Noether's theorem — symmetry ↔ conservation law: Visual walkthrough

2.1.9 · D2 · Physics › Analytical Mechanics › Noether's theorem — symmetry ↔ conservation law

Step 0 — Woh picture jise hum explain karne ki koshish kar rahe hain

Kisi bhi algebra se pehle, ek picture ke zariye agree karte hain ki ye words ka matlab kya hai.

KYA. Ek particle move karta hai. Woh ek curve trace karta hai. Hum poochte hain: kya koi aisa number hai jo us curve ke saath frozen rehta hai?

KYU. Physics "conserved quantities" se bhari hai — energy, momentum, spin. Hum jaanna chahte hain ki ye aate kahan se hain. Claim yeh hai: ye symmetry se aate hain, jo bas yeh hai ki "setup ko thoda sa nudge karo aur kuch important nahi badalta."

PICTURE. Ek bead ek smooth horizontal rail par slide kar rahi hai. Poori rail ko sideways slide karo — physics ekdum same hai (bead ko kahan rail hai pata nahi, sirf uski shape pata hai). Yeh "kahan se koi farak nahi padta" ek symmetry hai. Iske andar chhupa frozen number bead ka momentum hai.

Figure — Noether's theorem — symmetry ↔ conservation law

Step 1 se pehle hume ek aur idea chahiye: woh machine jo decide karti hai ki kaun sa path physical hai.

Step 1 — Paths ke liye scorecard: the action

KYA. Nature har path try nahi karti. Woh woh ek path chunti hai jo ek khaas total score, action, ko as flat-as-possible banata hai (minimum ya saddle). Hum yeh Lagrangian Mechanics se import kar rahe hain.

YEH tool kyun aur " $F = ma$ " kyun nahi? Kyunki poore paths ka ek scorecard exactly woh setting hai jahan "path ko nudge karo aur score nahi badalta" ka matlab banta hai. Symmetry nudges ke baare mein ek statement hai; action woh cheez hai jis par nudges act karte hain. Sirf forces se hume "coordinate nudge karo" ke baare mein baat karne ki koi natural jagah nahi milti.

PICTURE. Har candidate path ko ek number $S$ milta hai. $S$ ko "kaun sa path" ke against plot karo. Sachcha path valley ke bottom par baitha hai — jahan zameen flat hai. Flatness poora game hai.

"Valley ka bottom" condition, symbols mein likhi, Euler-Lagrange Equation hai — woh Step 2 hai.

Step 2 — Physical-path law (Euler–Lagrange)

KYA. "Score sachche path par flat hai" ek equation mein badal jaata hai jise path ko obey karna padta hai.

KYU. Derivation mein hume is equation ki zaroorat padegi ek bekar term ko ek tidy time-derivative mein convert karne ke liye. Isse ek tool ki tarah socho jo kehta hai "real path par force = rate of change of momentum," lekin action ki language mein likha.

PICTURE. Sachcha path (solid) aur ek thoda wiggled neighbour (dashed) jo dono endpoints par agree karta hai. Flatness ka matlab hai ki wiggle ka score sachche path ke score ke barabar hai first order tak — dono scores mein wiggle mein koi linear difference nahi.

Step 3 — Symmetry precisely kya hai

KYA. Har coordinate ko ek tiny, shape-controlled amount se nudge karo: $q_{i} \to q_{i}^{'} = q_{i} + ϵ δ q_{i} .$

YEH form kyun? Ek continuous symmetry ke liye ek continuous dial $ϵ$ chahiye. Hum nudge ko "kitna" ( $ϵ$ , ek chhota number) times "kis direction/shape mein" ( $δ q_{i}$ , ek pattern jo $q$ 's par depend kar sakta hai) mein split karte hain. Yahi separation baad mein $ϵ$ mein derivative lene deta hai.

PICTURE. Path ke har point par ek arrow $δ q_{i}$ "yahan yeh nudge mujhe push karega" ki taraf point karta hai. Translation mein sab parallel arrows hain; rotation mein arrows ek center ke around curl karte hain. Same path, do alag nudge-fields.

Step 4 — Nudge ke under score badalna (chain rule)

KYA. Compute karo ki nudge apply karne par $L$ kitna badalta hai, $δ L$ .

KYU chain rule aur kuch fancy nahi? $L$ depend karta hai $q_{i}$ aur $\overset{q}{˙}_{i}$ par. Agar dono thoda wiggle karein, toh change ka ek-ek honest first-order hisaab hai "har input ke liye $L$ kitna sensitive hai, times us input ne kitna move kiya." Yahi chain rule hai.

PICTURE. $L$ $(q, \overset{q}{˙})$ plane ke upar ek pahadi hai. $(δ q, δ \overset{q}{˙})$ se move karna height ko (slope-in- $q$ )×(step-in- $q$ ) + (slope-in- $\overset{q}{˙}$ )×(step-in- $\overset{q}{˙}$ ) se badalta hai.

