Linear AlgebraVector Spaces and Subspaces7 of 9
Vector Spaces and Subspaces · 7

Coordinates in a Basis

We are used to describing a vector by how far right and how far up it goes, but we can describe that exact same arrow using any set of step vectors we choose.

1

What coordinates in a basis mean

When we write a vector like [32]\begin{bmatrix} \first{3} \\ \second{2} \end{bmatrix} on the standard grid, those two numbers are instructions telling us how many horizontal steps ı^=[10]\first{\hat{\imath}} = \first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} and vertical steps ȷ^=[01]\second{\hat{\jmath}} = \second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} to take from the origin. If we switch from those standard steps to a new basis BB made of slanted step vectors b1=[21]\first{b_1} = \first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} and b2=[−11]\second{b_2} = \second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}}, we want to see what a coordinate pair like [21]B\begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B means and how to find where it lands in standard xx-yy coordinates.

What coordinates in a basis mean
x\htmlClass{lc-mute}{x}
y\htmlClass{lc-mute}{y}
[32]\begin{bmatrix} \htmlClass{lc-green}{3} \\ \htmlClass{lc-pink}{2} \end{bmatrix}
ı^=[10]\htmlClass{lc-green}{\hat{\imath}} = \htmlClass{lc-green}{\begin{bmatrix} 1 \\ 0 \end{bmatrix}}
ȷ^=[01]\htmlClass{lc-pink}{\hat{\jmath}} = \htmlClass{lc-pink}{\begin{bmatrix} 0 \\ 1 \end{bmatrix}}

The subscript BB on [21]B\begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B is not an operation: it just labels that these numbers count steps of b1\first{b_1} and b2\second{b_2} rather than standard horizontal and vertical steps.

Text solution
1. Multiply each basis vector by its BB-coordinate and addSince [21]B\begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B means taking 2\first{2} steps of b1\first{b_1} and 1\second{1} step of b2\second{b_2}, we multiply each basis vector by its weight and add their entries.
2[21]⏟b1+1[−11]⏟b2=[42]+[−11]=[33]\displaystyle \first{2}\underbrace{\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}}}_{\first{b_1}} + \second{1}\underbrace{\second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}}}_{\second{b_2}} = \first{\begin{bmatrix} 4 \\ 2 \end{bmatrix}} + \second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}} = \target{\begin{bmatrix} 3 \\ 3 \end{bmatrix}}
2. Put the basis vectors into the columns of a matrixPutting b1\first{b_1} and b2\second{b_2} side-by-side into a basis matrix P\op{P} lets us write that exact same linear combination as a single matrix multiplication.
[2−111]⏟basis matrix P[21]B=[33]\displaystyle \underbrace{\begin{bmatrix} \first{2} & \second{-1} \\ \first{1} & \second{1} \end{bmatrix}}_{\text{basis matrix } \op{P}}\begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B = \target{\begin{bmatrix} 3 \\ 3 \end{bmatrix}}

So [21]B\begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B and [33]\target{\begin{bmatrix} 3 \\ 3 \end{bmatrix}} are two different coordinate recipes that land on the exact same spot on the board.

Moving a vector parallel to one of the slanted grid lines changes only one of its BB-coordinates while changing both xx and yy, because b1\first{b_1} and b2\second{b_2} each move both horizontally and vertically. [c1c2]B\begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B and [xy]\target{\begin{bmatrix} x \\ y \end{bmatrix}} are two different step recipes for the exact same arrow.

Standard coordinates and basis coordinates are just two different ways of counting steps along their own grids to reach the exact same arrow tip.
2

Finding a vector's coordinates in a basis

When we already have the slanted recipe [c1c2]B\begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B, we just combine the basis vectors to see where we land. Most of the time, though, we are given the target vector in standard coordinates, such as v=[15]\target{v} = \target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}, and need to work backwards to find the weights [c1c2]B\begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B on b1=[21]\first{b_1} = \first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} and b2=[−11]\second{b_2} = \second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}} that reach it.

