Vectors in ℝⁿ — operations, geometric interpretation

---

WHAT is a vector in ℝⁿ?

WHY two pictures?

• As a point: the tuple is an address in space.
• As an arrow: it is the displacement from the origin $\mathbf{0}$ to that point. The arrow picture is what lets us add and scale; the point picture is what lets us locate.

---

The two fundamental operations

Everything else is built from exactly two operations.

---

Building length from first principles (the norm)

Derivation (in $\mathbb{R}^2$ then generalize): In $\mathbb{R}^2$, $\mathbf v=(v_1,v_2)$ is the hypotenuse of a right triangle with legs $v_1$ and $v_2$. By Pythagoras: $\|\mathbf v\|^2 = v_1^2 + v_2^2$.

In $\mathbb{R}^3$, drop to the $xy$-plane: that diagonal has squared length $v_1^2+v_2^2$, and it is perpendicular to the $z$-leg $v_3$. Apply Pythagoras again: $\|\mathbf v\|^2 = (v_1^2+v_2^2) + v_3^2.$ Repeating this "one new perpendicular axis at a time" $n$ times gives:

A vector with $\|\mathbf v\|=1$ is a unit vector. To make any nonzero $\mathbf v$ into one, normalize: $\hat{\mathbf v} = \dfrac{\mathbf v}{\|\mathbf v\|}$ (same direction, length 1).

---

The dot product — where geometry meets algebra

WHY does this single number encode the angle between arrows? Derive it from the Law of Cosines. Consider the triangle formed by $\mathbf u$, $\mathbf v$, and the side $\mathbf u-\mathbf v$: $\|\mathbf u - \mathbf v\|^2 = \|\mathbf u\|^2 + \|\mathbf v\|^2 - 2\|\mathbf u\|\|\mathbf v\|\cos\theta.$ Now expand the left side using $\|\mathbf w\|^2 = \mathbf w\cdot\mathbf w$ and bilinearity:

