Gauss's method for Lambert's problem: Visual walkthrough

3.2.29 · D2 · Physics › Orbital Mechanics & Astrodynamics › Gauss's method for Lambert's problem

Hum Lambert's problem solve kar rahe hain: do positions aur ek travel time diya gaya hai, orbit nikalni hai. Geometry ko ek line at a time badhte hue dekho.

Step 1 — Do dots aur ek chord (pure geometry, abhi tak koi orbit nahi)

KYA. Central body (Sun, Earth, ya jo bhi hum orbit kar rahe hain) ko ek point par rakho jise focus $F$ kehte hain. $F$ se pehli position ki taraf ek arrow khiincho, use $r_{1}$ bolo; doosri position ki taraf ek aur arrow, $r_{2}$ . Un do arrows ki tips ko jodne wali seedhi line chord $c$ hai.

KYUN. Orbit ke baare mein kuch bhi jaane bina, hum in teeno lengths ko diye gaye data se pahle se jaante hain. Gauss jo bhi karta hai woh is triangle se anchored hai, isliye hum ise pehle draw karte hain.

PICTURE. Figure dekho. $F$ se do arrows nikalte hain; unki tips ke beech dashed line $c$ hai. Arrows ke beech ka angle — $r_{1}$ se $r_{2}$ tak — transfer angle $θ$ hai.

Chord aati hai law of cosines se — wo rule jo triangle ki teesri side nikaalti hai jab do sides aur unke beech ka angle pata ho:

$c^{2} = side 1^{2} r_{1}^{2} + side 2^{2} r_{2}^{2} - correction for the angle 2 r_{1} r_{2} cos θ$

Aur $cos θ$ khud given vectors se dot product ke zariye free milta hai: $cos θ = \frac{r _{1} \cdot r _{2}}{r _{1} r _{2}} .$

Step 2 — Flat triangle ka area

KYA. Triangle $F$ –tip $_{1}$ –tip $_{2}$ ko bhar do aur uska area mapo.

KYUN. Chalti planet chord ke saath nahi chalti — woh curve karti hai. Lekin flat triangle sabse simple cheez hai jo hum maap sakte hain, isliye hum ise yardstick ki tarah use karte hain aur baad mein poochenge "true curved region kitni zyada badi hai?"

PICTURE. Shaded triangle. Iska area base $\times$ height $/2$ hai; $r_{1}$ ko base lekar, height hai $r_{2} sin θ$ .

Step 3 — Asli path ek pie slice hai, triangle nahi

KYA. Ab planet ko apni ellipse par tip $_{1}$ se tip $_{2}$ tak actually chalate ho. Jis region ko arrow $F \to planet$ sweep karta hai woh ek curved elliptic sector hai — ek pie slice jiska crust bahar ki taraf bula hua hai. Is region ko naam do: $A_{sector}$ .

KYUN. Yahaan orbit finally enter hoti hai. Sector flat triangle se bada hai kyunki asli path baahir (ya andar) ki taraf curve karti hai. Jitna yeh differ karta hai woh specific orbit ki pehchaan hai.

PICTURE. Mint pie-slice flat triangle ke upar baithi hai. Curved arc aur straight chord ke beech extra sliver "bulge" hai.

Kepler's second law, equation ki shakal mein, kehta hai ki sweep rate constant hai aur $\frac{1}{2} h$ ke barabar hai, jahan $h = μ p$ (constant) angular momentum hai. "Area = rate × time" ko rearrange karo: $Δ t = \frac{2 A _{sector}}{h}, h = μ p .$

Step 4 — Ratio $y$ define karo: pie triangle se kitna mota hai?

KYA. Curved sector area ko flat triangle area se divide karo. Use $y$ bolo.

KYUN. Raw area units aur orbit size par depend karti hai — messy. Do areas ka ratio ek clean, unit-free number hai jo sirf shape difference capture karta hai. Yahi wo ek quantity hai jise Gauss chase karta hai.

PICTURE. Dono regions side by side; $y$ literally "green area ÷ blue area" hai.

Recall

$y$ ko 1 ke paas first guess kyun lena achha hai? Doubt mein guess karo ::: Zyaadatar transfers mein arc thoda hi bulge karta hai, toh sector ≈ triangle, toh $y \approx 1$ se iteration answer ke paas shuru hoti hai.

