Multivariable-Calculus · Unit 4 · Video 3 · Interactive Practice

Three Planes, One Point: Why AX = B Usually Has Exactly One Solution, and When It Does Not

IKey Formulas

FormulaNameWhat you need
AX=BAX = BMatrix form of a 3×33 \times 3 systemCoefficients in AA, with a 00 for every missing variable
X=A1BX = A^{-1}BSolution of the systemA1A^{-1} exists — and the inverse multiplies on the left
P1P2\mathcal{P}_1 \cap \mathcal{P}_2 is a line along N1×N2\mathbf{N}_1 \times \mathbf{N}_2Line of the first two planesTwo non-parallel normals
dist(O,P)=dN\operatorname{dist}(O, \mathcal{P}) = \dfrac{|d|}{|\mathbf{N}|}Distance from the origin to ax+by+cz=dax + by + cz = dN=a,b,c\mathbf{N} = \langle a, b, c \rangle

Key Insight: Exactly two solutions is impossible — if PP and QQ both satisfy every equation, so does every point of the line through them. The solution set is empty, a single point, a line, or a plane, and nothing else.

IITwo Planes Fix a Line, the Third Picks a Point

Every point of P1P2\mathcal{P}_1 \cap \mathcal{P}_2 satisfies the first two equations; the third keeps exactly one of them.

💡 The picture assumes P1\mathcal{P}_1 and P2\mathcal{P}_2 really cross. Were they parallel but distinct there would be no line to start from and no solution whatever P3\mathcal{P}_3 does; were they the same plane the line would widen into that plane, and three copies of one equation make the solution set that whole plane.

IIIWhen the Third Plane Misses the Line

Tilt or shift P3\mathcal{P}_3 and the single point becomes no solution — or the whole line.

💡 The coefficient of tt in that substitution is N3(N1×N2)=detA\mathbf{N}_3 \cdot (\mathbf{N}_1 \times \mathbf{N}_2) = \det A, here 1+k1 + k: that one number decides whether the line pierces P3\mathcal{P}_3 at all, and when it vanishes the constant dd settles which degenerate case you are in — the whole line at d=3d = 3, nothing otherwise. Reading it off AA is the next video.

IVWhy the Inverse Goes on the Left

Left-multiplying AX=BAX = B by A1A^{-1} collapses the left side to XX; the reversed product does not exist.

Step 1 — Three equations, one matrix equation
(101110123)(xyz)=(123)\begin{pmatrix} 1 & 0 & 1 \\ 1 & 1 & 0 \\ 1 & 2 & 3 \end{pmatrix}\begin{pmatrix} x \\ y \\ z \end{pmatrix} = \begin{pmatrix} 1 \\ 2 \\ 3 \end{pmatrix}
Row 1 of AA dotted with XX reproduces x+z=1x + z = 1; the 00 holds the place of the missing yy.

VQuiz Questions

Problem 1 · Reading Off the Coefficient Matrix

Given: the system x+z=1x + z = 1,  x+y=2\ x + y = 2,  x+2y+3z=3\ x + 2y + 3z = 3 written as AX=BAX = Bfind the second row of AA.

✅ Correct! x+y=2x + y = 2 carries no zz, so a 00 holds that column.
❌ That is the first row. x+z=1x + z = 1 gives (101)\begin{pmatrix} 1 & 0 & 1 \end{pmatrix}; the second equation is x+y=2x + y = 2.
❌ A row needs one entry per variable. A missing variable contributes 00, not a missing entry — otherwise AA would not be 3×33 \times 3 and AXAX would not exist.
❌ That is the third row. x+2y+3z=3x + 2y + 3z = 3 gives (123)\begin{pmatrix} 1 & 2 & 3 \end{pmatrix}.
❌ Not quite. Read the coefficients of xx, yy, zz — in that order — out of x+y=2x + y = 2.
Show solution

Write every equation with all three variables showing:

x+0y+  z=1x+  y+0z=2x+2y+3z=3\begin{aligned} x + 0y + \ \ z &= 1 \\ x + \ \ y + 0z &= 2 \\ x + 2y + 3z &= 3 \end{aligned}

The coefficients, row by row, are the rows of AA:

A=(101110123),B=(123)A = \begin{pmatrix} 1 & 0 & 1 \\ 1 & 1 & 0 \\ 1 & 2 & 3 \end{pmatrix}, \qquad B = \begin{pmatrix} 1 \\ 2 \\ 3 \end{pmatrix}

The second row is (110)\begin{pmatrix} 1 & 1 & 0 \end{pmatrix}, and row 2 of AA dotted with XX returns x+y=2x + y = 2.

Problem 2 · Which Side Does the Inverse Go On?

Given: AA is an invertible 3×33 \times 3 matrix and BB is the 3×13 \times 1 column of constants in AX=BAX = Bfind the expression that gives XX.

