3.3 Adjoint, Inverse and Systems of Linear Equations
The adjoint of a square matrix turns its cofactors into an inverse: $A^{-1}=\frac{1}{|A|}\operatorname{adj}A$ whenever $|A|\ne0$. With the inverse, a system of linear equations $AX=B$ is solved in one line — and when $|A|=0$, the same cofactors decide whether there are no solutions or infinitely many.
To solve $5x=15$ you divide by $5$ — that is, you multiply both sides by $5^{-1}=\tfrac15$, the number that undoes multiplication by $5$. A system such as $2x+y+z=17$, $x+2y+z=15$, $x+y+3z=20$ is a single matrix equation $AX=B$, and we would like to solve it the same way: $X=A^{-1}B$. But matrices cannot be divided, and some of them cannot be undone at all — the nilpotent matrix of Section 3.1 squashes the whole plane onto a line, and no matrix can unsquash it. Which matrices have an inverse, and how do you find it?
The answer comes straight from Section 3.2. Arrange the cofactors of $A$ into a matrix and transpose it: this adjoint multiplied by $A$ gives $|A|$ times the identity. So if $|A|\ne0$, dividing the adjoint by $|A|$ produces $A^{-1}$; if $|A|=0$, there is no inverse. In this section we build the adjoint and its properties, find inverses, and then solve systems of two or three linear equations: by $X=A^{-1}B$ (or Cramer's rule) when $|A|\ne0$, and by the test of consistency when $|A|=0$. JEE Main asks about this in nearly every paper: a chain of properties such as $|\operatorname{adj}(\operatorname{adj}A)|$, an inverse found from a matrix equation, or "for which $\lambda$ and $\mu$ does the system have infinitely many solutions?".
The Adjoint of a Square Matrix
The adjoint (adjugate) of a square matrix $A=[a_{ij}]$ is the transpose of the matrix of its cofactors: $\operatorname{adj}A=[C_{ij}]^{T}$, so the $(i,j)$ entry of $\operatorname{adj}A$ is $C_{ji}$. For a $3\times3$ matrix,
$$\operatorname{adj}A=\begin{bmatrix}C_{11}&C_{21}&C_{31}\\C_{12}&C_{22}&C_{32}\\C_{13}&C_{23}&C_{33}\end{bmatrix}.$$For a $2\times2$ matrix the cofactors are single entries, and $\operatorname{adj}\begin{bmatrix}a&b\\c&d\end{bmatrix}=\begin{bmatrix}d&-b\\-c&a\end{bmatrix}$: swap the diagonal entries and change the signs of the other two.
For every square matrix $A$ of order $n$,
$$A\,(\operatorname{adj}A)=(\operatorname{adj}A)\,A=|A|\,I_n.$$Proof. The $(i,k)$ entry of $A\,(\operatorname{adj}A)$ is row $i$ of $A$ times column $k$ of $\operatorname{adj}A$, and column $k$ of $\operatorname{adj}A$ holds the cofactors of row $k$ of $A$:
$$\big(A\,\operatorname{adj}A\big)_{ik}=a_{i1}C_{k1}+a_{i2}C_{k2}+a_{i3}C_{k3}=\begin{cases}|A|,&i=k\ \ \text{(Theorem 3.7)},\\0,&i\ne k\ \ \text{(Theorem 3.10)}.\end{cases}$$So $A\,(\operatorname{adj}A)$ has $|A|$ down the diagonal and $0$ elsewhere: it is $|A|\,I$. The same argument with columns gives $(\operatorname{adj}A)\,A=|A|\,I$. $\blacksquare$ This is the whole reason the transpose appears in the definition: it puts the cofactors of row $k$ into column $k$, where row-by-column multiplication meets them.
$S=\begin{bmatrix}1&2&1\\2&1&3\\3&3&4\end{bmatrix}$ has third row equal to the sum of the first two, so $|S|=0$. Its adjoint is not the zero matrix. What is $S\,(\operatorname{adj}S)$?