Step 5 — Eccentric anomaly, aur half-angle kahan se aata hai

Do equations se pehle, humein woh variable chahiye jo maape ki ship ne ellipse ka kitna hissa cover kiya — aur yeh dekhna hai ki half-angle $θ /2$ baar baar kyun aata hai.

KYA. Ellipse ko uski same width ka ek helper circle le kar draw karo. Orbit par kisi bhi point ke liye, use seedha circle tak upar drop karo; ellipse ke centre se (focus se nahi!) us circle point tak ka angle eccentric anomaly $E$ hai. Do tip values ko $E_{1}$ aur $E_{2}$ bolo.

KYUN. Ellipse mein time-of-flight sirf $E$ ke terms mein clean hoti hai (yeh Kepler's equation hai). Aur chord jo do tips ko jodti hai, $E$ mein nikali jaaye toh hamesha $sin (\frac{E _{2} - E _{1}}{2})$ ke proportional aati hai — ek half-difference — kyunki radius $a$ ke circle ki chord jo angle $Δ E$ subtend kare uski length $2 a sin (Δ E /2)$ hoti hai. Woh ek "chord = $2 a sin (half-angle)$ " identity hi har $θ /2$ aur har half- $E$ ka source hai jo tumhe dikhega.

PICTURE. Ellipse, uska helper circle, centre $O$ , focus $F$ , do eccentric-anomaly angles, aur chord $2 a sin (Δ E /2)$ ki tarah drawn.

Step 6 — Do equations, aur woh kahan se aate hain

Humein ab ek hi orbit ke baare mein do independent facts mile hain, donon mein variable $x$ hai. Ek pure geometry hai; doosra time carry karta hai. Asli orbit donon ko ek saath satisfy karta hai.

Do equations kyun? Ek unknown pair $(x, y)$ , do independent physical facts (geometry, time) → ek determined system. Unka simultaneous solution orbit hai.

Step 7 — Constants $ℓ$ aur $m$ , motivated

KYA. Do numbers, givens se ek baar compute karo.

KYUN. Woh sirf Step 6 ki derivation ke shape-only aur time-only leftovers hain, jinhe naam diya gaya hai taaki iteration clean rahe.

PICTURE. Label diagram: $ℓ$ triangle ko tag karta hai, $m$ stopwatch ko.

Formula Do constants

m=\frac{\mu\,\Delta t^2}{\big(2\sqrt{r_1 r_2}\,\cos(\theta/2)\big)^3}$$ - $\ell$ mein $\dfrac{r_1+r_2}{\sqrt{r_1 r_2}\cos(\theta/2)}$ — **shape** combination: yeh sum-of-radii hai jo Step 5 ke chord-scale $\sqrt{r_1r_2}\cos(\theta/2)$ se measure kiya gaya. $-\tfrac12$ ise centre karta hai taaki $\ell=0$ reference triangle mark kare. - $\cos(\theta/2)$ — Step 5 ki chord-matching identity ka half-angle, koi coincidence nahi. - $m$ mein $\Delta t^2$ — time **squared** enter karta hai. Yeh Kepler's period law ka fingerprint hai: swept area $\propto \Delta t$, lekin length-scale jo ise normalise kare $\propto \Delta t^{2/3}$, toh ek baar denominator cube karo toh packed constant $\Delta t^2$ carry karta hai. - Denominator $\big(2\sqrt{r_1 r_2}\cos(\theta/2)\big)^3$ — chord-scale **cubed**, exactly Step 6 ka $\sqrt{p}$-through-Kepler bookkeeping.

Step 8 — Iteration ek picture ki tarah: guess, slide, correct

KYA. $y = 1$ se shuru karo. Equation A use karo wo $x$ nikaalne ke liye jo yeh guess imply karta hai. Equation B use karo ek better $y$ compute karne ke liye. Repeat.

KYUN. Do curves ka crossing closed form mein solve nahi ho sakta, toh hum unke beech bounce karte hain; kyunki $y$ true value ke paas shuru hota hai, bounces tezi se sidhte hain (ek contraction map).

PICTURE. Do curves ke crossing point par converge hoti staircase.