Finding a vector's coordinates in a basis
x\htmlClass{lc-mute}{x}
y\htmlClass{lc-mute}{y}
b1=[21]\htmlClass{lc-green}{b_1} = \htmlClass{lc-green}{\begin{bmatrix} 2 \\ 1 \end{bmatrix}}
b2=[−11]\htmlClass{lc-pink}{b_2} = \htmlClass{lc-pink}{\begin{bmatrix} -1 \\ 1 \end{bmatrix}}
v=[15]\htmlClass{lc-blue}{v} = \htmlClass{lc-blue}{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}
[v]B=  ?[v]_B = \;?
Text solution
1. Set up a linear combination with unknown weights c1\first{c_1} and c2\second{c_2}Asking for the BB-coordinates of [15]\target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}} is asking how many steps of b1\first{b_1} and b2\second{b_2} add up to [15]\target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}.
c1[21]+c2[−11]=[15]\displaystyle \first{c_1}\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}} = \target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}
2. Add the two row equations to solve for c1\first{c_1} and c2\second{c_2}Reading across the top and bottom rows gives a system of two equations where adding them cancels c2\second{c_2} right away.
2c1−c2=1c1+c2=5⟹3c1=6⟹c1=2,c2=5−2=3\displaystyle \begin{aligned} \first{2c_1} - \second{c_2} &= \target{1} \\ \first{c_1} + \second{c_2} &= \target{5} \end{aligned} \quad\Longrightarrow\quad \first{3c_1} = 6 \quad\Longrightarrow\quad \first{c_1} = 2, \quad \second{c_2} = 5 - 2 = 3

So [v]B=[23]B[v]_B = \begin{bmatrix} \first{2} \\ \second{3} \end{bmatrix}_B, meaning 2\first{2} steps of b1\first{b_1} followed by 3\second{3} steps of b2\second{b_2} land right on [15]\target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}.

When the numbers are harder to solve by hand or we have more dimensions, we can solve that same equation in two standard ways: either by row-reducing the augmented matrix with the basis vectors as columns, or by multiplying by the inverse matrix P−1\op{P^{-1}}:

[2−11115]→row-reduce[102013]or[c1c2]B=P−1[15]⏟since P[v]B=v\displaystyle \begin{gathered} \left[\begin{array}{cc|c} \first{2} & \second{-1} & \target{1} \\ \first{1} & \second{1} & \target{5} \end{array}\right] \xrightarrow{\text{row-reduce}} \left[\begin{array}{cc|c} 1 & 0 & \first{2} \\ 0 & 1 & \second{3} \end{array}\right] \\ \text{or} \\ \underbrace{\begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B = \op{P^{-1}}\target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}}_{\text{since } \op{P}[v]_B = \target{v}} \end{gathered}
3

Why a basis must span and be linearly independent

A set of step vectors only gives every point a single, reliable coordinate address if the vectors form a basis, meaning they both span the space and are linearly independent. Let's see how coordinates break down when either requirement fails.

Why a basis must span and be linearly independent
x\htmlClass{lc-mute}{x}
y\htmlClass{lc-mute}{y}
[21]\htmlClass{lc-green}{\begin{bmatrix} 2 \\ 1 \end{bmatrix}}
[42]\htmlClass{lc-pink}{\begin{bmatrix} 4 \\ 2 \end{bmatrix}}
[15]  ?\htmlClass{lc-blue}{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}\;?

If the step vectors don't span the space, like [21]\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} and [42]\second{\begin{bmatrix} 4 \\ 2 \end{bmatrix}} on the same line, vectors off that line like [15]\target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}} have no coordinates at all. And if the step vectors aren't linearly independent, like b1=[10]\first{b_1} = \first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}}, b2=[01]\second{b_2} = \second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}}, and an extra b3=[11]=b1+b2\third{b_3} = \third{\begin{bmatrix} 1 \\ 1 \end{bmatrix}} = \first{b_1} + \second{b_2}, the same vector [22]\target{\begin{bmatrix} 2 \\ 2 \end{bmatrix}} gets infinitely many coordinate recipes, such as [220]B\begin{bmatrix} \first{2} \\ \second{2} \\ \third{0} \end{bmatrix}_B, [002]B\begin{bmatrix} \first{0} \\ \second{0} \\ \third{2} \end{bmatrix}_B, and [111]B\begin{bmatrix} \first{1} \\ \second{1} \\ \third{1} \end{bmatrix}_B:

2[10]+2[01]+0[11]=0[10]+0[01]+2[11]=1[10]+1[01]+1[11]=[22]\displaystyle \begin{aligned} &\first{2}\first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} + \second{2}\second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} + \third{0}\third{\begin{bmatrix} 1 \\ 1 \end{bmatrix}} \\ ={}& \first{0}\first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} + \second{0}\second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} + \third{2}\third{\begin{bmatrix} 1 \\ 1 \end{bmatrix}} \\ ={}& \first{1}\first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} + \second{1}\second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} + \third{1}\third{\begin{bmatrix} 1 \\ 1 \end{bmatrix}} \\ ={}& \target{\begin{bmatrix} 2 \\ 2 \end{bmatrix}} \end{aligned}

As long as b2\second{b_2} points off the line of b1\first{b_1}, their slanted grid covers the board and v\target{v} has one unique coordinate pair [c1c2]B\begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B, but the instant b2\second{b_2} lines up with b1\first{b_1}, the grid collapses onto a single line and v\target{v} has no coordinates at all.

As long as b2\second{\mathbf{b}_2} points off the line of b1\first{\mathbf{b}_1}, the two vectors make a full grid with one unique pair of coordinates for v\target{\mathbf{v}}, until b2\second{\mathbf{b}_2} lines up with b1\first{\mathbf{b}_1} and the grid collapses onto a single line.
4

Coordinates in a basis for a subspace

Now suppose we have just two basis vectors in 3D space, b1=[121]\first{b_1} = \first{\begin{bmatrix} 1 \\ 2 \\ 1 \end{bmatrix}} and b2=[102]\second{b_2} = \second{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}}, which span a tilted 2D plane through the origin. We want to find the BB-coordinates of a 3D vector v=[567]\target{v} = \target{\begin{bmatrix} 5 \\ 6 \\ 7 \end{bmatrix}} on that plane, and see how the algebra tells us when a vector like w=[568]\target{w} = \target{\begin{bmatrix} 5 \\ 6 \\ 8 \end{bmatrix}} sits off the plane altogether.

Coordinates in a basis for a subspace
[v]B=  ?[v]_B = \;?
seen edge-on

Because the basis matrix [112012]\begin{bmatrix} \first{1} & \second{1} \\ \first{2} & \second{0} \\ \first{1} & \second{2} \end{bmatrix} has 33 rows and 22 columns, it is not square and has no inverse P−1\op{P^{-1}}, so we find the coordinates by solving the system row by row or row-reducing.

Text solution
1. Set up the linear combination with two unknown weightsEven though v=[567]\target{v} = \target{\begin{bmatrix} 5 \\ 6 \\ 7 \end{bmatrix}} has three standard entries, basis BB has only two vectors, so [v]B[v]_B has only two coordinates c1\first{c_1} and c2\second{c_2}.
c1[121]+c2[102]=[567]\displaystyle \first{c_1}\first{\begin{bmatrix} 1 \\ 2 \\ 1 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}} = \target{\begin{bmatrix} 5 \\ 6 \\ 7 \end{bmatrix}}
2. Solve two rows to find c1\first{c_1} and c2\second{c_2}The middle row gives c1\first{c_1} right away, and plugging c1=3\first{c_1} = 3 into the top row gives c2\second{c_2}.
2c1+0c2=6  ⟹  c1=3,3+c2=5  ⟹  c2=2\displaystyle \begin{aligned} \first{2c_1} + \second{0c_2} &= \target{6} \implies \first{c_1} = 3, \\ \first{3} + \second{c_2} &= \target{5} \implies \second{c_2} = 2 \end{aligned}
3. Check the remaining row to confirm the vector lies on the planePlugging c1=3\first{c_1} = 3 and c2=2\second{c_2} = 2 into the bottom row tests whether those steps land at the right zz-height.
1(3)+2(2)\displaystyle \first{1(3)} + \second{2(2)} =3+4\displaystyle = 3 + 4 =7(matches the third entry 7)\displaystyle = 7 \quad\text{(matches the third entry } \target{7}\text{)}

So v\target{v} lies on the plane and its coordinates are [v]B=[32]B[v]_B = \begin{bmatrix} \first{3} \\ \second{2} \end{bmatrix}_B.

