Linear AlgebraLinear Transformations6 of 7
Linear Transformations · 6

Invertibility and Isomorphism

When a linear transformation moves vectors from one space to another, we often want to know whether we can undo it to get every starting vector back.

1

Undoing a Linear Transformation

For a linear transformation T:V→WT: V \to W that sends a vector v\built{v} to an output w\built{w}, its inverse transformation T−1:W→VT^{-1}: W \to V is the rule that sends every output w\built{w} right back to the single input v\built{v} it came from.

What Makes a Linear Map Invertible
VV
WW
TT
v\built{v}
w\built{w}
T−1 ?T^{-1}\,?

To reverse every arrow cleanly, TT has to pass two tests at the same time:

One-to-one (no squishing)⏟only 0 maps to 0: Kernel={0}andOnto (full coverage)⏟the image fills all of W\displaystyle \begin{gathered} \underbrace{\text{One-to-one (no squishing)}}_{\text{only } 0 \text{ maps to } 0\text{: } \text{Kernel} = \{0\}} \\ \text{and} \\ \underbrace{\text{Onto (full coverage)}}_{\text{the image fills all of } W} \end{gathered}

When you invert a matrix in Rn\mathbb{R}^n by setting up [A∣I][A \mid I] and row-reducing to [I∣A−1][I \mid \inverse{A^{-1}}], the columns on the left are where the basis vectors landed, and row operations unscramble that stretched grid step-by-step back to the standard grid:

[12101301]→R2←R2−R1[121001−11]→R1←R1−2R2[103−201−11]\displaystyle \begin{aligned} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{2} & \inverse{1} & \inverse{0} \\ \blockone{1} & \blocktwo{3} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] \xrightarrow{\rowop{R_2 \leftarrow R_2 - R_1}} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{2} & \inverse{1} & \inverse{0} \\ \blockone{0} & \blocktwo{1} & \inverse{-1} & \inverse{1} \end{array}\right] \\[6pt] \xrightarrow{\rowop{R_1 \leftarrow R_1 - 2R_2}} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{0} & \inverse{3} & \inverse{-2} \\ \blockone{0} & \blocktwo{1} & \inverse{-1} & \inverse{1} \end{array}\right] \end{aligned}

If row reduction hits a row of all zeros on the left, the matrix squished a dimension flat, so no inverse exists.

As long as the transformed basis vectors point in independent directions, T−1T^{-1} unscrambles the stretched grid in WW right back to the standard square grid in VV. The exact instant one basis vector lands on the line of the other, the grid in WW collapses flat onto a single line and fails both one-to-one and onto at once.

A transformation can be undone as long as the stretched grid covers the whole plane, and stops being reversible the instant the grid flattens onto a line.
2

Isomorphism: Same Space, Different Labels

Once we know when a linear map can be undone, we can compare two different-looking vector spaces. An isomorphism is simply an invertible linear transformation between two spaces VV and WW, telling us that both spaces work the exact same way under addition and scaling and only differ in the labels on their basis vectors.

Isomorphism: Same Space, Different Labels
degree≤2\text{degree} \le 2
3+2x+5x23 + 2x + 5x^2
R3\mathbb{R}^3
[325]\begin{bmatrix} 3 \\ 2 \\ 5 \end{bmatrix}

In the space of polynomials up to degree 22, the powers 1\blockone{1}, x\blocktwo{x}, and x2\blockthree{x^2} act as labels keeping three numbers from mixing during addition, doing the exact same job as the three vertical slots in R3\mathbb{R}^3:

a(1)⏟slot 1+b(x)⏟slot 2+c(x2)⏟slot 3  ⟷  [abc]\displaystyle \blockone{a}\underbrace{\blockone{(1)}}_{\text{slot } 1} + \blocktwo{b}\underbrace{\blocktwo{(x)}}_{\text{slot } 2} + \blockthree{c}\underbrace{\blockthree{(x^2)}}_{\text{slot } 3} \;\longleftrightarrow\; \begin{bmatrix} \blockone{a} \\ \blocktwo{b} \\ \blockthree{c} \end{bmatrix}

Because more starting directions than target directions forces a squish, and fewer starting directions leaves empty space, two spaces are isomorphic only when they have the exact same number of basis vectors:

dim⁡V\displaystyle \dim V =dim⁡W\displaystyle = \dim W
3

Building an Isomorphism Between Spaces

For spaces like polynomials a+bx+cx2\blockone{a} + \blocktwo{bx} + \blockthree{cx^2} or 2×22 \times 2 matrices, the basis is easy to spot right away. What if a space hides its true dimension behind an equation, like the plane x+2y−z=0x + 2y - z = 0 inside R3\mathbb{R}^3, and we want to build an explicit isomorphism T:V→WT: V \to W between that plane and the space of polynomials of degree up to 11, a+bt\blockone{a} + \blocktwo{bt}?