= \|\mathbf u\|^2 - 2\,\mathbf u\cdot\mathbf v + \|\mathbf v\|^2.$$ Set the two expressions equal; the $\|\mathbf u\|^2$ and $\|\mathbf v\|^2$ cancel: $$-2\,\mathbf u\cdot\mathbf v = -2\|\mathbf u\|\|\mathbf v\|\cos\theta.$$ > [!formula] Geometric meaning of the dot product > $$\boxed{\;\mathbf u\cdot\mathbf v = \|\mathbf u\|\,\|\mathbf v\|\cos\theta\;}$$ > Hence $\cos\theta = \dfrac{\mathbf u\cdot\mathbf v}{\|\mathbf u\|\|\mathbf v\|}$, and crucially: > $$\mathbf u \perp \mathbf v \iff \mathbf u\cdot\mathbf v = 0.$$ ![[4.5.01-Vectors-in-Rn-—-operations,-geometric-interpretation.png]] --- ## Worked examples > [!example] 1 — Addition & scaling > Let $\mathbf u=(1,2)$, $\mathbf v=(3,-1)$. Find $2\mathbf u - \mathbf v$. > - $2\mathbf u = (2,4)$ — *Why?* scalar mult is component-wise. > - $2\mathbf u - \mathbf v = (2-3,\;4-(-1)) = (-1,5)$ — *Why?* subtraction = add $(-1)\mathbf v$, component-wise. > [!example] 2 — Length & unit vector > $\mathbf w=(3,4)$. Then $\|\mathbf w\|=\sqrt{3^2+4^2}=\sqrt{25}=5$. > - *Why?* Pythagoras: it's the hypotenuse of a 3–4 triangle. > - Unit vector: $\hat{\mathbf w}=\frac15(3,4)=(0.6,0.8)$. *Why divide by 5?* dividing by the length > rescales to length exactly 1 without rotating. > [!example] 3 — Angle between vectors > $\mathbf a=(1,0,1)$, $\mathbf b=(0,1,1)$. > - $\mathbf a\cdot\mathbf b = 1\cdot0+0\cdot1+1\cdot1 = 1$ — *Why?* sum of products of matching components. > - $\|\mathbf a\|=\|\mathbf b\|=\sqrt2$ — *Why?* $\sqrt{1+0+1}$. > - $\cos\theta = \dfrac{1}{\sqrt2\cdot\sqrt2}=\tfrac12 \Rightarrow \theta=60^\circ$. > *Why?* plug into the boxed formula; $\arccos\frac12=60^\circ$. > [!example] 4 — Forecast-then-Verify > **Forecast:** are $(2,-3)$ and $(3,2)$ perpendicular? *Guess first!* > **Verify:** $2\cdot3 + (-3)\cdot2 = 6-6 = 0$. Yes — dot product zero $\Rightarrow$ perpendicular. ✓ --- ## Common mistakes (Steel-manned) > [!mistake] "Vectors multiply component-wise like they add." > **Why it feels right:** addition is component-wise, so by symmetry multiplication should be too, > giving $(u_1v_1,\dots)$. **The fix:** that "product" isn't useful geometrically. The meaningful > products are the **dot product** (gives a *scalar*, encodes angle) and the cross product (only in > $\mathbb{R}^3$). Component-wise product carries no geometric meaning here. > [!mistake] "The dot product gives another vector." > **Why it feels right:** every other operation so far ($+$, scaling) returns a vector. > **The fix:** $\mathbf u\cdot\mathbf v$ is a **single number (scalar)** — it *measures* alignment, > it doesn't point anywhere. If you got a vector, you computed the wrong thing. > [!mistake] "Normalizing means subtracting the mean / making length 0." > **Why it feels right:** "normalize" sounds like statistics. **The fix:** here normalize = > *divide by the norm* so length becomes $1$. Direction is preserved. --- > [!recall]- Feynman: explain to a 12-year-old > Imagine a treasure map. A vector is an instruction like "go 3 steps right and 4 steps up." > To **add** two instructions, you just follow one then the other — and you can do the right-steps > and up-steps separately. The **length** is how far you actually walked in a straight line from > start to finish (3-right + 4-up = 5 steps straight, the classic triangle!). The **dot product** > is a cleverness meter: if two instructions point the same way, it's big; if they cross at a right > angle, it's exactly zero. That zero is how a computer "knows" two directions are perpendicular. > [!mnemonic] Remember the operations > **"Add Across, Scale All, Length is the Long way (Pythagoras), Dot is Direction."** > And for the dot: **"Dot = product, Cross-axis = 0"** → matching components multiply, perpendicular ⇒ 0. --- ## Active-recall flashcards #flashcards/maths What is a vector in ℝⁿ? ::: An ordered $n$-tuple $(v_1,\dots,v_n)$ of real numbers; geometrically a point or an arrow from the origin. How do you add two vectors in ℝⁿ? ::: Component-wise: $(u_i+v_i)$ for each $i$; geometrically tip-to-tail / parallelogram rule. Define the Euclidean norm. ::: $\|\mathbf v\|=\sqrt{\sum v_i^2}$, derived from repeated Pythagoras. How do you normalize a vector? ::: Divide by its length: $\hat{\mathbf v}=\mathbf v/\|\mathbf v\|$, giving length 1, same direction. Define the dot product. ::: $\mathbf u\cdot\mathbf v=\sum u_i v_i$ — a scalar. Geometric meaning of dot product? ::: $\mathbf u\cdot\mathbf v=\|\mathbf u\|\|\mathbf v\|\cos\theta$. When are two vectors perpendicular? ::: When $\mathbf u\cdot\mathbf v=0$. What is $\mathbf v\cdot\mathbf v$? ::: $\|\mathbf v\|^2$, the squared length. Derive the angle formula's key cancellation. ::: Expand $\|\mathbf u-\mathbf v\|^2$ two ways (Law of Cosines vs dot expansion); the $\|\mathbf u\|^2,\|\mathbf v\|^2$ cancel leaving $\mathbf u\cdot\mathbf v=\|\mathbf u\|\|\mathbf v\|\cos\theta$. What does scaling by $c<0$ do geometrically? ::: Stretches length by $|c|$ and flips direction. --- ## Connections - [[Linear Combinations and Span]] — addition + scaling are the engine of spans. - [[Dot Product and Orthogonality]] — deeper consequences (projections, Cauchy–Schwarz). - [[Norms and Distance in Rn]] — generalizing the length idea. - [[Vector Spaces — Axioms]] — ℝⁿ is the prototype that the axioms abstract. - [[Projections and Orthogonal Decomposition]] — built directly from $\mathbf u\cdot\mathbf v$. - [[Cross Product (R3 only)]] — contrast: a *vector*-valued product unique to 3D. ## 🖼️ Concept Map ```mermaid flowchart TD V[Vector in Rn as n-tuple] P[Point / address] A[Arrow / displacement] ADD[Vector addition] SCALE[Scalar multiplication] CW[Component-wise rule] PYTH[Pythagorean theorem] NORM[Euclidean norm / length] UNIT[Unit vector / normalize] DOT[Dot product] V -->|interpreted as| P V -->|interpreted as| A A -->|enables| ADD A -->|enables| SCALE ADD -->|defined| CW SCALE -->|defined| CW ADD -->|geometrically| A PYTH -->|applied per axis| NORM NORM -->|scale to length 1| UNIT DOT -->|self product gives| NORM DOT -->|sums products of| CW ``` ## 🔊 Hinglish (regional understanding) > [!intuition]- Hinglish mein samjho > Dekho, ek **vector** matlab numbers ki ek ordered list — jaise $(3,4)$. Lekin iska asli kamaal yeh > hai ki yeh do cheezein ek saath represent karta hai: ek **point** (kahan ho space mein) aur ek > **arrow** (kaise move karna hai origin se). Isi dual picture pe poori linear algebra khadi hai. > > Do hi basic operations hain. **Addition** aur **scalar multiplication**, dono **component-wise** > hote hain — har axis alag hai, isliye east mein chalo toh north pe asar nahi padta. Addition ka > geometric matlab parallelogram ya tip-to-tail rule hai. **Length (norm)** hum Pythagoras se derive > karte hain: $\|\mathbf v\|=\sqrt{v_1^2+\dots+v_n^2}$. Length 1 wale vector ko **unit vector** kehte > hain, aur kisi vector ko uski length se divide karke normalize kar dete hain. > > Sabse important: **dot product** $\mathbf u\cdot\mathbf v=\sum u_i v_i$. Yeh ek single number deta > hai (vector nahi!) jo angle batata hai: $\mathbf u\cdot\mathbf v=\|\mathbf u\|\|\mathbf v\|\cos\theta$. > Agar dot product **zero** hai, toh vectors **perpendicular** hain — yahi rule computer use karta hai > direction check karne ke liye. Ek common galti: log sochte hain dot product bhi vector deta hai ya > multiply bhi component-wise hota hai — nahi! Dot ka result hamesha scalar hota hai. Yeh basics itne > strong rakho ki span, projection, orthogonality sab automatic samajh aayega. ![[audio/4.5.01-Vectors-in-Rn-—-operations,-geometric-interpretation.mp3]]