When we try the same thing for w=[568]\target{w} = \target{\begin{bmatrix} 5 \\ 6 \\ 8 \end{bmatrix}}, the first two rows still give c1=3\first{c_1} = 3 and c2=2\second{c_2} = 2, but the third row fails:

1(3)+2(2)=7≠8(or [001] when row-reduced)\displaystyle \begin{gathered} \first{1(3)} + \second{2(2)} = 7 \fails{\neq 8} \\ \left(\text{or } \left[\begin{array}{cc|c} 0 & 0 & \fails{1} \end{array}\right] \text{ when row-reduced}\right) \end{gathered}

Never stop after two equations give you c1\first{c_1} and c2\second{c_2} for a 3D vector: always plug them into the third row, because when the third row doesn't match, that tells you a vector like w=[568]\target{w} = \target{\begin{bmatrix} 5 \\ 6 \\ 8 \end{bmatrix}} isn't on the plane at all and has no BB-coordinates.

5

Choosing a basis to make a problem simple

Dropping from three coordinates to two on a plane in 3D is one reason to change bases. Even when we stay in 2D, picking basis vectors lined up with the geometry of a problem, such as reflecting a vector across a mirror line tilted at 30∘30^\circ, turns a mess of sines and cosines into a single sign flip.

Choosing a basis to make a problem simple
x\htmlClass{lc-mute}{x}
y\htmlClass{lc-mute}{y}
30∘30^\circ
v\htmlClass{lc-blue}{v}
reflect v=  ?\text{reflect } \htmlClass{lc-blue}{v} = \;?

With b1\first{b_1} along the 30∘30^\circ mirror line and b2\second{b_2} perpendicular to it, the standard [xy]\target{\begin{bmatrix} x \\ y \end{bmatrix}} coordinates change in complicated ways when we reflect, but in the mirror basis the step along b1\first{b_1} stays the same and the step along b2\second{b_2} just flips sign:

[c1c2]B→reflect across b1[c1−c2]B\displaystyle \begin{bmatrix} \first{c_1} \\ \second{c_2} \end{bmatrix}_B \op{\xrightarrow{\text{reflect across } \first{b_1}}} \begin{bmatrix} \first{c_1} \\ \second{-c_2} \end{bmatrix}_B
In the standard grid, reflecting across a tilted mirror changes both coordinates in messy ways, while in the mirror's own grid it just flips the second sign.

Key concepts

Every coordinate question comes down to which direction you are converting between the basis B={b1,b2}B = \{\first{b_1}, \second{b_2}\} and standard coordinates:

Given [v]B[v]_B, find standard coordinates v\target{v} (combine the steps)

Do not set up a system or invert a matrix, because [v]B[v]_B already gives you the weights. Multiply the basis vectors by those weights and add, which is the same as multiplying by the basis matrix P=[b1b2]\op{P} = \begin{bmatrix} \first{b_1} & \second{b_2} \end{bmatrix}:

For b1=[21],  b2=[−11],  [v]B=[21]B:v=P[v]B=2[21]+1[−11]=[33]\displaystyle \begin{aligned} &\text{For } \first{b_1} = \first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}},\; \second{b_2} = \second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}},\; \\ &[v]_B = \begin{bmatrix} \first{2} \\ \second{1} \end{bmatrix}_B: \\ &\target{v} = \op{P}[v]_B = \first{2}\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} + \second{1}\second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}} = \target{\begin{bmatrix} 3 \\ 3 \end{bmatrix}} \end{aligned}

Given standard coordinates v\target{v}, find [v]B[v]_B (solve for the recipe)

Set up c1b1+c2b2=v\first{c_1 b_1} + \second{c_2 b_2} = \target{v} and solve by row-reducing [b1b2v]\left[\begin{array}{cc|c} \first{b_1} & \second{b_2} & \target{v} \end{array}\right], or, when P\op{P} is square and P−1\op{P^{-1}} is already known, multiply by P−1\op{P^{-1}}:

For v=[15]:c1[21]+c2[−11]=[15]or[v]B=P−1v=[23]B\displaystyle \begin{aligned} &\text{For } \target{v} = \target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}}: \\ &\first{c_1}\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} -1 \\ 1 \end{bmatrix}} = \target{\begin{bmatrix} 1 \\ 5 \end{bmatrix}} \\ &\text{or}\quad [v]_B = \op{P^{-1}}\target{v} = \begin{bmatrix} \first{2} \\ \second{3} \end{bmatrix}_B \end{aligned}