Building an Isomorphism Between Spaces
x\htmlClass{lc-mute}{x}
y\htmlClass{lc-mute}{y}
z\htmlClass{lc-mute}{z}
V ⁣: x+2y−z=0V\colon\ x + 2y - z = 0
TT
W ⁣: a+btW\colon\ \blockone{a} + \blocktwo{bt}
isomorphic?\text{isomorphic?}
Text solution
Solve the equation to find the basis of VVRewrite the plane equation x+2y−z=0x + 2y - z = 0 as z=x+2yz = x + 2y and split a general vector on the plane by the free weights x\blockone{x} and y\blocktwo{y} to find its independent basis vectors.
[xyz]=[xyx+2y]=x[101]⏟first basis vector+y[012]⏟second basis vector\displaystyle \begin{aligned} \built{\begin{bmatrix} x \\ y \\ z \end{bmatrix}} &= \begin{bmatrix} \blockone{x} \\ \blocktwo{y} \\ \blockone{x} + \blocktwo{2y} \end{bmatrix} \\ &= \blockone{x}\underbrace{\blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}}}_{\text{first basis vector}} \\ &\quad + \blocktwo{y}\underbrace{\blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}}}_{\text{second basis vector}} \end{aligned}
Match dimensions and pair the two bases one-to-oneSince the plane VV is built from 22 basis vectors and WW is built from the 22 basis polynomials {1,t}\{\blockone{1}, \blocktwo{t}\}, both spaces have dimension 22, and we build TT by matching their basis vectors directly.
[101]⟷1,[012]⟷t\displaystyle \blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}} \longleftrightarrow \blockone{1}, \qquad \blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}} \longleftrightarrow \blocktwo{t}
Keep the same basis weights going forward and backwardLinearity carries the weights x\blockone{x} and y\blocktwo{y} straight onto 1\blockone{1} and t\blocktwo{t} to form TT, while T−1T^{-1} swaps the plane's basis vectors back in for 1\blockone{1} and t\blocktwo{t}.
T ⁣([xyz])=x(1)+y(t)=x+yt⟹T ⁣([3513])=3+5t\displaystyle \begin{aligned} T\!\left(\built{\begin{bmatrix} x \\ y \\ z \end{bmatrix}}\right) &= \blockone{x(1)} + \blocktwo{y(t)} \\ &= \blockone{x} + \blocktwo{yt} \\ \Longrightarrow \quad T\!\left(\built{\begin{bmatrix} 3 \\ 5 \\ 13 \end{bmatrix}}\right) &= \blockone{3} + \blocktwo{5t} \end{aligned}
T−1(3+5t)=3[101]+5[012]=[3513]\displaystyle \begin{aligned} T^{-1}(\blockone{3} + \blocktwo{5t}) &= \blockone{3}\blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}} + \blocktwo{5}\blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}} \\ &= \built{\begin{bmatrix} 3 \\ 5 \\ 13 \end{bmatrix}} \end{aligned}

Every vector on the tilted plane x+2y−z=0x + 2y - z = 0 in R3\mathbb{R}^3 splits along the two basis directions [101]\blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}} and [012]\blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}} with weights x\blockone{x} and y\blocktwo{y}, matching the polynomial x+yt\blockone{x} + \blocktwo{yt} built from 1\blockone{1} and t\blocktwo{t} with the exact same weights.

Every vector on this tilted 3D plane is built from just two basis directions, matching it one-to-one with the polynomial x + yt, built from the exact same two weights.
4

The Equal-Dimension Shortcut

Checking both one-to-one and onto separately takes two steps, so we want a faster way to test whether a linear map T:V→WT: V \to W between two spaces of the same dimension is invertible.

