Foundations — Constraints using Lagrange multipliers

This is the vocabulary page for the parent topic. Read it once and every later equation reads like plain English.

0. The scene we are describing

Everything in this chapter is about one object being forced to stay somewhere: a bead threaded on a wire, a mass hanging from a rope, a ball rolling without slipping. The object wants to move freely (gravity pulls it), but something forbids certain motions. Our whole job is to describe both the freedom and the forbidding with symbols.

Look at the picture: the bead is stuck on the curve. It can slide along the wire (that arrow is allowed) but it can never move off the wire (that arrow is forbidden). Keep this split — along vs off — in your head. Every symbol below is a tool for talking about one of those two directions.

1. Coordinates $q_k$ — the numbers that pin down where the object is

The picture: to locate the bead in the plane you could give its horizontal position $x$ and vertical position $y$. That's two numbers, so $q_1=x$, $q_2=y$. Or, since it lives on a circle, you could give the radius $r$ and the angle $\theta$ — the polar coordinates the parent note loves. Either way, a coordinate is a dial you can turn to move the object.

Why the topic needs it: every equation in the parent note runs "for $k=1,\dots ,n$". That $n$ is how many dials you have. You cannot even state a constraint without first having coordinates to constrain. See Generalized Coordinates and Degrees of Freedom for the full story of choosing them.

2. The constraint function $f(q,t)=0$ — writing the wall as an equation

The picture: for a bead on a hoop of radius $R$, the rule is "your distance from the centre is always $R$." Written as a function that must vanish: $f=r-R=0.$ If $r=R$, then $f=0$ — allowed. If $r=1.2R$, then $f=0.2R\ne 0$ — forbidden. The set of allowed points ($f=0$) is a surface (here, the circle). Think of $f$ as a height map whose "sea level" ($f=0$) is exactly the wall.

Why the $t$? Sometimes the wall itself moves in time (a hoop being lifted, a rope being reeled in). Allowing $f$ to depend on time $t$ covers those cases. If the wall is fixed, $f$ just doesn't contain $t$ — but we write $t$ in to stay general.

3. The partial derivative $\frac{\partial f}{\partial q_k}$ — the slope of the wall in one direction

Here is the first genuinely new tool, so we slow down and earn it.

Why we need this tool and no other: we want to know which way is "off the wall." A single ordinary derivative can't tell us that, because "off" could mean nudging $x$, or nudging $y$, or some mix. The partial derivative lets us test each direction separately, and then assemble the answer.

The picture: stand on the height-map surface $f$. $\partial f/\partial q_k$ is the slope you feel if you step in the $q_k$ direction. For $f=r-R$: $\frac{\partial f}{\partial r}=1,\frac{\partial f}{\partial \theta }=0.$ Stepping in $r$ changes $f$ at rate $1$ (you climb straight out of sea level). Stepping in $\theta$ changes $f$ not at all — you slide along the wall. That zero is the mathematics saying "the $\theta$ direction is the allowed one."

4. The gradient $\nabla f$ — the arrow pointing straight off the wall

The picture: on the hoop, $\nabla f=(\partial f/\partial r,\partial f/\partial \theta )=(1,0)$ — an arrow pointing radially outward, dead perpendicular to the circle. That is precisely the direction a normal force from a wall can push. This is the secret hinge of the whole topic.

5. Virtual displacement $\delta q_k$ — an imaginary "allowed" nudge

The picture: grab the bead and shift it a hair along the wire — never off it. That whisper of a move is $\delta q$. It is "virtual" because you don't actually let time pass; you just ask "what if I nudged it this way — is that allowed?"

Why frozen time? Real motion mixes the object moving with the wall possibly moving. To isolate the object's freedom on the wall right now, we hold $t$ still. That is why the constraint's variation drops its $\partial f/\partial t$ term: $\delta f={\sum }_k\frac{\partial f}{\partial q_k}\delta q_k=0.$

6. Kinetic and potential energy $T$, $V$, and the Lagrangian $L=T-V$

The picture: imagine the bead partway up the hoop. It has some speed (that's $T$) and some height (that's $V$). $L$ blends them into one number that, when we differentiate it just right, spits out Newton's laws — but in any coordinates, even weird ones. That machinery is the Euler-Lagrange Equations.

Why "minus"? The combination $T-V$ (not $T+V$) is the one whose variation gives the correct equations of motion — a deep fact from the principle of least action. For this foundations page, just accept $L=T-V$ as the recipe; the parent note uses it as a black box.

7. The Euler–Lagrange operator $\frac{d}{dt}\frac{\partial L}{\partial {\dot{q}}_k}-\frac{\partial L}{\partial q_k}$

Why the topic needs it: for a free coordinate, setting this machine to zero is the law of motion. Under a constraint, it doesn't equal zero — it equals the leftover push, the constraint force $\lambda \partial f/\partial q_k$. So the whole parent equation reads: (rate of momentum) − (applied force) = (constraint force), which is just $ma=$ (real forces) rewritten.

8. The star of the show: the multiplier $\lambda$

Why it must exist — the tidy logic: from §5, the allowed nudge $\delta q$ is perpendicular to $\nabla f$. From d'Alembert, that same $\delta q$ is perpendicular to the Euler–Lagrange bracket. Two arrows perpendicular to the same flat plane must be parallel — so the bracket $=\lambda \times \nabla f$ for some number $\lambda$. That number is forced to exist; we just name it.

9. Counting: $n$, $m$, and "balanced equations"

Why it matters: the classic panic — "$\lambda$ seems undetermined!" — dissolves once you count. The constraint equation $f=0$ itself is the extra relation that pins $\lambda$ down. Bring your bookkeeping and you never lose a variable.

Prerequisite map

Equipment checklist

Say the answer out loud before revealing. If any stalls, reread that section.