Further matrix algebra
FP3 Topic 6, spec point by spec point (Further matrix algebra). Each spec point has the method, worked examples written the way the mark scheme wants them, how the marks go and the mistakes that cost the most marks.
Spec6.1Linear transformations in two and three dimensions
Represent linear transformations of 2D and 3D column vectors by matrices.
The idea
As in FP1, the columns are the images of the unit vectors. In 3D:
\[\mathbf{M} = \begin{pmatrix} \uparrow & \uparrow & \uparrow \\ \mathbf{M}\mathbf{i} & \mathbf{M}\mathbf{j} & \mathbf{M}\mathbf{k} \\ \downarrow & \downarrow & \downarrow \end{pmatrix}\]
Examples: reflection in the plane \(y = 0\) is \(\begin{pmatrix} 1 & 0 & 0 \\ 0 & -1 & 0 \\ 0 & 0 & 1 \end{pmatrix}\); rotation by \(\theta\) about the \(z\)-axis is \(\begin{pmatrix} \cos\theta & -\sin\theta & 0 \\ \sin\theta & \cos\theta & 0 \\ 0 & 0 & 1 \end{pmatrix}\).
\(|\det\mathbf{M}|\) is the volume scale factor.
Images of lines
Multiply the general point of the line by \(\mathbf{M}\). The image is another line (unless \(\mathbf{M}\) is singular); write it in vector or Cartesian form.
image of a line
\(\mathbf{M} = \begin{pmatrix} 2 & 0 & 1 \\ 0 & 1 & -1 \\ 1 & 1 & 0 \end{pmatrix}\). Find the image of the line \((1 + t, 2t - 1, -3t)\).
\(\mathbf{M}\begin{pmatrix} 1 + t \\ 2t - 1 \\ -3t \end{pmatrix} = \begin{pmatrix} 2 - t \\ -1 + 5t \\ 3t \end{pmatrix}\).
Cartesian: \(\dfrac{x - 2}{-1} = \dfrac{y + 1}{5} = \dfrac{z}{3}\).
- M1 for multiplying the parametric point by the matrix, A1 for the image point, M1 for eliminating \(t\), A1 for a Cartesian form (WFM03 Jan 2026 Q6c).
- Volumes: set \(\frac16|\text{triple product}| \times |\det\mathbf{M}|\) equal to the given volume and solve; there may be two values (WFM03 Oct 2021 Q4c).
Often asked as: image of a point, line or plane; find a constant from a volume.
Writing images of \(\mathbf{i}\), \(\mathbf{j}\), \(\mathbf{k}\) as rows.
Spec6.2Combining transformations
Combine transformations using matrix products, knowing \(\mathbf{AB}\) means \(\mathbf{B}\) first, then \(\mathbf{A}\).
The idea
Exactly as in FP1 6.3, in 3D: "\(\mathbf{S}\) followed by \(\mathbf{T}\)" is \(\mathbf{TS}\). Matrix multiplication is associative but not commutative, so the order is crucial.
order
\(\mathbf{S}\) is a reflection in the plane \(z = 0\) and \(\mathbf{T}\) is a rotation of \(90°\) about the \(z\)-axis. The matrix for "\(\mathbf{S}\) then \(\mathbf{T}\)" is
\(\mathbf{TS} = \begin{pmatrix} 0 & -1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 1 \end{pmatrix}\begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & -1 \end{pmatrix} = \begin{pmatrix} 0 & -1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & -1 \end{pmatrix}\).
- M1 for the product in the right order, A1 for six correct elements, A1 for the full matrix (WFM03 Jun 2026 Q5b).
Often asked as: one part of a longer matrix question.
\(\mathbf{ST}\) for "S then T".
Spec6.3The transpose
Find the transpose of a matrix and use \((\mathbf{AB})^{\mathsf{T}} = \mathbf{B}^{\mathsf{T}}\mathbf{A}^{\mathsf{T}}\).
The idea
\(\mathbf{A}^{\mathsf{T}}\) swaps rows and columns: row 1 becomes column 1, and so on.
- \((\mathbf{AB})^{\mathsf{T}} = \mathbf{B}^{\mathsf{T}}\mathbf{A}^{\mathsf{T}}\) (order reverses, like inverses).
- Symmetric: \(\mathbf{A}^{\mathsf{T}} = \mathbf{A}\) (mirror image across the leading diagonal).