For $A=\begin{bmatrix}5&2\\-3&4\end{bmatrix}$, $\operatorname{adj}A$ is
The Inverse of a Square Matrix
A square matrix $A$ of order $n$ is invertible if there is a square matrix $B$ of the same order with $AB=BA=I_n$. Such a $B$ is called the inverse of $A$ and written $A^{-1}$.
- An inverse, if it exists, is unique.
- $A$ is invertible if and only if $|A|\ne0$ ($A$ is non-singular), and then $$A^{-1}=\frac{1}{|A|}\operatorname{adj}A.$$
- For square matrices, $AB=I$ alone already implies $BA=I$, so $B=A^{-1}$.
In particular, for $ad-bc\ne0$: $\ \begin{bmatrix}a&b\\c&d\end{bmatrix}^{-1}=\dfrac{1}{ad-bc}\begin{bmatrix}d&-b\\-c&a\end{bmatrix}$.
Proof. (1) If $B$ and $C$ are both inverses of $A$, then $B=BI=B(AC)=(BA)C=IC=C$. (2) If $AB=I$, then $|A||B|=|I|=1$ by Theorem 3.9, so $|A|\ne0$. Conversely, if $|A|\ne0$, divide Theorem 3.12 by $|A|$: $A\big(\tfrac{1}{|A|}\operatorname{adj}A\big)=\big(\tfrac1{|A|}\operatorname{adj}A\big)A=I$. (3) If $AB=I$, then $|A|\ne0$ as in (2), so $A^{-1}$ exists, and $B=A^{-1}(AB)=A^{-1}$. $\blacksquare$
If $A$ and $B$ are invertible matrices of the same order and $k\ne0$:
And cancelling is allowed for invertible matrices: if $A$ is invertible and $AB=AC$, then $B=C$.
Proofs. Each one is "check that the claimed inverse works" plus Theorem 3.13(3): $(AB)(B^{-1}A^{-1})=A(BB^{-1})A^{-1}=AA^{-1}=I$; $\ A^{T}(A^{-1})^{T}=(A^{-1}A)^{T}=I^{T}=I$; $\ |A||A^{-1}|=|AA^{-1}|=1$. For cancelling, multiply $AB=AC$ on the left by $A^{-1}$. $\blacksquare$ Compare Section 3.1: $AB=AC$ does not allow cancelling in general — invertibility is exactly what is needed.
With the inverse in hand, the properties of the adjoint that JEE Main loves follow in a line or two each.
For square matrices $A$, $B$ of order $n$ ($n=2$ or $3$) and a number $k$:
$$|\operatorname{adj}A|=|A|^{n-1},\qquad \operatorname{adj}(\operatorname{adj}A)=|A|^{n-2}A,\qquad |\operatorname{adj}(\operatorname{adj}A)|=|A|^{(n-1)^2},$$ $$\operatorname{adj}(kA)=k^{n-1}\operatorname{adj}A,\qquad \operatorname{adj}(A^{T})=(\operatorname{adj}A)^{T},\qquad \operatorname{adj}(AB)=(\operatorname{adj}B)(\operatorname{adj}A).$$For $n=3$: $|\operatorname{adj}A|=|A|^2$, $\operatorname{adj}(\operatorname{adj}A)=|A|\,A$, $|\operatorname{adj}(\operatorname{adj}A)|=|A|^4$.