Step 5 — Position-slope ko momentum-rate se trade karo (EL use karo)

KYA. Awkward pehle term $\frac{\partial L}{\partial q _{i}}$ ko Step 2 ke law se replace karo.

KYU. Hum ek single time-derivative $\frac{d}{d t} (kuch)$ dhundh rahe hain, kyunki wahi ek conserved quantity produce karta hai. Term $\partial L / \partial q_{i}$ hume rok raha hai; Euler–Lagrange hume isse $\frac{d}{d t} p_{i}$ se swap karne deta hai, jo pattern mein fit baith jaata hai.

PICTURE. Ek chhota swap-card: "force" slot $\partial L / \partial q_{i}$ ko "rate of momentum" card $\frac{d}{d t} p_{i}$ se exchange kiya gaya — sirf ek physical path par valid.

Step 6 — Ek time-derivative mein collapse → the conserved charge

KYA. Product rule recognize karo aur do terms ko ek single $\frac{d}{d t} (\cdot)$ mein fold karo.

KYU. Identity $\frac{d}{d t} (uv) = \overset{u}{˙} v + u \overset{v}{˙}$ hamare do-term mess ko ek derivative mein badal deta hai. Agar poori cheez $δ L = 0$ (ek symmetry!) ke barabar hai, toh "derivative $= 0$ " ka matlab hai "andar ki cheez constant hai." Woh constant prize hai.

PICTURE. Do puzzle pieces $\overset{u}{˙} v$ aur $u \overset{v}{˙}$ ek single block $\frac{d}{d t} (uv)$ mein snap hote hain. Uske neeche ek flat line "constant" likhi hai.

Step 7 — Boundary case: jab $L$ badlta hai

KYA. Kabhi kabhi score perfectly invariant nahi hota: balki $δ L = ϵ \frac{d F}{d t}$ hota hai kisi function $F$ ke liye. Action phir bhi invariant hai, kyunki $\int \frac{d F}{d t} d t$ sirf endpoint values padhta hai, aur nudge endpoints par vanish hone ke liye defined hai.

KYU yeh cheating nahi hai. Ek total time-derivative integrate hokar "value at $t_{2}$ minus value at $t_{1}$ " deta hai — pure boundary. Kyunki nudge karte waqt endpoints fixed rakhte hain, yeh action ke variation mein kuch contribute nahi karta.

PICTURE. Ek "extra score" ka ramp jo sirf endpoints par depend karta hai; path ka interior untouched hai, isliye woh affect nahi kar sakta ki kaun sa path chuna jaata hai.

Worked check — free particle, rotation ↔ angular momentum

Ek-picture summary

Upar sab kuch ek arrow diagram hai: ek nudge jo score ko flat chhod deta hai, ek time-derivative ko zero hone par majboor karta hai, aur us derivative ke andar ki cheez conserved charge hai.

Recall Feynman retelling — poora walkthrough simple words mein

Socho ki har tarike se ek marble roll ho sakti hai, har tarike ko ek score do. Nature hamesha woh tarika chunti hai jiska score sabse flat ho. Ab gently poora setup shove karo — slide karo, spin karo, thodi der ruko. Agar score nahi hila, toh tumne ek "doesn't-matter" dhundh liya. Phir maine likha ki jab main shove karta hun toh score kitna badalta hai (woh chain rule hai), ek clumsy piece swap kiya us rule se jo real paths obey karte hain (Euler–Lagrange), aur achanak do bacha hua pieces "ek akeli cheez ki rate of change" mein snap ho gaye. Agar shove karne se score nahi badla, toh woh rate of change zero hai — isliye woh ek akeli cheez kabhi nahi hilti. Sliding momentum freeze karta hai; spinning angular momentum freeze karta hai; wait karna energy freeze karta hai (jab tak rules khud clock ke saath tick nahi karte). Har "doesn't-matter" ek aisa number chhupata hai jo change karne se inkar karta hai.

Recall

Ek cyclic coordinate woh hai jis par $L$ ... ::: depend nahi karta; tab uska conjugate momentum conserved hota hai. Noether charge formula (no boundary term) hai ::: $Q = \sum_{i} p_{i} δ q_{i} = const$ . Boundary term $δ L = d F / d t$ ke saath, charge ban jaata hai ::: $Q = \sum_{i} p_{i} δ q_{i} - F$ . Time-translation shape $δ q_{i}$ barabar hai ::: $\overset{q}{˙}_{i}$ , jo energy function $h = \sum_{i} p_{i} \overset{q}{˙}_{i} - L$ deta hai. Discrete symmetries Noether charge kyun nahi deti? ::: Koi continuous parameter $ϵ$ nahi hota jiske respect se differentiate kar sakein.