- Orthogonal: \(\mathbf{P}^{\mathsf{T}}\mathbf{P} = \mathbf{I}\), so \(\mathbf{P}^{-1} = \mathbf{P}^{\mathsf{T}}\). Its columns are perpendicular unit vectors. This is what makes 6.8 work.
transpose
\(\mathbf{A} = \begin{pmatrix} 1 & 2 & 0 \\ 3 & -1 & 4 \end{pmatrix}\) has \(\mathbf{A}^{\mathsf{T}} = \begin{pmatrix} 1 & 3 \\ 2 & -1 \\ 0 & 4 \end{pmatrix}\).
- B1 for the transpose written out, M1 for \(\mathbf{B}^{\mathsf{T}}\mathbf{A}^{\mathsf{T}}\) multiplied in that order with at least 5 entries correct (WFM03 Jun 2017 Q2).
Often asked as: a short "show that \((\mathbf{AB})^{\mathsf{T}} = \mathbf{B}^{\mathsf{T}}\mathbf{A}^{\mathsf{T}}\)" for given matrices; inside diagonalisation.
Transposing only the square part of a non-square matrix.
Spec6.43 × 3 determinants
Evaluate \(3 \times 3\) determinants, and know singular and non-singular matrices.
Method: expand along a row
\[\begin{vmatrix} a & b & c \\ d & e & f \\ g & h & i \end{vmatrix} = a\begin{vmatrix} e & f \\ h & i \end{vmatrix} - b\begin{vmatrix} d & f \\ g & i \end{vmatrix} + c\begin{vmatrix} d & e \\ g & h \end{vmatrix}\]
Signs alternate \(+ - +\). You can expand along any row or column (with the sign pattern \(\begin{smallmatrix} + & - & + \\ - & + & - \\ + & - & + \end{smallmatrix}\)); choose one with zeros.
\(\det = 0\): singular (no inverse; the transformation squashes space flat).
when is it singular?
Find \(k\) such that \(\mathbf{A} = \begin{pmatrix} 2 & k & 2 \\ 2 & 2 & k \\ 1 & 2 & 2 \end{pmatrix}\) is singular.
\(\det\mathbf{A} = 2(4 - 2k) - k(4 - k) + 2(4 - 2) = k^2 - 8k + 12 = (k - 2)(k - 6)\).
Singular when \(k = 2\) or \(k = 6\).
- M1 for a correct expansion along some row or column, A1 for the determinant simplified, then solving (WFM03 Jan 2021 Q3a, WFM03 Jun 2026 Q5a).
Often asked as: singular values of \(k\); the first step of an inverse or eigenvalue question.
Forgetting the minus sign on the middle term.
Spec6.53 × 3 inverses
Find the inverse of a \(3 \times 3\) matrix and use \((\mathbf{AB})^{-1} = \mathbf{B}^{-1}\mathbf{A}^{-1}\).
Method
- Determinant (6.4). If it is 0, there is no inverse.
- Minors: for each entry, the \(2 \times 2\) determinant left when you delete its row and column.
- Cofactors: apply the sign pattern \(\begin{smallmatrix} + & - & + \\ - & + & - \\ + & - & + \end{smallmatrix}\).
- Transpose the cofactor matrix (this is the adjugate).
- Divide by the determinant.
Check one entry of \(\mathbf{AA}^{-1}\) if you have time.
a 3 × 3 inverse
Find the inverse of \(\mathbf{B} = \begin{pmatrix} 1 & -1 & 1 \\ 1 & 1 & 1 \\ 1 & 2 & 4 \end{pmatrix}\).
\(\det\mathbf{B} = 1(2) + 1(3) + 1(1) = 6\).
Minors: \(\begin{pmatrix} 2 & 3 & 1 \\ -6 & 3 & 3 \\ -2 & 0 & 2 \end{pmatrix}\). Cofactors: \(\begin{pmatrix} 2 & -3 & 1 \\ 6 & 3 & -3 \\ -2 & 0 & 2 \end{pmatrix}\).
\(\mathbf{B}^{-1} = \dfrac16\begin{pmatrix} 2 & 6 & -2 \\ -3 & 3 & 0 \\ 1 & -3 & 2 \end{pmatrix}\).
- M1 for the matrix of minors (about six correct), M1 for cofactor signs, A1, M1 for transposing and dividing by the determinant, A1ft (WFM03 Jun 2021 Q3b, WFM03A Jan 2026 Q5a).