When not to use P−1\op{P^{-1}} (basis for a subspace)

If you have two basis vectors in 3D, like b1=[121]\first{b_1} = \first{\begin{bmatrix} 1 \\ 2 \\ 1 \end{bmatrix}} and b2=[102]\second{b_2} = \second{\begin{bmatrix} 1 \\ 0 \\ 2 \end{bmatrix}}, the 3×23 \times 2 basis matrix has no inverse. Solve two rows for c1\first{c_1} and c2\second{c_2} and check the third row: if 1(3)+2(2)=7\first{1(3)} + \second{2(2)} = 7 matches v=[567]\target{v} = \target{\begin{bmatrix} 5 \\ 6 \\ 7 \end{bmatrix}}, then [v]B=[32]B[v]_B = \begin{bmatrix} \first{3} \\ \second{2} \end{bmatrix}_B has just two coordinates; if 1(3)+2(2)=7≠8\first{1(3)} + \second{2(2)} = 7 \fails{\neq 8} for w=[568]\target{w} = \target{\begin{bmatrix} 5 \\ 6 \\ 8 \end{bmatrix}}, the vector lies off the plane and has no BB-coordinates.

What makes BB a valid basis

Spanning guarantees the system has at least one solution (every vector gets coordinates), and linear independence guarantees no step vector is a combination of the others (every vector's coordinates are unique).

6

Practice

1

Let BB be the basis b1=[31]\first{b_1} = \first{\begin{bmatrix} 3 \\ 1 \end{bmatrix}} and b2=[−12]\second{b_2} = \second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}}. Find the standard coordinates of the vector v\target{v} whose coordinates in basis BB are [v]B=[2−3]B[v]_B = \begin{bmatrix} \first{2} \\ \second{-3} \end{bmatrix}_B.

You already have the step recipe [2−3]B\begin{bmatrix} \first{2} \\ \second{-3} \end{bmatrix}_B: do you need to solve a system of equations, or combine b1\first{b_1} and b2\second{b_2} using these weights?

Use the BB-coordinates 2\first{2} and −3\second{-3} directly as the weights on b1\first{b_1} and b2\second{b_2} (setting up c1[31]+c2[−12]=[2−3]\first{c_1}\first{\begin{bmatrix} 3 \\ 1 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}} = \begin{bmatrix} 2 \\ -3 \end{bmatrix} solves the backwards problem and gives [17−117]\fails{\begin{bmatrix} \tfrac{1}{7} \\ -\tfrac{11}{7} \end{bmatrix}}):
v=2[31]−3[−12]\displaystyle \target{v} = \first{2}\first{\begin{bmatrix} 3 \\ 1 \end{bmatrix}} - \second{3}\second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}}
Watch the double negative (−3)(−1)=+3(-3)(-1) = +3 in the top entry (keeping it negative gives 6−3=36 - 3 = \fails{3} and the wrong answer [3−4]\fails{\begin{bmatrix} 3 \\ -4 \end{bmatrix}}):
v=[62]+[3−6]=[9−4]\displaystyle \target{v} = \first{\begin{bmatrix} 6 \\ 2 \end{bmatrix}} + \second{\begin{bmatrix} 3 \\ -6 \end{bmatrix}} = \target{\begin{bmatrix} 9 \\ -4 \end{bmatrix}}
2

Let BB be the basis b1=[13]\first{b_1} = \first{\begin{bmatrix} 1 \\ 3 \end{bmatrix}} and b2=[2−1]\second{b_2} = \second{\begin{bmatrix} 2 \\ -1 \end{bmatrix}}. Find the coordinates [v]B[v]_B of the vector v=[70]\target{v} = \target{\begin{bmatrix} 7 \\ 0 \end{bmatrix}} in basis BB.

Set up c1b1+c2b2=[70]\first{c_1 b_1} + \second{c_2 b_2} = \target{\begin{bmatrix} 7 \\ 0 \end{bmatrix}}; what does the bottom row tell you about c2\second{c_2} in terms of c1\first{c_1}?

