Visual walkthrough — Transformation between frames — direction cosine matrices

3.4.2 · D2 · Physics › Rocket Flight Mechanics › Transformation between frames — direction cosine matrices

Ye page parent DCM note ka picture-companion hai. Agar koi word unfamiliar lage, hum usse use karne se pehle build karenge.

---

Step 1 — Ek arrow, abhi koi numbers nahi

KYA. Space mein ek single arrow $\vec{v}$ rakho. "Ground frame mein velocity" nahi — bas ek arrow. Uski ek length hai aur ek direction, aur abhi ke liye bas itna hi hai. Koi numbers nahi.

KYUN. Ye DCM ka poora secret hai: arrow ek real physical cheez hai (ek thrust, ek velocity, ek position). Use parwah nahi ki tum kaunse axes use karte ho. Jo numbers hum abhi attach karne wale hain wo hamaari description hai, arrow khud nahi. Agar hum ye bhool gaye, toh baad ke har superscript ka matlab gibberish ho jayega.

PICTURE. Lal arrow akela float kar raha hai. Jaanbujhkar koi axes nahi draw ki gayi hain — deliberately. Ek arrow jiske paas koi ruler nahi, uske koi components nahi hote. Components wo cheez hai jo hum baad mein add karte hain.

---

Step 2 — Frame $A$ bichhaao aur shadows padho

KYA. Ek set of perpendicular axes daal do. Hum unhe ${\hat{a}}_1$ aur ${\hat{a}}_2$ naam dete hain (2-D mein; teesra, ${\hat{a}}_3$, page se bahar point karta hai). Chhoti hat $\hat{}$ matlab "length exactly 1" — ek unit ruler. Ab $\vec{v}$ ki tip se har axis par ek perpendicular daal do. Un shadows ki lengths numbers $v_1^A$ aur $v_2^A$ hain.

KYUN. Hum perpendicular shadows (right-angle drop) use karte hain kyunki exactly yahi matlab hai "arrow ka kitna hissa is axis ke along hai." Kisi aur tarah ka projection ya toh double-count karega ya arrow ka kuch part miss kar dega. Superscript $A$ kehta hai "frame $A$ ke against measure kiya gaya."

PICTURE. Lal arrow ke ab do dashed shadow lines hain jo axes se right angles par milti hain. $v_1^A$ horizontal ke along, $v_2^A$ vertical ke along padho. Milke wo arrow rebuild karte hain: $\vec{v}=v_1^A{\hat{a}}_1+v_2^A{\hat{a}}_2$.

---

Step 3 — Dot product ek shadow-measuring machine hai

KYA. Ek shadow length algebra se extract karne ke liye, hum dot product use karte hain. Kisi bhi unit axis $\hat{u}$ ke liye, number $\vec{v}\cdot \hat{u}$ hai $\hat{u}$ par $\vec{v}$ ki shadow ki length.

YE TOOL KYUN. Dot product kyun aur kuch aur nahi? Kyunki $\vec{v}\cdot \hat{u}=|\vec{v}||\hat{u}|\cos \alpha =|\vec{v}|\cos \alpha$ (kyunki $|\hat{u}|=1$), aur $|\vec{v}|\cos \alpha$ exactly $\hat{u}$ par $\vec{v}$ ki perpendicular shadow ki length hai. Dot product precisely woh question answer karta hai jo humein matter karta hai — "is arrow ka kitna hissa us taraf point karta hai?" — ek hi operation mein. Isliye yahi, aur kuch nahi, DCM mein appear karta hai.

PICTURE. Lal arrow aur axis $\hat{u}$ ke beech angle $\alpha$; shadow length arrow length times $\cos \alpha$ hai. Jaise $\alpha \to 0$ shadow full length ke equal hoti hai; $\alpha =90^{\circ }$ par shadow gayab ho jaati hai ($\cos 90^{\circ }=0$).

---

Step 4 — Ek doosra frame $B$ laao, $A$ se twist kiya hua

KYA. Ab ek doosra set of axes ${\hat{b}}_1,{\hat{b}}_2$ overlay karo, jo $A$-axes se kisi angle se rotated hai. Wohi lal arrow ab $B$-axes par alag shadows daalta hai: $v_1^B$ aur $v_2^B$.

