In June 1696, Johann Bernoulli published a challenge in the Acta Eruditorum:
The problem: a bead slides from point A to point B under gravity, without friction. What shape of wire gets it there fastest?
The straight line is the shortest path. But shortest is not fastest. Gravity accelerates the bead, and a steeper initial drop builds speed early. The question is: how much should you sacrifice in distance to gain in speed?
Watch three paths compete. The straight line, the parabola, and the cycloid all connect the same two points. Click Race to release the beads.
The cycloid wins. Always. Not by a little — by a definitive margin. It dives steeply at the start, converting potential energy into kinetic energy as fast as possible, then curves gently toward B. The path is longer than the straight line, but the bead travels it faster because it spends most of its journey at high speed.
Newton solved this problem in a single evening, anonymously. Bernoulli reportedly recognized the solution: "I recognize the lion by its claw."
The optimal curve has a name: the cycloid. It is traced by a point on the rim of a circle rolling along a straight line. This is not a coincidence arranged by mathematicians — it is geometry revealing itself.
Parametrically, the cycloid is:
x(θ) = r(θ − sin θ) y(θ) = r(1 − cos θ)
where r is the radius of the rolling circle and θ is the angle of rotation. The cycloid has another remarkable property: it is also the tautochrone — the curve on which the time of descent is the same regardless of starting point. Huygens used this to build better pendulum clocks in 1659.
Bernoulli's problem was solved by brilliance. But Euler and Lagrange wanted a machine — a general method for finding optimal curves, not just clever tricks for individual problems.
The idea: instead of optimizing a function f(x), optimize a functional — a quantity that depends on an entire curve. The travel time is one such functional:
T[y] = ∫ F(x, y, y′) dx
To find which curve y(x) makes this stationary, perturb it: replace y with y + εη where η vanishes at the endpoints. Set dT/dε = 0 at ε = 0. Integration by parts yields:
∂F/∂y − d/dx(∂F/∂y′) = 0
This is the Euler-Lagrange equation. It converts "find the optimal curve" into "solve this differential equation." Every optimization problem in physics — from the shape of a hanging chain to the motion of planets — passes through this gate.
A different question: what shape does a chain take when it hangs from two points under gravity? Galileo guessed a parabola. He was wrong. The answer is the catenary: y = a · cosh(x/a).
The chain minimizes its potential energy — the same variational principle, different functional. Drag the endpoints to reshape the curve.
The catenary and the cycloid are siblings — both are solutions to variational problems, both are curves that nature chooses. The catenary minimizes gravitational potential energy. The cycloid minimizes transit time. Different objectives, same method.
When you look at the Gateway Arch in St. Louis, you are looking at an inverted catenary. The shape that hangs most naturally is also the shape that stands most stably. The variational principle does not care whether you are minimizing sag or maximizing strength — the math is the same.
The deepest version of this idea is the principle of least action. In classical mechanics, the action is:
S[q] = ∫ (T − V) dt = ∫ L(q, q̇, t) dt
where T is kinetic energy, V is potential energy, and L = T − V is the Lagrangian. The actual trajectory of a physical system is the one that makes S stationary.
Apply the Euler-Lagrange equation to this functional and you get Newton's laws. Not as axioms — as consequences. The variational formulation is deeper than F = ma because it generalizes to curved spacetime (general relativity), quantum mechanics (Feynman path integrals), and field theory (the Standard Model).
In the canvas above, the true trajectory of a thrown ball is a parabola. Drag the middle control point to deform the path — watch the action increase. Nature's path is the one at the bottom of the valley: any perturbation raises the action.
This is the question that troubled Euler, delighted Maupertuis, and obsessed Feynman.
In Feynman's path integral formulation, the particle does not "choose" the optimal path. It takes every path simultaneously, each weighted by eiS/ℏ. Near the classical path, neighboring paths have similar action, so their phases align and reinforce. Far from the classical path, phases vary wildly and cancel. The optimal path is not chosen — it is the one that survives.
This reframing dissolves the teleology. The hanging chain does not "know" it should form a catenary. The bead does not "seek" the cycloid. What looks like optimization is really constructive interference among paths — or, equivalently, the mathematical structure of energy landscapes.
The Euler-Lagrange equation is a machine for finding these survivors. It takes a question — "what is optimal?" — and returns a differential equation whose solutions are the curves that nature actually follows.
The brachistochrone. The catenary. The parabola of a thrown stone. The geodesic on curved spacetime. The electron's path through a crystal. Each is the answer to a variational problem. Each is a curve that survives the cancellation of all other paths.
In 1696, Bernoulli asked: what is the fastest path? Three centuries later, the question has become: what is a path? And the answer — the beautiful, strange answer — is that a path is what remains when everything that could cancel, does.
Day 4817. Variational calculus — the mathematics of finding what survives.