Write the combination with unknown weights c1\first{c_1} and c2\second{c_2}:
c1[13]+c2[2−1]=[70]⟹c1+2c2=73c1−c2=0\displaystyle \first{c_1}\first{\begin{bmatrix} 1 \\ 3 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} 2 \\ -1 \end{bmatrix}} = \target{\begin{bmatrix} 7 \\ 0 \end{bmatrix}} \quad\Longrightarrow\quad \begin{aligned} \first{c_1} + \second{2c_2} &= \target{7} \\ \first{3c_1} - \second{c_2} &= \target{0} \end{aligned}
The bottom equation gives c2=3c1\second{c_2} = \first{3c_1} (writing c1=3c2\fails{c_1 = 3c_2} or swapping the order of c1c_1 and c2c_2 at the end gives the wrong answer [31]B\fails{\begin{bmatrix} 3 \\ 1 \end{bmatrix}}_B). Plug c2=3c1\second{c_2} = \first{3c_1} into the top equation:
c1+2(3c1)\displaystyle \first{c_1} + 2(\first{3c_1}) =7  ⟹  7c1\displaystyle = \target{7} \implies \first{7c_1} =7  ⟹  c1\displaystyle = 7 \implies \first{c_1} =1,c2\displaystyle = 1, \quad \second{c_2} =3(1)\displaystyle = 3(1) =3\displaystyle = 3
[v]B=[13]B\displaystyle [v]_B = \begin{bmatrix} \first{1} \\ \second{3} \end{bmatrix}_B
3

A basis B={b1,b2,b3}B = \{\first{b_1}, \second{b_2}, \third{b_3}\} in 3D has basis matrix P=[b1b2b3]\op{P} = \begin{bmatrix} \first{b_1} & \second{b_2} & \third{b_3} \end{bmatrix} whose inverse is P−1=[10−12100−11]\op{P^{-1}} = \begin{bmatrix} 1 & 0 & -1 \\ 2 & 1 & 0 \\ 0 & -1 & 1 \end{bmatrix}. Find the coordinates [v]B[v]_B of the standard vector v=[3−21]\target{v} = \target{\begin{bmatrix} 3 \\ -2 \\ 1 \end{bmatrix}}, and state whether converting from standard coordinates to BB uses P\op{P} or P−1\op{P^{-1}}.

Since P[v]B=v\op{P}[v]_B = \target{v} converts from basis BB to standard coordinates, which matrix solves for [v]B[v]_B when starting from the standard vector v\target{v}?

Converting from standard coordinates to BB uses P−1\op{P^{-1}}:
[v]B=P−1v=[10−12100−11][3−21]\displaystyle [v]_B = \op{P^{-1}}\target{v} = \begin{bmatrix} 1 & 0 & -1 \\ 2 & 1 & 0 \\ 0 & -1 & 1 \end{bmatrix}\target{\begin{bmatrix} 3 \\ -2 \\ 1 \end{bmatrix}}
In the bottom row, be careful with (−1)(−2)=+2(-1)(-2) = +2 (treating it as −2-2 gives −2+1=−1-2 + 1 = \fails{-1} and the wrong answer [24−1]B\fails{\begin{bmatrix} 2 \\ 4 \\ -1 \end{bmatrix}}_B):
[v]B=[1(3)+0(−2)+(−1)(1)2(3)+1(−2)+0(1)0(3)+(−1)(−2)+1(1)]B=[243]B\displaystyle [v]_B = \begin{bmatrix} 1(\target{3}) + 0(\target{-2}) + (-1)(\target{1}) \\ 2(\target{3}) + 1(\target{-2}) + 0(\target{1}) \\ 0(\target{3}) + (-1)(\target{-2}) + 1(\target{1}) \end{bmatrix}_B = \begin{bmatrix} \first{2} \\ \second{4} \\ \third{3} \end{bmatrix}_B
4

A tilted plane through the origin in 3D has basis BB with b1=[112]\first{b_1} = \first{\begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix}} and b2=[02−1]\second{b_2} = \second{\begin{bmatrix} 0 \\ 2 \\ -1 \end{bmatrix}}. Find the BB-coordinates [v]B[v]_B of the vector v=[3−18]\target{v} = \target{\begin{bmatrix} 3 \\ -1 \\ 8 \end{bmatrix}}.