- If \(\mathbf{TU} = \mathbf{I}\) with unknowns, multiply and solve the resulting equations (WFM03 Jan 2024 Q2a).
Often asked as: inverse in terms of \(k\); then use it to reverse a transformation (6.6).
Forgetting to transpose.
Sign pattern applied to the wrong entries.
Spec6.6Inverse transformations
Find the inverse (when it exists) of a transformation or a combination of transformations.
The idea
To find what maps onto a given line or plane, apply \(\mathbf{M}^{-1}\) to it:
- Line \(\mathbf{r} = \mathbf{a} + t\mathbf{d}\): the original line is \(\mathbf{M}^{-1}\mathbf{a} + t\,\mathbf{M}^{-1}\mathbf{d}\).
- Plane: write it in parametric form, apply \(\mathbf{M}^{-1}\) to the point and both directions, then convert to Cartesian.
Alternatively, substitute \(\mathbf{M}\mathbf{x}\) (the image of a general point) into the image's equation.
the line that maps onto a given line
\(\mathbf{B}\) (from 6.5) maps \(l_1\) onto \(l_2\): \(\left(\mathbf{r} - (12\mathbf{i} + 4\mathbf{j} + 6\mathbf{k})\right) \times (-6\mathbf{i} + 2\mathbf{j} + 3\mathbf{k}) = \mathbf{0}\). Find \(l_1\).
\(\mathbf{B}^{-1}(12, 4, 6) = \frac16(36, -24, 12) = (6, -4, 2)\). \(\mathbf{B}^{-1}(-6, 2, 3) = \frac16(-6, 24, -6) = (-1, 4, -1)\).
\(l_1\): \(\mathbf{r} = (6\mathbf{i} - 4\mathbf{j} + 2\mathbf{k}) + s(-\mathbf{i} + 4\mathbf{j} - \mathbf{k})\).
- B1ft for the inverse, M1 for multiplying the point and direction (or the parametric form) by it, A1 for the parametric form, A1 for the equation (WFM03 Oct 2020 Q6b, WFM03 Jan 2024 Q2b).
- Planes: M1 for a two-parameter form of the image plane, ddM1 for applying the inverse, then eliminating the parameters (WFM03A Jan 2026 Q5b).
Often asked as: after finding an inverse in terms of \(k\), set \(k\) and use it.
Multiplying by \(\mathbf{M}\) instead of \(\mathbf{M}^{-1}\).
Spec6.7Eigenvalues and eigenvectors
Find eigenvalues and eigenvectors of \(2 \times 2\) and \(3 \times 3\) matrices, including normalised eigenvectors.
The idea
An eigenvector of \(\mathbf{M}\) is a non-zero vector whose direction doesn't change under the transformation: it only gets scaled by the eigenvalue \(\lambda\).
\[\mathbf{M}\mathbf{x} = \lambda\mathbf{x}\]
The line through the origin in that direction is an invariant line.
Method
- Eigenvalues: solve the characteristic equation \(\det(\mathbf{M} - \lambda\mathbf{I}) = 0\).
- Eigenvectors: for each \(\lambda\), solve \((\mathbf{M} - \lambda\mathbf{I})\mathbf{x} = \mathbf{0}\). Two equations are enough in 3D; the third is a check. There are infinitely many answers (any multiple), so pick a simple one.
- Normalised: divide by the magnitude to get length 1.
Checks: the eigenvalues add up to the trace (sum of the leading diagonal) and multiply to the determinant.
2 × 2
Find the eigenvalues and eigenvectors of \(\begin{pmatrix} 4 & 1 \\ 2 & 3 \end{pmatrix}\).
\((4 - \lambda)(3 - \lambda) - 2 = \lambda^2 - 7\lambda + 10 = 0\), so \(\lambda = 2\) or \(5\).
\(\lambda = 5\): \(-x + y = 0\), eigenvector \(\begin{pmatrix} 1 \\ 1 \end{pmatrix}\). \(\lambda = 2\): \(2x + y = 0\), eigenvector \(\begin{pmatrix} 1 \\ -2 \end{pmatrix}\).
3 × 3 with a repeated eigenvalue
\(\mathbf{M} = \begin{pmatrix} 2 & 0 & 3 \\ 0 & -4 & -3 \\ 0 & -4 & 0 \end{pmatrix}\). Find the eigenvalues and an eigenvector for the smaller one.