Step-4 update Equation B hai jisme $y^{2}$ ko $m / (ℓ + x)$ se replace kiya gaya hai: $y - 1 = \frac{m}{y ^{2}} (x - \frac{1}{2} + X) = (ℓ + x) (x - \frac{1}{2} + X) .$

Step 9 — Degenerate aur non-elliptic cases jo tumhe ZAROOR dekhne chahiye

KYA / KYUN. Har solver ko corners mein survive karna chahiye. Yahan woh hain jo diktein karte hain.

PICTURE. Panels: near- $18 0^{\circ}$ blow-up, short/long-way branches, straight-line $θ = 0$ limit, aur parabolic/hyperbolic ( $e \geq 1$ ) regime.

Common mistake Parabolic (

$e = 1$ ) aur hyperbolic ( $e > 1$ ) transfers Sahi lagta hai: humne sab kuch ek ellipse ke liye derive kiya. Trap: agar required $Δ t$ bahut chhota hai (ek fast, energetic transfer), toh connecting orbit bound nahi hai — yeh ek parabola ( $e = 1$ ) ya hyperbola ( $e > 1$ ) hai, jiske liye eccentric anomaly $E$ aur ellipse-only series $X (x)$ ab applicable nahi hain (Step 5 ka "helper circle" ek helper hyperbola ban jaata hai). Lambert's problem ka phir bhi ek solution hai, lekin Gauss ki specific elliptic series uski taraf converge nahi karti. Fix: Universal variable formulation ellipse, parabola, aur hyperbola ko ek set of equations (Stumpff functions) se handle karta hai, yahi wajah hai ki Battin's method aur Izzo Lambert solver modern default hain. Physically: given $r_{1}, r_{2}$ , chhote flight times ⇒ hyperbola, ek special time ⇒ parabola, lambe times ⇒ ellipse.

Ek-picture summary

Sab kuch ek canvas par: flat triangle (yardstick), mota sector $A_{sector}$ (asli path), unka ratio $y$ , do curves jinka crossing $y$ satisfy karta hai, aur staircase jo us par land karta hai. $y$ se hum $p$ padhte hain, phir Lagrange f and g functions $f, g, \overset{g}{˙}$ , phir velocities jo do dots aur ek stopwatch ko ek poori orbit mein badal dete hain — ek Porkchop plot ka raw material.

Recall Feynman retelling — poori walkthrough seedhe alfaazon mein

Mujhe Sun se ek spaceship ki do positions tak arrows khiinche, aur unki tips ko ek seedhi line se joda — woh flat triangle mera ruler hai. Lekin ship seedhi nahi chalti; woh curve karti hai, ek mota pie slice sweep karti hai, aur mein us curved region ko sector kehta hoon. Kepler ne mujhe bataya ki area actually ek clock hai: ship ne kitna sweep kiya yeh batata hai kitna waqt laga — aur unhone jo stopwatch time $Δ t$ diya woh batata hai wo pie slice kitni badi honi chahiye. Toh mein pie slice ko flat triangle se compare karta hoon aur us ratio ko $y$ kehta hoon. $y$ pin karne ke liye mujhe ek aur number chahiye: ship ellipse par kitni door gayi, eccentric-anomaly angle $E$ se measure kiya — aur kyunki circle ki chord hamesha "twice-radius-times-sine-of-half-the-angle" hoti hai, half-angles $θ /2$ har jagah aate hain. Isse mujhe ek hi orbit ke baare mein do statements milti hain: ek pure geometry, ek clock carry karta hai. Mujhe $y$ nahi pata abhi, toh maan leta hoon yeh lagbhag 1 hai (barely curved), dekhta hoon us arc ka kya matlab hai, time check karta hoon, aur $y$ nudge karta hoon jab tak geometry aur stopwatch agree na kar lein. Woh agreement hi orbit hai. Edges dhyaan mein rakho: Sun ke opposite sides mere numbers explode karte hain, which way around bhool jaana galat path deta hai, perfectly-lined-up points koi triangle nahi chodtein maapne ke liye, aur bahut fast trip ellipse bhi nahi hoti — woh parabola ya hyperbola hoti hai jiske liye universal-variable method chahiye.