Set up c1b1+c2b2=v\first{c_1 b_1} + \second{c_2 b_2} = \target{v}: how many coordinates will [v]B[v]_B have, and which row gives one of the weights right away?

Set up the combination for the two basis vectors:
c1[112]+c2[02−1]=[3−18]\displaystyle \first{c_1}\first{\begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} 0 \\ 2 \\ -1 \end{bmatrix}} = \target{\begin{bmatrix} 3 \\ -1 \\ 8 \end{bmatrix}}
The top row gives c1=3\first{c_1} = 3. Plug c1=3\first{c_1} = 3 into the middle row c1+2c2=−1\first{c_1} + \second{2c_2} = \target{-1} and subtract 33 from both sides (adding 33 gives 2c2=2  ⟹  c2=1\second{2c_2} = 2 \implies \second{c_2} = \fails{1} and the wrong answer [31]B\fails{\begin{bmatrix} 3 \\ 1 \end{bmatrix}}_B, which fails the bottom row check 2(3)−1(1)=5≠8\first{2(3)} - \second{1(1)} = 5 \fails{\neq 8}):
3+2c2\displaystyle \first{3} + \second{2c_2} =−1  ⟹  2c2\displaystyle = \target{-1} \implies \second{2c_2} =−4  ⟹  c2\displaystyle = -4 \implies \second{c_2} =−2\displaystyle = -2
Check the bottom row: 2(3)−1(−2)=6+2=8\first{2(3)} - \second{1(-2)} = 6 + 2 = 8, which agrees:
[v]B=[3−2]B\displaystyle [v]_B = \begin{bmatrix} \first{3} \\ \second{-2} \end{bmatrix}_B
5

Using the same plane basis b1=[112]\first{b_1} = \first{\begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix}} and b2=[02−1]\second{b_2} = \second{\begin{bmatrix} 0 \\ 2 \\ -1 \end{bmatrix}} in 3D, determine whether the vector w=[3−15]\target{w} = \target{\begin{bmatrix} 3 \\ -1 \\ 5 \end{bmatrix}} has coordinates in basis BB. If so, find [w]B[w]_B; if not, explain what the system tells you geometrically.

Once the first two rows give c1\first{c_1} and c2\second{c_2}, check what happens when you plug those weights into the third row.

Setting up c1[112]+c2[02−1]=[3−15]\first{c_1}\first{\begin{bmatrix} 1 \\ 1 \\ 2 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} 0 \\ 2 \\ -1 \end{bmatrix}} = \target{\begin{bmatrix} 3 \\ -1 \\ 5 \end{bmatrix}}, the first two rows again give c1=3\first{c_1} = 3 and c2=−2\second{c_2} = -2. Do not stop there (stopping after two rows gives the wrong answer [3−2]B\fails{\begin{bmatrix} 3 \\ -2 \end{bmatrix}}_B); plugging c1=3\first{c_1} = 3 and c2=−2\second{c_2} = -2 into the third row gives:
2(3)−1(−2)\displaystyle \first{2(3)} - \second{1(-2)} =8≠5\displaystyle = 8 \fails{\neq 5}
Because the system has no solution, w\target{w} does not lie on the plane spanned by b1\first{b_1} and b2\second{b_2} and has no coordinates in basis BB.
6

Suppose someone uses the three vectors b1=[10]\first{b_1} = \first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}}, b2=[01]\second{b_2} = \second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}}, and b3=[2−1]\third{b_3} = \third{\begin{bmatrix} 2 \\ -1 \end{bmatrix}} as step vectors on a 2D board. Find two different coordinate recipes [c1c2c3]B\begin{bmatrix} \first{c_1} \\ \second{c_2} \\ \third{c_3} \end{bmatrix}_B that both land on v=[4−2]\target{v} = \target{\begin{bmatrix} 4 \\ -2 \end{bmatrix}}, and name the basis requirement that this set of vectors breaks.

Reach [4−2]\target{\begin{bmatrix} 4 \\ -2 \end{bmatrix}} once using only b1\first{b_1} and b2\second{b_2} (with c3=0\third{c_3} = 0), and a second time using only b3\third{b_3} (with c1=c2=0\first{c_1} = \second{c_2} = 0).