✅ Correct! Left-multiplying by A1A^{-1} turns AXAX into IX=XIX = X, and (3×3)(3×1)(3 \times 3)(3 \times 1) returns the column XX.
❌ The sizes forbid that order. BA1BA^{-1} is (3×1)(3×3)(3 \times 1)(3 \times 3): the inner sizes 11 and 33 disagree, so the product does not exist.
ABAB is a legal product, but the wrong one. Multiplying by AA again does not undo AA — only A1A=IA^{-1}A = I does.
BB has no inverse. It is 3×13 \times 1, not square, so B1B^{-1} is not defined.
❌ Not quite. Undo AA by left-multiplying both sides of AX=BAX = B by A1A^{-1}.
Show solution

Left-multiply both sides of AX=BAX = B by A1A^{-1} and use associativity:

A1(AX)=A1B(A1A)X=A1BIX=X=A1BA^{-1}(AX) = A^{-1}B \quad\Longrightarrow\quad (A^{-1}A)X = A^{-1}B \quad\Longrightarrow\quad IX = X = A^{-1}B

The sizes decide the side. A1BA^{-1}B is (3×3)(3×1)(3 \times 3)(3 \times 1): inner sizes 33 and 33 agree and the product is the 3×13 \times 1 column XX. Reversed, BA1BA^{-1} is (3×1)(3×3)(3 \times 1)(3 \times 3): the inner sizes are 11 and 33, so no product exists — matrix multiplication is not commutative, and here the other order is not even defined.

Problem 3 · Parallel, or Contained?

Given: x+z=1x + z = 1 and x+y=2x + y = 2 meet in the line (1t, 1+t, t)(1 - t,\ 1 + t,\ t), and the third equation is x+2yz=dx + 2y - z = dfind what the third equation becomes on that line, and then the value of dd that makes the system have infinitely many solutions.

Substituting the line into x+2yzx + 2y - z gives

So the system has infinitely many solutions when

✅ Correct! The tt terms cancel, so the third equation reads 3=d3 = d: at d=3d = 3 it is 0=00 = 0 and the whole line solves the system; at any other dd it is a contradiction and nothing does.
❌ Check the substitution. (1t)+2(1+t)t(1-t) + 2(1+t) - t: the tt terms are t+2tt-t + 2t - t.
❌ Compare the two sides. The left side is the constant 33 for every point of the line, so the equation can only hold if dd is that same constant.
Show solution

Put x=1tx = 1 - t, y=1+ty = 1 + t, z=tz = t into the left side of the third equation:

(1t)+2(1+t)t=1t+2+2tt=3(1 - t) + 2(1 + t) - t = 1 - t + 2 + 2t - t = 3

Every tt cancels, so the third equation reads 3=d3 = d at every point of the line — the line is parallel to P3\mathcal{P}_3, which the normals confirm: 1,2,11,1,1=1+21=0\langle 1, 2, -1 \rangle \cdot \langle -1, 1, 1 \rangle = -1 + 2 - 1 = 0.

  • d=3d = 3: the third equation says 3=33 = 3, that is 0=00 = 0. It adds nothing, every point of the line solves the system, and there are infinitely many solutions — the line lies inside P3\mathcal{P}_3.
  • d3d \ne 3, say d=5d = 5: the third equation says 3=53 = 5, a contradiction of the 1=21 = 2 kind. The line runs alongside P3\mathcal{P}_3 and never meets it, so there is no solution.

Note what cannot happen: two solutions. Two distinct solutions would force every point of the line through them to be a solution as well.

Problem 4 · What the Constant Measures

Given: the plane x+y+z=4x + y + z = 4find its distance from the origin.

✅ Correct! 43=4332.31\frac{4}{\sqrt{3}} = \frac{4\sqrt{3}}{3} \approx 2.31, the constant measured in units of N|\mathbf{N}|.
❌ The constant is the distance only for a unit normal. Here N=1,1,1\mathbf{N} = \langle 1, 1, 1 \rangle has length 3\sqrt{3}, not 11.
❌ That divides by N2=3|\mathbf{N}|^2 = 3. The distance divides by the length N=3|\mathbf{N}| = \sqrt{3}.
❌ Multiplied instead of divided. A longer normal means the same constant pushes the plane less far from the origin.
❌ Not quite. Divide the constant by the length of the normal vector 1,1,1\langle 1, 1, 1 \rangle.
Show solution

The normal vector is N=1,1,1\mathbf{N} = \langle 1, 1, 1 \rangle, so N=1+1+1=3|\mathbf{N}| = \sqrt{1 + 1 + 1} = \sqrt{3}, and the constant is d=4d = 4:

dist(O,P)=dN=43=4332.31\operatorname{dist}(O, \mathcal{P}) = \frac{|d|}{|\mathbf{N}|} = \frac{4}{\sqrt{3}} = \frac{4\sqrt{3}}{3} \approx 2.31

Changing the constant slides the plane to a parallel one, since the normal does not change: x+y+z=0x + y + z = 0 passes through the origin, and x+y+z=4x + y + z = 4 has been pushed 44 units out along N\mathbf{N} — but those units are N=3|\mathbf{N}| = \sqrt{3} long each, which is why the distance is 4/34/\sqrt{3} rather than 44.

Solved: 0 / 4