Linear-Algebra-Done-Right · Unit 9 · Video 1 · Interactive Practice

The Dual Basis Reads Coefficients and T′ Reverses Arrows

IKey Formulas

FormulaNameWhat it says
V′=L(V,F)V' = \mathcal{L}(V, \mathbf{F}),   dim⁡V′=dim⁡V\;\dim V' = \dim VDual spaceThe linear functionals on VV; for finite-dimensional VV, dim⁡L(V,F)=(dim⁡V)(1)\dim \mathcal{L}(V, \mathbf{F}) = (\dim V)(1)
φj(vk)={1k=j0k≠j\varphi_j(v_k) = \begin{cases} 1 & k = j \\ 0 & k \neq j \end{cases}Dual basisOne functional per basis vector, fixed by the linear map lemma; φ1,…,φn\varphi_1, \dots, \varphi_n is a basis of V′V'
v=φ1(v) v1+⋯+φn(v) vnv = \varphi_1(v)\,v_1 + \cdots + \varphi_n(v)\,v_nCoefficient readerφj(v)\varphi_j(v) is the coefficient of vjv_j when vv is written in the basis
T′(φ)=φ∘TT'(\varphi) = \varphi \circ T,   (ST)′=T′S′\;(ST)' = T'S'Dual mapT ⁣:V→WT \colon V \to W gives T′ ⁣:W′→V′T' \colon W' \to V'; the order of a product reverses

Key Insight: φj\varphi_j depends on the whole basis, not only on vjv_j: with v1=(1,0)v_1 = (1, 0) and v2=(1,1)v_2 = (1, 1), the functional φ1(x,y)=x−y\varphi_1(x, y) = x - y is not the first coordinate, because it must also vanish on v2v_2.

IIVisualization 1 — The Dual Basis Reads Coefficients

For a basis v1,v2v_1, v_2 of R2\mathbf{R}^2, the dual basis returns the two coefficients of vv: v=φ1(v) v1+φ2(v) v2v = \varphi_1(v)\,v_1 + \varphi_2(v)\,v_2.

💡 Challenge: keep v1=(1,0)v_1 = (1, 0) and move only v2v_2 until φ1\varphi_1 becomes the first-coordinate functional (x,y)↦x(x, y) \mapsto x.

IIIVisualization 2 — A Functional on P2(R)\mathcal{P}_2(\mathbf{R}) Is Three Numbers

A linear functional on P2(R)\mathcal{P}_2(\mathbf{R}) is fixed by its values on 11, xx, x2x^2 — those values are its coordinates in the dual basis.

IVVisualization 3 — T′ Reverses the Arrow

For T(x,y)=(x+2y, 3x, x−y)T(x, y) = (x + 2y,\ 3x,\ x - y), each φ∈(R3)′\varphi \in (\mathbf{R}^3)' becomes T′(φ)=φ∘T∈(R2)′T'(\varphi) = \varphi \circ T \in (\mathbf{R}^2)'.

VQuiz Questions

Problem 1 · Dual Basis in R²

Given: the basis v1=(1,1)v_1 = (1, 1), v2=(1,2)v_2 = (1, 2) of R2\mathbf{R}^2 with dual basis φ1,φ2\varphi_1, \varphi_2 — find φ1(x,y)\varphi_1(x, y).

✅ Correct! φ1(1,1)=2−1=1\varphi_1(1, 1) = 2 - 1 = 1 and φ1(1,2)=2−2=0\varphi_1(1, 2) = 2 - 2 = 0, exactly the defining values.
❌ Not the first coordinate. The first-coordinate functional gives 11 at v2=(1,2)v_2 = (1, 2), but φ1\varphi_1 must vanish there.
❌ That is φ2\varphi_2. −x+y-x + y takes the value 00 at v1v_1 and 11 at v2v_2 — it reads the coefficient of v2v_2.
❌ That belongs to another basis. x−yx - y is φ1\varphi_1 for (1,0),(1,1)(1, 0), (1, 1); here it gives φ1(1,1)=0\varphi_1(1, 1) = 0, not 11.
❌ Not quite. Solve (x,y)=c1(1,1)+c2(1,2)(x, y) = c_1(1, 1) + c_2(1, 2) for c1c_1.
Show solution

By the coefficient-reader theorem, φ1(x,y)\varphi_1(x, y) is the coefficient c1c_1 in

(x,y)=c1(1,1)+c2(1,2)=(c1+c2, c1+2c2).(x, y) = c_1(1, 1) + c_2(1, 2) = (c_1 + c_2,\ c_1 + 2c_2).