Remember from the Rank-Nullity Theorem that the dimension of the starting space acts as a fixed budget split between dimensions that get squished to zero and dimensions that survive into the image:

dim⁡(Kernel)⏟squished to 0+dim⁡(Image)⏟survives in W\displaystyle \underbrace{\lost{\dim(\text{Kernel})}}_{\text{squished to } 0} + \underbrace{\built{\dim(\text{Image})}}_{\text{survives in } W} =dim⁡(Starting Space)\displaystyle = \dim(\text{Starting Space})
The Equal-Dimension Shortcut
VV
WW
TT
one-to-one?
no squishing
onto?
full coverage
dim⁡V=dim⁡W\dim V = \dim W

When the starting space and target space have the same dimension (dim⁡V=dim⁡W\dim V = \dim W), checking that nothing gets squished (dim⁡(Kernel)=0\lost{\dim(\text{Kernel})} = 0) forces the image to have the full dimension of WW:

0+dim⁡(Image)=dim⁡W⟹dim⁡(Image)=dim⁡W\displaystyle \begin{aligned} \lost{0} + \built{\dim(\text{Image})} &= \dim W \\ \Longrightarrow \quad \built{\dim(\text{Image})} &= \dim W \end{aligned}

An image of dimension dim⁡W\dim W inside WW automatically fills the entire target space, giving onto for free.

5

Inverting a Map by Crossing to Rn\mathbb{R}^n

Now suppose we are given the calculus transformation T(p)=p(x)+p′(x)T(p) = p(x) + p'(x) on polynomials up to degree 22 (a+bx+cx2\blockone{a} + \blocktwo{bx} + \blockthree{cx^2}), and we want to check whether TT is invertible and find the starting polynomial p(x)p(x) that produces 5+3x+x25 + 3x + x^2. Instead of guessing how to undo a derivative in polynomial space, we can use our coordinate isomorphism to cross into R3\mathbb{R}^3, invert the matrix there, and cross back.

