the infimum is not attained and the method collapses.

The decisive insight, developed in the mid-twentieth century by Tonelli, Morrey, and others, is that lower semicontinuity of integral functionals is governed by convexity of the integrand in the gradient variable1112. If the function $L$ is convex in its slope argument, the functional is lower semicontinuous and minimizers exist under mild growth conditions. This is the content of what is often called the Tonelli existence theory, and it explains why convex analysis became indispensable to the field1312. When convexity fails, minimizers may fail to exist in the ordinary sense, and one must either pass to a relaxed problem or accept generalized solutions. The theory of relaxation, and the study of microstructure in materials that oscillate at finer and finer scales, grew from exactly this difficulty.

Regularity: How Smooth Is a Minimizer?

Existence settled, a new question arose: how regular is the minimizer? A weak solution might in principle be very rough, and the classical Euler–Lagrange equation presupposes enough differentiability to be meaningful. The regularity theory for minimizers, developed through the twentieth century, shows that under suitable conditions—convexity, growth, and smoothness of the integrand—weak minimizers are in fact smooth in the interior. The methods are deep, drawing on comparison techniques, potential theory, and the theory of elliptic partial differential equations.

The theory also revealed its own limits. In the late 1920s and 1930s, examples showed that minimizers of seemingly reasonable variational problems can fail to be smooth, and the study of such singularities became a field of its own. The regularity problem for minimal surfaces—surfaces of least area spanning a given boundary—was a central testing ground, and its resolution for higher-dimensional surfaces required geometric measure theory, developed by Federer and others from the 1960s onward. The general question of when minimizers are smooth, and what singular sets they can have, remains an active research area.

Variational Geometry and Geometric Analysis

The calculus of variations has always been intertwined with geometry. Geodesics—shortest paths on a curved surface—are minimizers of length, and the study of their existence and multiplicity is a variational problem on a manifold114. Minimal surfaces, surfaces of prescribed mean curvature, and harmonic maps (maps between manifolds that minimize a Dirichlet-type energy) are all variational objects, and their study became the core of geometric analysis in the second half of the twentieth century.

The methods here are global and often topological. One does not merely solve an equation; one looks for critical points of a functional on an infinite-dimensional space and uses the topology of that space to count them. Morse theory, the mountain-pass theorem, and minimax methods provide existence results for critical points that are not minima—saddle points of the functional—and these techniques proved decisive for problems in geometry and nonlinear analysis. The solutions of the Yamabe problem, the existence of harmonic maps, and the structure of minimal surfaces all rest on such arguments.

The Present: A Field of Methods

Today the calculus of variations is less a single technique than a family of interlocking methods. The direct method and Sobolev-space formulations provide the existence framework; convex analysis and lower semicontinuity supply the structural conditions; regularity theory determines how smooth the resulting minimizers are; and geometric and topological methods locate critical points that are not minima. Optimal control extends the classical constrained problems into engineering and economics, while the variational formulation of mechanics remains the backbone of theoretical physics.

The open questions are as varied as the methods. When do minimizers exist without convexity, and what do the relaxed problems describe? How singular can minimizers be, and what is the structure of their singular sets? How do variational methods interact with the geometry of the underlying space, and what can topology force into existence? These questions descend directly from the brachistochrone challenge of 1696: the field still asks which way is best, and still finds that the answer depends on how generously one is willing to define "way."

References

  1. Calculus of variations (introduction, Example, History, Euler–Lagrange equation), Wikipedia
  2. Isoperimetric inequality, Wikipedia
  3. Brachistochrone curve, Wikipedia
  4. Euler–Lagrange equation, Wikipedia
  5. Action (physics) (History), Wikipedia
  6. Hamiltonian mechanics, Wikipedia
  7. Noether's theorem, Wikipedia
  8. Lagrange multiplier, Wikipedia
  9. Hilbert's twentieth problem, Wikipedia
  10. Hilbert's twenty-third problem, Wikipedia
  11. Quasiconvexity (calculus of variations), Wikipedia
  12. Tonelli's theorem (functional analysis), Wikipedia
  13. Leonida Tonelli, Wikipedia
  14. Geodesic, Wikipedia

The frameworks of Calculus Of Variations

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1730-1900
  2. 1750-Present

    Analytical mechanics

    How can motion be derived from a single quantity?

    Fly to its tower in Nousfarer

  3. 1760-Present
  4. 1760-Present

    Lagrange multiplier

    How do we optimize when a constraint must hold?

    Fly to its tower in Nousfarer

  5. 1900-Present
  6. 1920-Present
  7. 1920-Present

    Global variational and minimax methods

    How do we find global minima and saddle points?

    Fly to its tower in Nousfarer

  8. 1930-Present

    Regularity theory

    How smooth is an optimal function?

    Fly to its tower in Nousfarer

  9. 1930-Present

    Sobolev space

    What spaces contain weak solutions of variational problems?

    Fly to its tower in Nousfarer

  10. 1930-Present
  11. 1950-Present
  12. 1960-Present

    Geometric measure theory and minimal surfaces

    What are the optimal surfaces and how regular are they?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Classical Euler–Lagrange variational …Analytical mechanicsConstrained and isoperimetric variati…Lagrange multiplierDirect method in the calculus of vari…Convex analysis and lower semicontinu…Global variational and minimax methodsRegularity theorySobolev spaceVariational geometry and geometric an…Optimal control and the Pontryagin ma…Geometric measure theory and minimal …

Beyond the field

Where Calculus Of Variations's ideas came from in other fields, and what they led to there: the strongest 12 of 72 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Calculus Of Variations is a planet in Nousfarer

Its city has a tower for each of these 12 frameworks. Land, walk in, and climb one concept at a time.

Land on Calculus Of Variations

Related fields

Fields its ideas are linked to

More in Analysis

See where Calculus Of Variations sits on the Map of Human Knowledge