Proofs (for $|A|\ne0$; the formulas stay true when $|A|=0$). Take determinants in Theorem 3.12: $|A|\,|\operatorname{adj}A|=\big||A|\,I_n\big|=|A|^n$, and divide by $|A|$. Next, $\operatorname{adj}A=|A|A^{-1}$; apply this to $\operatorname{adj}A$ itself: $\operatorname{adj}(\operatorname{adj}A)=|\operatorname{adj}A|\,(\operatorname{adj}A)^{-1}$ $=|A|^{n-1}\cdot\tfrac{1}{|A|}A=|A|^{n-2}A$, using $(\operatorname{adj}A)^{-1}=\big(|A|A^{-1}\big)^{-1}=\tfrac1{|A|}A$. Its determinant is $\big(|A|^{n-2}\big)^n|A|=|A|^{n^2-2n+1}=|A|^{(n-1)^2}$. Every cofactor of $kA$ is an $(n-1)\times(n-1)$ determinant with each row multiplied by $k$, so it gains a factor $k^{n-1}$. Finally $\operatorname{adj}(AB)=|AB|(AB)^{-1}$ $=|A||B|B^{-1}A^{-1}$ $=\big(|B|B^{-1}\big)\big(|A|A^{-1}\big)$ $=(\operatorname{adj}B)(\operatorname{adj}A)$ — the order reverses again. $\blacksquare$
$A$ is a $3\times3$ matrix with $|A|=2$. Then $|\operatorname{adj}A|$ is
Systems of Linear Equations: Solution and Consistency
A system of three linear equations in $x$, $y$, $z$,
$$a_1x+b_1y+c_1z=d_1,\quad a_2x+b_2y+c_2z=d_2,\quad a_3x+b_3y+c_3z=d_3,$$is the matrix equation $AX=B$ with coefficient matrix $A=\begin{bmatrix}a_1&b_1&c_1\\a_2&b_2&c_2\\a_3&b_3&c_3\end{bmatrix}$, $X=\begin{bmatrix}x\\y\\z\end{bmatrix}$, $B=\begin{bmatrix}d_1\\d_2\\d_3\end{bmatrix}$ (two equations in $x$, $y$ work the same way with $2\times2$ matrices). The system is consistent if it has at least one solution and inconsistent if it has none. It is homogeneous if $B=O$; a homogeneous system always has the trivial solution $X=O$.
Geometry explains the possible answers. Two equations in $x$, $y$ are two lines: they meet in one point, are parallel (no point in common) or coincide (infinitely many common points). Three equations in $x$, $y$, $z$ are three planes in space, which meet in a single point, along a whole line, in a whole plane, or not at all. So a linear system has exactly one solution, none, or infinitely many — never exactly two.
If $|A|\ne0$, the system $AX=B$ has exactly one solution,
$$X=A^{-1}B=\frac{1}{|A|}(\operatorname{adj}A)\,B.$$Equivalently (Cramer's rule), with $\Delta=|A|$ and $\Delta_1$, $\Delta_2$, $\Delta_3$ the determinants obtained by replacing the first, second, third column of $A$ by $B$:
$$x=\frac{\Delta_1}{\Delta},\qquad y=\frac{\Delta_2}{\Delta},\qquad z=\frac{\Delta_3}{\Delta}.$$Proof. If $AX=B$, multiply on the left by $A^{-1}$: $X=A^{-1}B$, so there is at most one solution. And $X=A^{-1}B$ really is a solution: $A(A^{-1}B)=B$. For Cramer's rule, the first entry of $(\operatorname{adj}A)B$ is $C_{11}d_1+C_{21}d_2+C_{31}d_3$ — the expansion along column $1$ of the determinant whose first column is $B$ and whose other columns are those of $A$. That is $\Delta_1$; similarly for $\Delta_2$, $\Delta_3$. $\blacksquare$ So $(\operatorname{adj}A)B=\begin{bmatrix}\Delta_1\\\Delta_2\\\Delta_3\end{bmatrix}$ — a fact we use again in a moment.
Let $|A|=0$.
- If $(\operatorname{adj}A)B\ne O$ — that is, at least one of $\Delta_1$, $\Delta_2$, $\Delta_3$ is non-zero — the system $AX=B$ is inconsistent (no solution).
- If $(\operatorname{adj}A)B=O$ ($\Delta_1=\Delta_2=\Delta_3=0$), the system has either infinitely many solutions or none; eliminate to decide.
A homogeneous system $AX=O$ has only the trivial solution if $|A|\ne0$, and infinitely many solutions (non-trivial ones included) if $|A|=0$.