Inverting a Map by Crossing to ℝⁿ
T(p)\displaystyle T(p) =p(x)+p′(x)\displaystyle = p(x) + p'(x)
p(x)\displaystyle p(x) =  ?  →    T      5+3x+x2\displaystyle = \;? \;\xrightarrow{\;\;T\;\;}\; 5 + 3x + x^2
invertible?\displaystyle \text{invertible?}
p(x)=4+x+x2p(x) = 4 + x + x^2
5+3x+x25 + 3x + x^2
TT
44
11
11
1\blockone{1}
x\blocktwo{x}
x2\blockthree{x^2}
55
33
11
1\blockone{1}
x\blocktwo{x}
x2\blockthree{x^2}
A−1\inverse{A^{-1}}
R3\mathbb{R}^3
labels off
labels on
Text solution
Plug the basis into TT to build the matrix in R3\mathbb{R}^3Plug each vector from the basis {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\} into T(p)=p(x)+p′(x)T(p) = p(x) + p'(x) and record the output coordinates as the columns of a 3×33 \times 3 matrix.
T(1)=1  ⟷  [100]T(x)=1+x  ⟷  [110]T(x2)=2x+x2  ⟷  [021]⟹A=[110012001]\displaystyle \begin{gathered}\begin{aligned} T(\blockone{1}) &= 1 \;\longleftrightarrow\; \blockone{\begin{bmatrix} 1 \\ 0 \\ 0 \end{bmatrix}} \\ T(\blocktwo{x}) &= 1 + x \;\longleftrightarrow\; \blocktwo{\begin{bmatrix} 1 \\ 1 \\ 0 \end{bmatrix}} \\ T(\blockthree{x^2}) &= 2x + x^2 \;\longleftrightarrow\; \blockthree{\begin{bmatrix} 0 \\ 2 \\ 1 \end{bmatrix}} \end{aligned} \\ \Longrightarrow \quad A = \begin{bmatrix} \blockone{1} & \blocktwo{1} & \blockthree{0} \\ \blockone{0} & \blocktwo{1} & \blockthree{2} \\ \blockone{0} & \blocktwo{0} & \blockthree{1} \end{bmatrix}\end{gathered}
Check that AA is invertible and row-reduce next to the identityThe 11s down the diagonal, with 00s below them, show that no column is squished into the others (dim⁡(Kernel)=0\lost{\dim(\text{Kernel})} = 0, so the equal-dimension shortcut guarantees TT is invertible), and two row operations unscramble AA into the identity to give A−1\inverse{A^{-1}}.
[110100012010001001]→R2←R2−2R3[11010001001−2001001]→R1←R1−R2[1001−1201001−2001001]\displaystyle \begin{aligned} & \left[\begin{array}{ccc|ccc} 1 & 1 & 0 & \inverse{1} & \inverse{0} & \inverse{0} \\ 0 & 1 & 2 & \inverse{0} & \inverse{1} & \inverse{0} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] & \xrightarrow{\rowop{R_2 \leftarrow R_2 - 2R_3}} \\ & \left[\begin{array}{ccc|ccc} 1 & 1 & 0 & \inverse{1} & \inverse{0} & \inverse{0} \\ 0 & 1 & 0 & \inverse{0} & \inverse{1} & \inverse{-2} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] & \xrightarrow{\rowop{R_1 \leftarrow R_1 - R_2}} \\ & \left[\begin{array}{ccc|ccc} 1 & 0 & 0 & \inverse{1} & \inverse{-1} & \inverse{2} \\ 0 & 1 & 0 & \inverse{0} & \inverse{1} & \inverse{-2} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \end{aligned}
Multiply by A−1\inverse{A^{-1}} and put the polynomial labels back onTranslate the target polynomial 5+3x+x25 + 3x + x^2 into [531]\begin{bmatrix} 5 \\ 3 \\ 1 \end{bmatrix}, multiply by A−1\inverse{A^{-1}}, and put the 1,x,x2\blockone{1}, \blocktwo{x}, \blockthree{x^2} labels back onto the three slots.
[1−1201−2001][531]=[5−3+23−21]=[411]  ⟷  p(x)=4+x+x2\displaystyle \begin{aligned} & \inverse{\begin{bmatrix} 1 & -1 & 2 \\ 0 & 1 & -2 \\ 0 & 0 & 1 \end{bmatrix}}\begin{bmatrix} 5 \\ 3 \\ 1 \end{bmatrix} \\ &= \begin{bmatrix} 5 - 3 + 2 \\ 3 - 2 \\ 1 \end{bmatrix} \\ &= \begin{bmatrix} 4 \\ 1 \\ 1 \end{bmatrix} \\ &\;\longleftrightarrow\; p(x) = 4 + x + x^2 \end{aligned}
Read the general calculus inverse from the columns of A−1\inverse{A^{-1}}Putting the {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\} labels back onto each column of A−1\inverse{A^{-1}} shows that 1↦1\blockone{1} \mapsto 1, x↦−1+x\blocktwo{x} \mapsto -1 + x, and x2↦2−2x+x2\blockthree{x^2} \mapsto 2 - 2x + x^2, which subtracts the first derivative and adds the second derivative.
T−1(q)\displaystyle T^{-1}(q) =q(x)−q′(x)+q′′(x)\displaystyle = q(x) - q'(x) + q''(x)

Key concepts

Everything here comes down to matching dimensions and tracking basis vectors. Before testing whether a map T:V→WT: V \to W is invertible, compare dim⁡V\dim V and dim⁡W\dim W by counting independent basis vectors, remembering to substitute any subspace equations first (so the plane z=x+2yz = x + 2y in R3\mathbb{R}^3 has basis {[101],[012]}\left\{\blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}}, \blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}}\right\} and dimension 22, not 33). If dim⁡V≠dim⁡W\dim V \neq \dim W, stop immediately: you don't need to check the kernel or row-reduce, because unequal dimensions force either a squish or empty space, so no isomorphism can exist (for instance, polynomials of degree up to 22, a+bx+cx2\blockone{a} + \blocktwo{bx} + \blockthree{cx^2}, have dimension 33 and can never be isomorphic to 2×22 \times 2 matrices of dimension 44).