KYUN. DCM ka poora point yehi hai. Do observers, same arrow, alag-alag turned rulers. Yahan se hamara goal: $A$-numbers diye gaye, $B$-numbers compute karo arrow ko kabhi touch kiye bina.

PICTURE. Do axis crosses ek origin share karte hain: kale $A$-axes, aur $B$-axes jo angle $\theta$ se turned hain (faintly drawn). Lal arrow unchanged hai. $B$-axes par uski shadows clearly $A$-axes par uski shadows se alag lengths hain.

---

Step 5 — Arrow ko EK $B$-axis par project karo

KYA. Ek single $B$-axis lo, jaise ${\hat{b}}_1$. Uski shadow-number hai $v_1^B=\vec{v}\cdot {\hat{b}}_1$ (Step 3 ki machine). Ab Step 2 ka arrow ka $A$-form substitute karo: $v_1^B=\vec{v}\cdot {\hat{b}}_1=(v_1^A{\hat{a}}_1+v_2^A{\hat{a}}_2+v_3^A{\hat{a}}_3)\cdot {\hat{b}}_1.$

KYUN. Hum $\vec{v}$ ko $A$-language mein rewrite karte hain kyunki $A$-numbers wo data hain jo hamare paas hain. Iss tarah likhne ke baad, dot product sum mein distribute ho jaata hai, aur har piece ek axis-dotted-with-axis ban jaata hai.

PICTURE. Lal ${\hat{b}}_1$ axis; teen faint kale arrows ${\hat{a}}_1,{\hat{a}}_2,{\hat{a}}_3$; har ek ${\hat{b}}_1$ par apni chhoti shadow daalta hai. Poori shadow $v_1^B$ un teen axis-shadows ka weighted sum hai, weights $A$-components hain.

---

Step 6 — Har axis-to-axis dot ek direction cosine hai

KYA. Ek term dekho, ${\hat{b}}_1\cdot {\hat{a}}_j$. Dono unit vectors hain, toh ${\hat{b}}_1\cdot {\hat{a}}_j=\cos {\theta }_{1j}$ — $B$-axis 1 aur $A$-axis $j$ ke beech angle ka cosine. Woh single number $C_{1j}^{BA}$ hai.

KYUN. Ye cosines hi reason hain ki object ko "direction cosine matrix" naam diya gaya hai. Ye sirf is baat par depend karte hain ki dono frames kaise twist hain, arrow par nahi — toh hum inhe ek baar compute karke har arrow ke liye reuse kar sakte hain.

PICTURE. ${\hat{b}}_1$ aur ${\hat{a}}_1$ ke beech angle ${\theta }_{11}$ ek wedge ki tarah draw kiya gaya; uska cosine matrix ke row 1, column 1 par stored number hai. Chhota angle → cosine nearly 1 (axes nearly aligned); $90^{\circ }$ → cosine 0 (axes perpendicular, zero contribution).

---

Step 7 — Teeno rows stack karo: matrix appear hoti hai

KYA. ${\hat{b}}_2$ aur ${\hat{b}}_3$ ke liye bhi Step 5 karo. Har ek teen cosines ki ek row deta hai. Teeno rows stack karna hi matrix hai, aur poori cheez ${\mathbf{v}}^B={\mathbf{C}}^{BA}{\mathbf{v}}^A$ mein collapse ho jaati hai.

KYUN. Matrix bas ek bookkeeping grid hai: row $i$ teen cosines hold karta hai jo $v_i^B$ build karte hain. Matrix–vector multiplication defined hai "row-dot-column" ke roop mein, jo exactly Step 5 ka sum hai. Matrix koi naya idea nahi — wahi teen projections hain ek tidy costume mein.

PICTURE. $3\times 3$ grid. Row 1 red mein highlighted = "${\hat{b}}_1$ ko $A$-coordinates mein likha gaya." Neeche, multiplication ${\mathbf{C}}^{BA}{\mathbf{v}}^A$ dikhaya gaya jisme har row poora column ${\mathbf{v}}^A$ "kha" raha hai.

---

Step 8 — Degenerate case: frame $B$ = frame $A$

KYA. Agar humne kabhi twist hi nahi kiya — $B$ exactly $A$ ke upar baitha hai? Toh ${\hat{b}}_i={\hat{a}}_i$, har same-index angle $0^{\circ }$ hai (cosine $1$), har cross angle $90^{\circ }$ hai (cosine $0$). Matrix identity $\mathbf{I}$ ban jaati hai.