Subtracting the first coordinate equation from the second gives c2=y−xc_2 = y - x, and then c1=x−c2=2x−yc_1 = x - c_2 = 2x - y. So

φ1(x,y)=2x−y,φ2(x,y)=−x+y.\varphi_1(x, y) = 2x - y, \qquad \varphi_2(x, y) = -x + y.

Check: φ1(1,1)=1\varphi_1(1, 1) = 1, φ1(1,2)=0\varphi_1(1, 2) = 0, φ2(1,1)=0\varphi_2(1, 1) = 0, φ2(1,2)=1\varphi_2(1, 2) = 1.

Problem 2 · Which Map Is a Linear Functional?

Given: four maps defined on P2(R)\mathcal{P}_2(\mathbf{R}) — find the one that is a linear functional, then its value at p(x)=1+3x−x2p(x) = 1 + 3x - x^2.

Which is a linear functional on P2(R)\mathcal{P}_2(\mathbf{R})?

What is φ(1+3x−x2)\varphi(1 + 3x - x^2) for that functional?

✅ Correct! Differentiation, evaluation and integration are linear, so 2p′(1)−∫01p2p'(1) - \int_0^1 p is linear, and here it equals 2−136=−162 - \tfrac{13}{6} = -\tfrac{1}{6}.
❌ Not linear. Doubling pp multiplies p(1) p(2)p(1)\,p(2) by 44, not 22.
❌ Not linear. A linear map sends 00 to 00, but this one sends the zero polynomial to 11.
❌ Linear, but not a functional. p↦p′p \mapsto p' takes values in P2(R)\mathcal{P}_2(\mathbf{R}), not in R\mathbf{R}; a functional must return a scalar.
❌ Check the definition. A linear functional is a linear map into the scalars F\mathbf{F}.
❌ Check the sign. The integral is subtracted: 2p′(1)−∫01p2p'(1) - \int_0^1 p.
❌ Recompute the two pieces. p′(x)=3−2xp'(x) = 3 - 2x gives p′(1)=1p'(1) = 1; ∫01(1+3x−x2) dx=1+32−13\int_0^1 (1 + 3x - x^2)\,dx = 1 + \tfrac{3}{2} - \tfrac{1}{3}.
Show solution

(a) fails homogeneity: φ(2p)=4 p(1) p(2)\varphi(2p) = 4\,p(1)\,p(2). (b) fails φ(0)=0\varphi(0) = 0. (d) is linear but lands in P2(R)\mathcal{P}_2(\mathbf{R}), so it is not a map into R\mathbf{R}. (c) is a linear combination of p↦p′(1)p \mapsto p'(1) and p↦∫01pp \mapsto \int_0^1 p, both linear functionals, so it is one.

For p(x)=1+3x−x2p(x) = 1 + 3x - x^2: p′(x)=3−2xp'(x) = 3 - 2x, so p′(1)=1p'(1) = 1, and

∫01(1+3x−x2) dx=1+32−13=136.\int_0^1 (1 + 3x - x^2)\,dx = 1 + \tfrac{3}{2} - \tfrac{1}{3} = \tfrac{13}{6}. φ(p)=2⋅1−136=−16.\varphi(p) = 2 \cdot 1 - \tfrac{13}{6} = -\tfrac{1}{6}.

Problem 3 · Dual Basis in R³

Given: the basis v1=(1,0,0)v_1 = (1, 0, 0), v2=(1,1,0)v_2 = (1, 1, 0), v3=(1,1,1)v_3 = (1, 1, 1) of R3\mathbf{R}^3 with dual basis φ1,φ2,φ3\varphi_1, \varphi_2, \varphi_3 — find φ2\varphi_2 and the coefficient φ1(4,7,2)\varphi_1(4, 7, 2).

What is φ2(x,y,z)\varphi_2(x, y, z)?

What is φ1(4,7,2)\varphi_1(4, 7, 2)?

✅ Correct! (4,7,2)=−3 v1+5 v2+2 v3(4, 7, 2) = -3\,v_1 + 5\,v_2 + 2\,v_3, so the dual basis reads off −3-3, 55, 22.
❌ Not the second coordinate. (x,y,z)↦y(x, y, z) \mapsto y gives 11 at v3=(1,1,1)v_3 = (1, 1, 1), but φ2(v3)\varphi_2(v_3) must be 00.
❌ Check the defining values. φ2\varphi_2 must give 00 at v1v_1, 11 at v2v_2 and 00 at v3v_3.
❌ That is the first coordinate. φ1\varphi_1 reads the coefficient of v1v_1, and v2v_2, v3v_3 also contribute to the first coordinate.
❌ Not quite. Solve for the coefficients from the last coordinate upward.
Show solution

Write (x,y,z)=c1(1,0,0)+c2(1,1,0)+c3(1,1,1)=(c1+c2+c3, c2+c3, c3)(x, y, z) = c_1(1, 0, 0) + c_2(1, 1, 0) + c_3(1, 1, 1) = (c_1 + c_2 + c_3,\ c_2 + c_3,\ c_3).