Proof. (1) If $X$ were a solution, then $(\operatorname{adj}A)B=(\operatorname{adj}A)AX=|A|X=O$. So $(\operatorname{adj}A)B\ne O$ rules out every solution. (2) Both cases really happen. The system $x+y+z=1$, $2x+2y+2z=3$, $3x+3y+3z=4$ has every $2\times2$ minor of $A$ equal to $0$, so $\operatorname{adj}A=O$ and all $\Delta_i=0$ — yet it has no solution, since twice the first equation says $2x+2y+2z=2\ne3$. Change the right-hand sides to $1,2,3$ and every point of the plane $x+y+z=1$ is a solution. For the homogeneous system, $|A|\ne0$ gives the unique solution $X=A^{-1}O=O$. If $|A|=0$ and $\operatorname{adj}A\ne O$, some column $v$ of $\operatorname{adj}A$ is non-zero, and $A\,\operatorname{adj}A=|A|I=O$ says $Av=O$; then every multiple $tv$ is a solution too. (If $\operatorname{adj}A=O$, all $2\times2$ minors vanish, the rows of $A$ are multiples of one row $(p,q,r)$, and a non-zero vector such as $(q,-p,0)$ or $(r,0,-p)$ solves every equation.) $\blacksquare$
- Compute $\Delta=|A|$. If $\Delta\ne0$: unique solution — find it by $X=A^{-1}B$ or by Cramer's rule.
- If $\Delta=0$: compute $\Delta_1$, $\Delta_2$, $\Delta_3$ (equivalently $(\operatorname{adj}A)B$). If any is non-zero: no solution.
- If all are zero: look for a combination of equations whose left sides cancel or agree (for example, "equation 3 = equation 1 + equation 2" on the left). If the right sides agree in the same way, there are infinitely many solutions — solve the independent equations with one variable as a parameter; otherwise none.
- Parameter questions ("find $\lambda$, $\mu$"): $\Delta(\lambda)=0$ gives the critical $\lambda$; then the $\Delta_i$ (or the combination of equations) decide $\mu$.
Consider $x+2y+z=4$, $\ 2x-y+3z=3$, $\ 3x+y+\lambda z=\mu$. With $\lambda=4$ and $\mu=9$, how many solutions are there?
1. Forgetting the transpose. $\operatorname{adj}A$ is the transpose of the cofactor matrix: its $(i,j)$ entry is $C_{ji}$. For $2\times2$: swap the diagonal, change the signs of the off-diagonal entries — do not swap them.
2. Powers in the adjoint chain. $|\operatorname{adj}A|=|A|^{n-1}$ (not $|A|^n$), $\operatorname{adj}(kA)=k^{n-1}\operatorname{adj}A$, $\operatorname{adj}(\operatorname{adj}A)=|A|^{n-2}A$. For $n=3$: $|\operatorname{adj}(\operatorname{adj}A)|=|A|^4$.
3. Reversal again. $(AB)^{-1}=B^{-1}A^{-1}$ and $\operatorname{adj}(AB)=\operatorname{adj}B\,\operatorname{adj}A$. And $(A+B)^{-1}\ne A^{-1}+B^{-1}$.
4. "All $\Delta_i=0$" does not prove infinitely many solutions. With $\Delta=0$, non-zero $\Delta_i$ means no solution; but $\Delta=\Delta_1=\Delta_2=\Delta_3=0$ can still be inconsistent ($x+y+z=1$, $2x+2y+2z=3$, $3x+3y+3z=4$). Check by combining the equations.
5. Homogeneous systems are always consistent. The question is only whether there is a non-trivial solution, and that happens exactly when $\Delta=0$ — the $\Delta_i$ are all $0$ anyway.
The system $x+2y=3$, $\ 2x+4y=k$ has infinitely many solutions when
Exercises
$I$ is the identity matrix. "Solve" means: decide whether the system has one, no or infinitely many solutions, and find them.
- 1 Find $\operatorname{adj}A$ for $A=\begin{bmatrix}4&-3\\2&1\end{bmatrix}$ and verify that $A\,\operatorname{adj}A=|A|\,I$.
- 2 Find the inverse of $\begin{bmatrix}5&3\\3&2\end{bmatrix}$.
- 3 Find $|A|$ and $\operatorname{adj}A$ for $A=\begin{bmatrix}1&0&2\\2&1&0\\0&3&1\end{bmatrix}$.