Expand down the first column: \((2 - \lambda)\left[(-4 - \lambda)(-\lambda) - 12\right] = (2 - \lambda)(\lambda^2 + 4\lambda - 12) = (2 - \lambda)(\lambda + 6)(\lambda - 2)\).
Eigenvalues 2 (repeated) and \(-6\).
\(\lambda = -6\): \(8x + 3z = 0\) and \(2y - 3z = 0\). With \(z = 8\): \(x = -3\), \(y = 12\). Eigenvector \((-3, 12, 8)\).
- M1 for the characteristic equation from \(\det(\mathbf{M} - \lambda\mathbf{I})\), dM1 for solving, A1 for the values, M1 for using \(\mathbf{M}\mathbf{x} = \lambda\mathbf{x}\), A1 for an eigenvector (WFM03 May 2024 Q2a).
- Setting a component to 0 without justification is not a correct method for finding an eigenvector (WFM03 Jun 2022 Q3b).
- Third eigenvalue from the trace or determinant: B1 (WFM03A Jun 2026 Q3c).
Often asked as: show \(\lambda\) is an eigenvalue; find the others; find eigenvectors; then diagonalise (6.8).
Sign errors in \(\det(\mathbf{M} - \lambda\mathbf{I})\): subtract \(\lambda\) only on the diagonal.
Giving the zero vector as an eigenvector.
Spec6.8Reducing a symmetric matrix to diagonal form
Find an orthogonal matrix \(\mathbf{P}\) such that \(\mathbf{P}^{\mathsf{T}}\mathbf{AP}\) is diagonal, for a symmetric matrix \(\mathbf{A}\).
The idea
For a symmetric matrix, eigenvectors for different eigenvalues are perpendicular. Normalise them and use them as the columns of \(\mathbf{P}\). Then \(\mathbf{P}\) is orthogonal (\(\mathbf{P}^{-1} = \mathbf{P}^{\mathsf{T}}\)), and
\[\mathbf{P}^{\mathsf{T}}\mathbf{AP} = \mathbf{D} = \begin{pmatrix} \lambda_1 & 0 & 0 \\ 0 & \lambda_2 & 0 \\ 0 & 0 & \lambda_3 \end{pmatrix}\]
The order must match: column \(k\) of \(\mathbf{P}\) is the eigenvector for the \(k\)th diagonal entry of \(\mathbf{D}\).
If you know two eigenvectors, the third is perpendicular to both: use their vector product.
3 × 3
\(\mathbf{A} = \begin{pmatrix} 6 & -2 & -1 \\ -2 & 6 & -1 \\ -1 & -1 & 5 \end{pmatrix}\) has eigenvalues 3, 6, 8, with eigenvectors \((1, 1, 1)\) for 3 and \((1, 1, -2)\) for 6. Find \(\mathbf{P}\) and \(\mathbf{D}\).
Third eigenvector: \((1, 1, 1) \times (1, 1, -2) = (-3, 3, 0)\), i.e. \((1, -1, 0)\). (Check: \(\mathbf{A}(1, -1, 0) = (8, -8, 0)\) ✓.)
\(\mathbf{P} = \begin{pmatrix} \frac{1}{\sqrt3} & \frac{1}{\sqrt6} & \frac{1}{\sqrt2} \\ \frac{1}{\sqrt3} & \frac{1}{\sqrt6} & -\frac{1}{\sqrt2} \\ \frac{1}{\sqrt3} & -\frac{2}{\sqrt6} & 0 \end{pmatrix}\), \(\quad\mathbf{D} = \begin{pmatrix} 3 & 0 & 0 \\ 0 & 6 & 0 \\ 0 & 0 & 8 \end{pmatrix}\).
- B1ft for \(\mathbf{D}\) with the eigenvalues on the diagonal, M1 for normalising, B1 or A1 for \(\mathbf{P}\) with columns in the same order as \(\mathbf{D}\) (WFM03 Jun 2026 Q1, WFM03 Jan 2024 Q4, WFM03 Jan 2021 Q5).
Often asked as: the last part of an eigenvalue question: "hence write down \(\mathbf{P}\) and \(\mathbf{D}\)".
Not normalising, so \(\mathbf{P}\) isn't orthogonal.
\(\mathbf{D}\) and \(\mathbf{P}\) in different orders.