From the bottom: c3=zc_3 = z, then c2=y−zc_2 = y - z, then c1=x−c2−c3=x−yc_1 = x - c_2 - c_3 = x - y. By the coefficient-reader theorem,

φ1(x,y,z)=x−y,φ2(x,y,z)=y−z,φ3(x,y,z)=z.\varphi_1(x, y, z) = x - y, \qquad \varphi_2(x, y, z) = y - z, \qquad \varphi_3(x, y, z) = z.

So φ1(4,7,2)=4−7=−3\varphi_1(4, 7, 2) = 4 - 7 = -3.

Check: −3(1,0,0)+5(1,1,0)+2(1,1,1)=(4,7,2)-3(1, 0, 0) + 5(1, 1, 0) + 2(1, 1, 1) = (4, 7, 2).

Problem 4 · Computing Dual Maps

Given: T∈L(R2,R3)T \in \mathcal{L}(\mathbf{R}^2, \mathbf{R}^3) with T(x,y)=(x+y, 2x, x−3y)T(x, y) = (x + y,\ 2x,\ x - 3y) and φ∈(R3)′\varphi \in (\mathbf{R}^3)' with φ(u1,u2,u3)=u1−u2+2u3\varphi(u_1, u_2, u_3) = u_1 - u_2 + 2u_3; also D∈L(P3(R))D \in \mathcal{L}(\mathcal{P}_3(\mathbf{R})), Dp=p′Dp = p', and ψ(p)=∫01p\psi(p) = \int_0^1 p — find T′(φ)T'(\varphi) and (D′(ψ))(x3−4x)\big(D'(\psi)\big)(x^3 - 4x).

What is (T′(φ))(x,y)\big(T'(\varphi)\big)(x, y)?

What is (D′(ψ))(x3−4x)\big(D'(\psi)\big)(x^3 - 4x)?

✅ Correct! T′(φ)=φ∘TT'(\varphi) = \varphi \circ T is the functional x−5yx - 5y on R2\mathbf{R}^2, and (D′(ψ))(p)=∫01p′=p(1)−p(0)=−3\big(D'(\psi)\big)(p) = \int_0^1 p' = p(1) - p(0) = -3.
❌ That is φ\varphi itself. T′T' goes from (R3)′(\mathbf{R}^3)' to (R2)′(\mathbf{R}^2)', so T′(φ)T'(\varphi) is a functional of (x,y)(x, y).
❌ Distribute the coefficient 22. 2(x−3y)=2x−6y2(x - 3y) = 2x - 6y, not 2x−3y2x - 3y.
❌ Not quite. Substitute T(x,y)T(x, y) into φ\varphi: (x+y)−2x+2(x−3y)(x + y) - 2x + 2(x - 3y).
❌ That is ψ(p)\psi(p). D′(ψ)=ψ∘DD'(\psi) = \psi \circ D differentiates first: it integrates p′p', not pp.
❌ Not quite. (D′(ψ))(p)=ψ(Dp)=∫01p′\big(D'(\psi)\big)(p) = \psi(Dp) = \int_0^1 p'.
Show solution

The map TT. By definition T′(φ)=φ∘TT'(\varphi) = \varphi \circ T, so

(T′(φ))(x,y)=φ(x+y, 2x, x−3y)=(x+y)−2x+2(x−3y)=x−5y.\big(T'(\varphi)\big)(x, y) = \varphi(x + y,\ 2x,\ x - 3y) = (x + y) - 2x + 2(x - 3y) = x - 5y.

It is a linear functional on R2\mathbf{R}^2: T′T' sends (R3)′(\mathbf{R}^3)' to (R2)′(\mathbf{R}^2)'.

The map DD. (D′(ψ))(p)=(ψ∘D)(p)=ψ(p′)=∫01p′\big(D'(\psi)\big)(p) = (\psi \circ D)(p) = \psi(p') = \int_0^1 p', which by the fundamental theorem of calculus is p(1)−p(0)p(1) - p(0). For p(x)=x3−4xp(x) = x^3 - 4x:

(D′(ψ))(p)=(1−4)−0=−3.\big(D'(\psi)\big)(p) = (1 - 4) - 0 = -3.

Directly: p′(x)=3x2−4p'(x) = 3x^2 - 4 and ∫01(3x2−4) dx=1−4=−3\int_0^1 (3x^2 - 4)\,dx = 1 - 4 = -3.

Solved: 0 / 4