When dim⁡V=dim⁡W\dim V = \dim W, the spaces are isomorphic, and you can build an explicit isomorphism by pairing their basis vectors one-to-one ([101]↔1\blockone{\begin{bmatrix} 1 \\ 0 \\ 1 \end{bmatrix}} \leftrightarrow \blockone{1} and [012]↔t\blocktwo{\begin{bmatrix} 0 \\ 1 \\ 2 \end{bmatrix}} \leftrightarrow \blocktwo{t}, giving T ⁣([xyz])=x+ytT\!\left(\built{\begin{bmatrix} x \\ y \\ z \end{bmatrix}}\right) = \blockone{x} + \blocktwo{yt}). If you are handed a specific linear map TT between two spaces of equal dimension, don't do the extra work of checking both one-to-one and onto: the Rank-Nullity budget dim⁡(Kernel)+dim⁡(Image)=dim⁡V\lost{\dim(\text{Kernel})} + \built{\dim(\text{Image})} = \dim V ties them together, so showing that only 00 maps to 00 (dim⁡(Kernel)=0\lost{\dim(\text{Kernel})} = 0) guarantees full coverage automatically. Finally, to invert a map like T(p)=p(x)+p′(x)T(p) = p(x) + p'(x), take off the basis labels {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\} to write the basis outputs as a matrix AA in Rn\mathbb{R}^n, row-reduce [A∣I]→[I∣A−1][A \mid I] \to [I \mid \inverse{A^{-1}}] to unscramble the grid, and put the basis labels back on at the end.

6

Practice

1

Try to invert the matrices A=[1327]A = \begin{bmatrix} 1 & 3 \\ 2 & 7 \end{bmatrix} and B=[1326]B = \begin{bmatrix} 1 & 3 \\ 2 & 6 \end{bmatrix} by row-reducing each next to the identity matrix. Find A−1\inverse{A^{-1}}, and explain what the final left side of BB's reduction tells you about what BB does to the 2D grid.

Start both reductions with the row operation that clears the 22 in the bottom-left corner, and watch whether the second column survives.

For AA, set up [A∣I][A \mid I] and clear the entries off the diagonal:
[13102701]→R2←R2−2R1[131001−21]→R1←R1−3R2[107−301−21]\displaystyle \begin{aligned} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{3} & \inverse{1} & \inverse{0} \\ \blockone{2} & \blocktwo{7} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] \xrightarrow{\rowop{R_2 \leftarrow R_2 - 2R_1}} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{3} & \inverse{1} & \inverse{0} \\ \blockone{0} & \blocktwo{1} & \inverse{-2} & \inverse{1} \end{array}\right] \\[6pt] \xrightarrow{\rowop{R_1 \leftarrow R_1 - 3R_2}} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{0} & \inverse{7} & \inverse{-3} \\ \blockone{0} & \blocktwo{1} & \inverse{-2} & \inverse{1} \end{array}\right] \end{aligned}
(Watch the sign on the second step: subtracting 33 times −2-2 gives 1−3(−2)=71 - 3(-2) = 7; dropping the minus sign is a common slip that produces the wrong top-left entry −5-5 and [−5−3−21]\begin{bmatrix} -5 & -3 \\ -2 & 1 \end{bmatrix}.)
Reading the right side gives A−1=[7−3−21]\inverse{A^{-1}} = \inverse{\begin{bmatrix} 7 & -3 \\ -2 & 1 \end{bmatrix}}.
For BB, applying the first row operation gives:
[13102601]→R2←R2−2R1[131000−21]\displaystyle \begin{aligned} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{3} & \inverse{1} & \inverse{0} \\ \blockone{2} & \blocktwo{6} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] \xrightarrow{\rowop{R_2 \leftarrow R_2 - 2R_1}} & \left[\begin{array}{cc|cc} \blockone{1} & \blocktwo{3} & \inverse{1} & \inverse{0} \\ \lost{0} & \lost{0} & \inverse{-2} & \inverse{1} \end{array}\right] \end{aligned}
The row of all zeros on the left shows that BB is not invertible: both basis columns [12]\blockone{\begin{bmatrix} 1 \\ 2 \end{bmatrix}} and [36]\blocktwo{\begin{bmatrix} 3 \\ 6 \end{bmatrix}} landed on the same line, squishing the 2D grid flat into a 1D line that row operations cannot un-squish.
2

Consider four vector spaces: UU (polynomials of degree up to 22, a+bx+cx2a + bx + cx^2), VV (polynomials of degree up to 33, a+bx+cx2+dx3a + bx + cx^2 + dx^3), WW (2×22 \times 2 matrices), and R3\mathbb{R}^3. Which pairs of spaces are isomorphic to each other, and why can there never be an isomorphism between UU and WW?

Count the number of independent basis vectors (or free number slots) in each space rather than looking at the highest exponent.

