Foundations — Isoperimetric problems — constraints (Lagrange multipliers in variational sense)

Before you can read the parent note Isoperimetric problems, every symbol on that page has to mean something to you. This page builds each one from nothing, in the order they depend on each other.

1. A function $y(x)$ — a curve you can bend

Start with the most basic object. When we write $y(x)$ we mean: ==give me a horizontal position $x$, and I hand you back a height $y$==. Sweep $x$ from left to right and the heights trace out a curve.

Why we need it: in this whole chapter, the unknown we are solving for is not a number — it is an entire curve. That is the single biggest jump from ordinary calculus. In "find $x$ that minimises $f(x)$" the answer is a point; here the answer is the shape of a rope or a fence.

2. The interval $[a,b]$ and boundary conditions

The symbols $a$ and $b$ are the two ground positions where the curve starts and ends — the two posts a chain hangs between, or the two pins holding a fence.

Picture: the two red dots in Figure s01 are frozen. Any curve we consider must pass through both.

Why we need it: without pinned ends the "best" curve would run off to infinity. Fixing the ends is what makes the problem have a finite answer.

3. The derivative $y^{\prime }$ — the steepness at each point

$y^{\prime }$ (read "y-prime") is the slope of the curve at a point: how many units up per one unit right, measured on the tiny straight piece hugging the curve there.

Why we need it: every integrand in the parent note (both $f$ and $g$) depends on $y^{\prime }$, because length and energy both care about steepness.

4. Arc length and the symbol $\sqrt{1+y^{\prime 2}}$

Here is a symbol that looks scary but is pure Pythagoras. Take a tiny step of width $dx$ along the ground. The curve rises by $dy=y^{\prime }dx$ over that step. The actual slanted piece of curve is the hypotenuse of a right triangle with legs $dx$ (across) and $dy$ (up).

Read it off the triangle: hypotenuse${}^2=$ base${}^2+$ height${}^2=d{x}^2+(y^{\prime }dx{)}^2$. Factor out $d{x}^2$ and take the square root — the $dx$ pops out, leaving $\sqrt{1+y^{\prime 2}}$.

Why we need it: the constraint in every isoperimetric example — "the string has fixed length" — is written $\int \sqrt{1+y^{\prime 2}}dx=\ell$. This is that symbol summed up.

5. The integral ${\int }_a^b\dots dx$ — adding up all the tiny pieces

The tall S sign $\int$ means sum up a continuous pile of tiny contributions. We chop the interval $[a,b]$ into countless tiny widths $dx$, evaluate the thing inside on each, and add them all.

Picture: lay infinitely many thin slabs side by side under the curve and glue them — the integral is the glued total.

Why we need it: length, area, energy are all whole-curve quantities, not point quantities. Only an integral can turn "the shape" into "one number to optimise or constrain."

6. A functional $J[y]$ — a machine that eats a whole curve

Now the key new object. An ordinary function eats a number. A functional eats an entire curve and returns a single number.

The little function $f(x,y,y^{\prime })$ inside is the integrand: a recipe telling you, at each $x$, how much this curve contributes — using where you are ($x$), how high ($y$), and how steep ($y^{\prime }$).

7. The variation $\delta y$, the bump $\eta (x)$, and the knob $ϵ$

To find the best curve we must nudge a candidate curve and see if the number $J$ goes up or down. The nudge is written $\delta y$ ("delta y"), a small deformation of the curve that keeps the endpoints pinned.

Turn the knob $ϵ$ and the whole family of red dashed curves in Figure s04 swings between the extremes, but always through the two fixed dots.

Why we need it: "the best curve" means any small legal nudge makes $J$ no better. That is the condition $\delta J=0$, and to write it we need $\eta$ and $ϵ$.

8. The Euler–Lagrange expression $f_y-\frac{d}{dx}{f}_{y^{\prime }}$

When you nudge $J$ and demand it not change for every bump $\eta$, the algebra (integration by parts + the Fundamental Lemma) squeezes out one master equation.

Read it as a balance law: at every point the direct pull of height ($f_y$) must exactly cancel the running change of the steepness-sensitivity ($\frac{d}{dx}{f}_{y^{\prime }}$). See Euler–Lagrange Equation for the full derivation this page prepares you for.

Why we need it: this is the tool that turns "$J$ is stationary" into a concrete differential equation you can solve for the curve.

9. The multiplier $\lambda$ — the price of the budget

Finally, the star of the parent note. $\lambda$ ("lambda") is a single fixed number — the fine per unit of budget. Whenever the problem has a constraint $K[y]=\ell$, we don't obey it directly. Instead we build the augmented recipe $F=f-\lambda g$ and optimise that freely.

Why exactly one number (not a function $\lambda (x)$)? Because an integral constraint is a single scalar equation ($\int gdx=\ell$), so it can only "charge" one fine. (Pointwise constraints would need a whole function $\lambda (x)$ — the last mistake in the parent note.) This mirrors ordinary finite-dimensional Lagrange multipliers exactly.

How it all feeds the topic

Trace it: curves and slopes build length; length summed is an integral; an integral of a whole curve is a functional; nudging a functional gives Euler–Lagrange; adding a priced budget gives the isoperimetric rule. Everything downstream — Beltrami Identity, the Catenary Curve, the Isoperimetric Inequality, the Brachistochrone Problem, and the Calculus of Variations — Fundamental Lemma — plugs in at exactly these nodes.

Equipment checklist

Self-test: cover the right side and see if you can state each from memory.