Using 4\first{4} steps of b1\first{b_1}, −2\second{-2} steps of b2\second{b_2}, and 0\third{0} steps of b3\third{b_3} (or 00 of b1\first{b_1} and b2\second{b_2} and 2\third{2} steps of b3\third{b_3}) lands on the same vector:
4[10]−2[01]+0[2−1]=[4−2]and0[10]+0[01]+2[2−1]=[4−2]\displaystyle \begin{aligned} \first{4}\first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} - \second{2}\second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} + \third{0}\third{\begin{bmatrix} 2 \\ -1 \end{bmatrix}} &= \target{\begin{bmatrix} 4 \\ -2 \end{bmatrix}} \\ \text{and}\quad \first{0}\first{\begin{bmatrix} 1 \\ 0 \end{bmatrix}} + \second{0}\second{\begin{bmatrix} 0 \\ 1 \end{bmatrix}} + \third{2}\third{\begin{bmatrix} 2 \\ -1 \end{bmatrix}} &= \target{\begin{bmatrix} 4 \\ -2 \end{bmatrix}} \end{aligned}
Because b3=2b1−b2\third{b_3} = \first{2b_1} - \second{b_2} is already a combination of b1\first{b_1} and b2\second{b_2}, the set breaks linear independence (not spanning, since the vectors still reach every point on the 2D board), giving multiple coordinate recipes such as [4−20]B\begin{bmatrix} \first{4} \\ \second{-2} \\ \third{0} \end{bmatrix}_B and [002]B\begin{bmatrix} \first{0} \\ \second{0} \\ \third{2} \end{bmatrix}_B.
7

A mirror line through the origin points along b1=[21]\first{b_1} = \first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}}, and b2=[−12]\second{b_2} = \second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}} is perpendicular to the mirror line. Using the basis B={b1,b2}B = \{\first{b_1}, \second{b_2}\}, find the BB-coordinates of v=[18]\target{v} = \target{\begin{bmatrix} 1 \\ 8 \end{bmatrix}}, reflect v\target{v} across the mirror line in basis BB, and convert the reflected vector back into standard coordinates.

Break the problem into three steps: solve c1b1+c2b2=[18]\first{c_1 b_1} + \second{c_2 b_2} = \target{\begin{bmatrix} 1 \\ 8 \end{bmatrix}} for [v]B[v]_B, flip the sign of the coordinate perpendicular to the mirror, and combine b1\first{b_1} and b2\second{b_2} with the reflected weights.

First find [v]B[v]_B by solving c1[21]+c2[−12]=[18]\first{c_1}\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} + \second{c_2}\second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}} = \target{\begin{bmatrix} 1 \\ 8 \end{bmatrix}}:
2c1−c2=1c1+2c2=8⟹c1=2,  c2=3⟹[v]B=[23]B\displaystyle \begin{aligned} \first{2c_1} - \second{c_2} &= \target{1} \\ \first{c_1} + \second{2c_2} &= \target{8} \end{aligned} \quad\Longrightarrow\quad \first{c_1} = 2,\; \second{c_2} = 3 \quad\Longrightarrow\quad [v]_B = \begin{bmatrix} \first{2} \\ \second{3} \end{bmatrix}_B
Reflecting across the mirror line along b1\first{b_1} flips the sign of the second coordinate c2\second{c_2} perpendicular to the mirror to give [2−3]B\begin{bmatrix} \first{2} \\ \second{-3} \end{bmatrix}_B (flipping c1\first{c_1} instead gives [−23]B\fails{\begin{bmatrix} -2 \\ 3 \end{bmatrix}}_B, which reflects across b2\second{b_2} and leads to the wrong vector [−74]\fails{\begin{bmatrix} -7 \\ 4 \end{bmatrix}}). Finally, combine b1\first{b_1} and b2\second{b_2} to convert back to standard coordinates:
2[21]−3[−12]=[42]+[3−6]=[7−4]\displaystyle \first{2}\first{\begin{bmatrix} 2 \\ 1 \end{bmatrix}} - \second{3}\second{\begin{bmatrix} -1 \\ 2 \end{bmatrix}} = \first{\begin{bmatrix} 4 \\ 2 \end{bmatrix}} + \second{\begin{bmatrix} 3 \\ -6 \end{bmatrix}} = \target{\begin{bmatrix} 7 \\ -4 \end{bmatrix}}