Count the basis vectors in each space:
UU has the 33-vector basis {1,x,x2}\{1, x, x^2\}, so dim⁡U=3\dim U = 3.
VV has the 44-vector basis {1,x,x2,x3}\{1, x, x^2, x^3\}, so dim⁡V=4\dim V = 4.
(Don't read the highest exponent as the dimension: forgetting the constant term 11 gives the wrong dimensions dim⁡U=2\dim U = 2 and dim⁡V=3\dim V = 3, which mistakenly pairs VV with R3\mathbb{R}^3.)
WW has 44 independent matrix slots, so dim⁡W=4\dim W = 4.
R3\mathbb{R}^3 has 33 coordinate slots, so dim⁡R3=3\dim \mathbb{R}^3 = 3.
Matching equal dimensions shows that UU and R3\mathbb{R}^3 are isomorphic (both dimension 33, isomorphic to R3\mathbb{R}^3), and VV and WW are isomorphic (both dimension 44, isomorphic to R4\mathbb{R}^4). There can never be an isomorphism between UU (dimension 33) and WW (dimension 44) because any linear map from UU to WW can sweep out at most a 33-dimensional subspace inside 44-dimensional WW (leaving empty space, so not onto), while any linear map from WW to UU must squish at least one of WW's 44 independent directions flat to fit into 33-dimensional UU (failing one-to-one).
3

Let TT be a linear transformation from the space of 2×22 \times 2 matrices to the space of polynomials of degree up to 33 (a+bx+cx2+dx3a + bx + cx^2 + dx^3). Suppose you check the kernel of TT and find that only the zero matrix [0000]\begin{bmatrix} 0 & 0 \\ 0 & 0 \end{bmatrix} maps to the zero polynomial. Without calculating the image of TT, determine whether every polynomial of degree up to 33 gets hit by TT and whether TT is an isomorphism. Explain why using the Rank-Nullity Theorem.

Compare the dimensions of the two spaces first, then plug dim⁡(Kernel)\lost{\dim(\text{Kernel})} into the Rank-Nullity dimension budget.

Both the starting space (2×22 \times 2 matrices with 44 slots) and the target space (polynomials up to degree 33 with basis {1,x,x2,x3}\{1, x, x^2, x^3\}) have dimension 44. Since only the zero matrix maps to zero, no directions get squished, so TT is one-to-one and:
dim⁡(Kernel)\displaystyle \lost{\dim(\text{Kernel})} =0\displaystyle = 0
(Be careful not to write dim⁡(Kernel)=1\lost{\dim(\text{Kernel})} = 1 just because the kernel contains one matrix: the zero matrix alone spans 00 directions, and using 11 gives the wrong image dimension 4−1=34 - 1 = 3 and the false conclusion that TT is not onto.)
Plugging dim⁡(Kernel)=0\lost{\dim(\text{Kernel})} = 0 into the Rank-Nullity Theorem gives:
dim⁡(Kernel)+dim⁡(Image)=dim⁡(Starting Space)0+dim⁡(Image)=4dim⁡(Image)=4\displaystyle \begin{aligned} & \lost{\dim(\text{Kernel})} + \built{\dim(\text{Image})} \\ &\quad = \dim(\text{Starting Space}) \\ & \lost{0} + \built{\dim(\text{Image})} = 4 \\ & \built{\dim(\text{Image})} = 4 \end{aligned}
Because a 44-dimensional image inside a 44-dimensional target space fills the entire space, every polynomial of degree up to 33 gets hit by TT (onto) and TT is an isomorphism.
4

Let VV be the subspace of R3\mathbb{R}^3 made up of all vectors [xyz]\begin{bmatrix} x \\ y \\ z \end{bmatrix} satisfying 2x−y−z=02x - y - z = 0, and let WW be the space of polynomials of degree up to 11, a+bt\blockone{a} + \blocktwo{bt}. Find a basis for VV to show that VV and WW are isomorphic, build an explicit isomorphism T:V→WT: V \to W by pairing your basis with {1,t}\{\blockone{1}, \blocktwo{t}\}, and compute both T ⁣([27−3])T\!\left(\built{\begin{bmatrix} 2 \\ 7 \\ -3 \end{bmatrix}}\right) and T−1(4−3t)T^{-1}(4 - 3t).

Solve the plane equation for zz and substitute it into [xyz]\begin{bmatrix} x \\ y \\ z \end{bmatrix} to split the vector into xx times one basis vector plus yy times another.