KYUN. Ye sanity anchor hai. "No rotation" ko har arrow ke numbers unchanged chhodni chahiye: ${\mathbf{v}}^B=\mathbf{I}{\mathbf{v}}^A={\mathbf{v}}^A$. Agar tumhara DCM formula kabhi zero twist ke liye $\mathbf{I}$ ke alawa kuch aur de, toh tumhare mein bug hai. Ye ye bhi dikhata hai ki elementary ${\mathbf{C}}_3(\theta )$ ka $\theta \to 0$ limit smoothly $\mathbf{I}$ ban jaata hai: $\cos 0=1$, $\sin 0=0$.

PICTURE. Dono axis crosses perfectly ek doosre ke upar; wedge angles sab $0^{\circ }$ ya $90^{\circ }$ hain; grid $1$s ki diagonal se fill ho jaata hai.

---

Ek-picture summary

Upar ki poori cheez ek single frame mein: lal arrow (unchanged), do twisted axis crosses, ek highlighted angle ${\theta }_{ij}$, aur arrow $\cos {\theta }_{ij}\Rightarrow C_{ij}^{BA}$ jo us cosine ko uski grid slot mein daal raha hai. Pipeline left-to-right padhi jaati hai: arrow → ${\hat{b}}_i$ par project karo → har axis-pair angle → uska cosine → matrix entry → $A$-numbers multiply karo → $B$-numbers.

Recall Feynman: poori walkthrough plain words mein retell karo

Main aur mera ek dost dono ek hi patang ki taraf point karte hain. Patang real hai — jab hum directions par argue karte hain tab wo hilti nahi. Main usse apne do rulers (apne axes) ke against measure karta hoon aur do numbers pata hai; wo numbers bas woh shadows hain jo patang ka arrow mere rulers par daalta hai, aur har shadow ki length arrow ki length times us angle ka cosine hai jo wo us ruler se banata hai. Meri dost ne apne rulers thoda ghuma liye hain, toh uske shadows alag nikle. Meri numbers ko uski numbers mein baadlne ke liye, main patang ko bilkul touch nahi karta: mujhe bas itna jaanna hai ki uske rulers mere rulers ke comparison mein kitne twisted hain — aur "kitne twisted" measure hota hai uske har ruler aur mere har ruler ke beech angle ke cosine se. Nau aise cosines (uske teen rulers times mere teen rulers) ek chhoti grid banate hain, direction cosine matrix. Mere teen numbers us grid se multiply karo aur uske teen numbers nikal aate hain. Agar usne apne rulers kabhi twist nahi kiye, toh har matching angle $0^{\circ }$ hai (cosine $1$) aur har crossing angle $90^{\circ }$ hai (cosine $0$), toh grid bas diagonal pe $1$s hai — identity — aur uske numbers mere barabar hain, exactly jaisa hona chahiye. Aur kyunki rulers ghoomane se arrow kabhi stretch nahi hota, patang ki distance dono hamaari descriptions mein same aati hai — ek free check ki humne sahi kiya.

Recall Quick self-test

• Dot product $\vec{v}\cdot {\hat{b}}_i$ kaunsa physical operation perform karta hai? ::: ye axis ${\hat{b}}_i$ par arrow ki perpendicular shadow ki length measure karta hai
• Har matrix entry ek cosine kyun hai? ::: har entry ek unit-axis dotted with ek unit-axis hai, aur ${\hat{b}}_i\cdot {\hat{a}}_j=\cos {\theta }_{ij}$
• ${\mathbf{C}}^{BA}$ ki row $i$ kya represent karti hai? ::: $B$-axis ${\hat{b}}_i$ ko $A$-coordinates mein likha gaya
• Jab $B=A$ ho toh matrix kya equal honi chahiye? ::: identity $\mathbf{I}$
• Arrow ki length dono frames mein same kyun rehti hai? ::: rulers ghoomane se arrow kabhi stretch nahi hota; orthogonality length preserve karta hai

Prerequisite frames aur tools Reference frames in rocketry — inertial, body, wind aur Orthogonal matrices and rotation groups SO(3) mein hain; multi-rotation build-up Euler angles — yaw pitch roll mein hai.