Rewrite 2x−y−z=02x - y - z = 0 as z=2x−yz = 2x - y and split a general vector on the plane by x\blockone{x} and y\blocktwo{y}:
[xyz]=[xy2x−y]=x[102]+y[01−1]\displaystyle \begin{aligned} \built{\begin{bmatrix} x \\ y \\ z \end{bmatrix}} &= \begin{bmatrix} \blockone{x} \\ \blocktwo{y} \\ \blockone{2x} - \blocktwo{y} \end{bmatrix} \\ &= \blockone{x}\blockone{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}} + \blocktwo{y}\blocktwo{\begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}} \end{aligned}
So {[102],[01−1]}\left\{\blockone{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}}, \blocktwo{\begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}}\right\} is a basis for VV, giving dim⁡V=2=dim⁡W\dim V = 2 = \dim W (with basis {1,t}\{\blockone{1}, \blocktwo{t}\}), which shows VV and WW are isomorphic.
Pairing [102]↔1\blockone{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}} \leftrightarrow \blockone{1} and [01−1]↔t\blocktwo{\begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}} \leftrightarrow \blocktwo{t} gives the explicit isomorphism and its inverse:
T ⁣([xyz])=x+yt,T−1(a+bt)=a[102]+b[01−1]=[ab2a−b]\displaystyle \begin{aligned} T\!\left(\built{\begin{bmatrix} x \\ y \\ z \end{bmatrix}}\right) &= \blockone{x} + \blocktwo{yt}, \\ T^{-1}(\blockone{a} + \blocktwo{bt}) &= \blockone{a}\blockone{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}} + \blocktwo{b}\blocktwo{\begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}} \\ &= \built{\begin{bmatrix} a \\ b \\ 2a - b \end{bmatrix}} \end{aligned}
Plugging the given inputs into both maps:
T ⁣([27−3])=2+7t\displaystyle T\!\left(\built{\begin{bmatrix} 2 \\ 7 \\ -3 \end{bmatrix}}\right) = \blockone{2} + \blocktwo{7t}
T−1(4−3t)=4[102]−3[01−1]=[4−311]\displaystyle \begin{aligned} T^{-1}(\blockone{4} - \blocktwo{3t}) &= \blockone{4}\blockone{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}} - \blocktwo{3}\blocktwo{\begin{bmatrix} 0 \\ 1 \\ -1 \end{bmatrix}} \\ &= \built{\begin{bmatrix} 4 \\ -3 \\ 11 \end{bmatrix}} \end{aligned}
(Watch the double negative in the third slot: 2(4)−(−3)=112(4) - (-3) = 11; adding instead of subtracting gives 8−3=58 - 3 = 5 and the wrong vector [4−35]\begin{bmatrix} 4 \\ -3 \\ 5 \end{bmatrix}, which doesn't lie on the plane 2x−y−z=02x - y - z = 0.)
5

On the space of polynomials up to degree 22 (a+bx+cx2\blockone{a} + \blocktwo{bx} + \blockthree{cx^2}), define the linear transformation T(p)=p(x)−p′(x)T(p) = p(x) - p'(x). Using the coordinate isomorphism with basis {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\}, write the 3×33 \times 3 matrix for TT in R3\mathbb{R}^3 and explain why TT is invertible. Then invert the matrix to find the polynomial p(x)p(x) that produces the output 7−x+2x27 - x + 2x^2, and read the columns of your inverse matrix to write a general calculus formula for T−1(q)T^{-1}(q).

Plug 1\blockone{1}, x\blocktwo{x}, and x2\blockthree{x^2} into p(x)−p′(x)p(x) - p'(x) one at a time, keeping the constant term in the top row and the x2\blockthree{x^2} term in the bottom row.

Plug each basis polynomial from {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\} into T(p)=p(x)−p′(x)T(p) = p(x) - p'(x) to form the columns of AA:
T(1)=1  ⟷  [100]T(x)=−1+x  ⟷  [−110]T(x2)=−2x+x2  ⟷  [0−21]⟹A=[1−1001−2001]\displaystyle \begin{gathered}\begin{aligned} T(\blockone{1}) &= 1 \;\longleftrightarrow\; \blockone{\begin{bmatrix} 1 \\ 0 \\ 0 \end{bmatrix}} \\ T(\blocktwo{x}) &= -1 + x \;\longleftrightarrow\; \blocktwo{\begin{bmatrix} -1 \\ 1 \\ 0 \end{bmatrix}} \\ T(\blockthree{x^2}) &= -2x + x^2 \;\longleftrightarrow\; \blockthree{\begin{bmatrix} 0 \\ -2 \\ 1 \end{bmatrix}} \end{aligned} \\ \Longrightarrow \quad A = \begin{bmatrix} \blockone{1} & \blocktwo{-1} & \blockthree{0} \\ \blockone{0} & \blocktwo{1} & \blockthree{-2} \\ \blockone{0} & \blocktwo{0} & \blockthree{1} \end{bmatrix}\end{gathered}
The 11s down the diagonal, with 00s below them, show that no column is squished into the others (dim⁡(Kernel)=0\lost{\dim(\text{Kernel})} = 0, one-to-one), and because both spaces have dimension 33, the equal-dimension shortcut guarantees TT is onto and therefore invertible.
Row-reduce [A∣I][A \mid I] to find A−1\inverse{A^{-1}}:
[1−1010001−2010001001]→R2←R2+2R3[1−10100010012001001]→R1←R1+R2[100112010012001001]\displaystyle \begin{aligned} & \left[\begin{array}{ccc|ccc} 1 & -1 & 0 & \inverse{1} & \inverse{0} & \inverse{0} \\ 0 & 1 & -2 & \inverse{0} & \inverse{1} & \inverse{0} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] & \xrightarrow{\rowop{R_2 \leftarrow R_2 + 2R_3}} \\ &\quad \left[\begin{array}{ccc|ccc} 1 & -1 & 0 & \inverse{1} & \inverse{0} & \inverse{0} \\ 0 & 1 & 0 & \inverse{0} & \inverse{1} & \inverse{2} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \\[6pt] & \xrightarrow{\rowop{R_1 \leftarrow R_1 + R_2}} \\ &\quad \left[\begin{array}{ccc|ccc} 1 & 0 & 0 & \inverse{1} & \inverse{1} & \inverse{2} \\ 0 & 1 & 0 & \inverse{0} & \inverse{1} & \inverse{2} \\ 0 & 0 & 1 & \inverse{0} & \inverse{0} & \inverse{1} \end{array}\right] \end{aligned}
(Because AA has negative entries above the diagonal, you must add rows to clear them, giving all positive entries in A−1=[112012001]\inverse{A^{-1}} = \inverse{\begin{bmatrix} 1 & 1 & 2 \\ 0 & 1 & 2 \\ 0 & 0 & 1 \end{bmatrix}}; subtracting rows out of habit gives [1−1201−2001]\begin{bmatrix} 1 & -1 & 2 \\ 0 & 1 & -2 \\ 0 & 0 & 1 \end{bmatrix} and the wrong polynomial 12−5x+2x212 - 5x + 2x^2.)
Multiply A−1\inverse{A^{-1}} by the coordinate vector [7−12]\begin{bmatrix} 7 \\ -1 \\ 2 \end{bmatrix} for 7−x+2x27 - x + 2x^2:
[112012001][7−12]=[7−1+4−1+42]=[1032]  ⟷  p(x)=10+3x+2x2\displaystyle \begin{aligned} & \inverse{\begin{bmatrix} 1 & 1 & 2 \\ 0 & 1 & 2 \\ 0 & 0 & 1 \end{bmatrix}}\begin{bmatrix} 7 \\ -1 \\ 2 \end{bmatrix} \\ &\quad = \begin{bmatrix} 7 - 1 + 4 \\ -1 + 4 \\ 2 \end{bmatrix} \\ &\quad = \begin{bmatrix} 10 \\ 3 \\ 2 \end{bmatrix} \;\longleftrightarrow\; p(x) \\ &\quad = 10 + 3x + 2x^2 \end{aligned}
Putting the {1,x,x2}\{\blockone{1}, \blocktwo{x}, \blockthree{x^2}\} labels back onto the three columns of A−1\inverse{A^{-1}} shows that 1↦1\blockone{1} \mapsto 1, x↦1+x\blocktwo{x} \mapsto 1 + x, and x2↦2+2x+x2\blockthree{x^2} \mapsto 2 + 2x + x^2, which gives the general inverse formula:
T−1(q)\displaystyle T^{-1}(q) =q(x)+q′(x)+q′′(x)\displaystyle = q(x) + q'(x) + q''(x)