- 4 Hence find $A^{-1}$ for the matrix of part 3, and give its $(1,2)$ entry as a decimal.
- 5 Find every $k$ for which $\begin{bmatrix}k&2&2\\2&k&2\\2&2&k\end{bmatrix}$ has no inverse.
- 6 For $A=\begin{bmatrix}2&3\\1&2\end{bmatrix}$, verify that $(A^{T})^{-1}=(A^{-1})^{T}$.
- 7 Show that $A=\begin{bmatrix}2&5\\1&3\end{bmatrix}$ satisfies $A^2-5A+I=O$ and use it to find $A^{-1}$. Give the $(1,2)$ entry.
- 8 $A$ and $B$ are invertible $3\times3$ matrices with $|A|=2$ and $|B|=5$. Find $|(AB)^{-1}|$ and $|\operatorname{adj}(A^{-1}B)|$.
- 9 Prove that if $A$ is invertible and $A^2=A$, then $A=I$.
- 1 $A$ is $3\times3$ with $|A|=5$. Find $|\operatorname{adj}A|$, $|\operatorname{adj}(\operatorname{adj}A)|$, $|A^{-1}|$ and $|A\,\operatorname{adj}A|$.
- 2 $A$ is $3\times3$ with $|A|=-2$. Find $|\operatorname{adj}(-A)|$ and $|\operatorname{adj}(2A)|$.
- 3 Show that $\operatorname{adj}(\operatorname{adj}A)=A$ for every $3\times3$ matrix $A$ with $|A|=1$.
- 4 Find $\operatorname{adj}A$ and $A^{-1}$ for $A=\begin{bmatrix}1&1&1\\0&1&1\\0&0&1\end{bmatrix}$.
- 5 A $3\times3$ matrix $A$ has $\operatorname{adj}A=\begin{bmatrix}1&2&-1\\0&3&1\\2&1&0\end{bmatrix}$. Find the possible values of $|A|$.
- 6 Prove that the adjoint of a symmetric matrix is symmetric.
- 7 $A$ is $3\times3$ with $|A|=2$. Find $|3\operatorname{adj}(A^{-1})|$.
- 1 Solve $3x+2y=4$, $\ x-3y=5$ by the matrix method. Give $x$.
- 2 Solve $x+y-z=1$, $\ 2x-y+z=5$, $\ x+2y+3z=10$ by the matrix method. Give $z$.
- 3 Use Cramer's rule for $2x-y+z=3$, $\ x+3y-2z=1$, $\ 3x+y+z=8$. Give $\Delta$ and $y$.
- 4 Show that $x+y+2z=3$, $\ 2x-y+z=1$, $\ 3x+3z=5$ is inconsistent.
- 5 Show that $x+y+2z=3$, $\ 2x-y+z=1$, $\ 3x+3z=4$ has infinitely many solutions, and find them.
- 6 Find $k$ for which $kx+2y=5$, $\ 3x+y=1$ has no solution.
- 7 Find $k$ for which $x+2y+3z=0$, $\ 2x+ky+4z=0$, $\ x+y+z=0$ has a non-trivial solution.
- 8 Asha buys $2$ kg of apples, $1$ kg of bananas and $3$ kg of oranges for $\text{₹}\,530$; Ravi buys $1$ kg, $2$ kg and $1$ kg for $\text{₹}\,300$; Meera buys $3$ kg, $1$ kg and $2$ kg for $\text{₹}\,570$. Find the price of each fruit per kg.
- 1 For $x+y+z=2$, $\ 2x+3y+z=5$, $\ x+2y+\lambda z=\mu$, find when the system has a unique solution, no solution and infinitely many solutions. Give $\lambda+\mu$ in the last case.
- 2 Discuss the system $x+ay=1$, $\ ax+y=1$ for all real $a$. For which $a$ is it inconsistent?
- 3 Show that $2x+y-z=0$, $\ x-y+2z=0$, $\ 3x+z=0$ has non-trivial solutions, and find them all.
- 4 Show that $x+y+z=1$, $\ 2x+2y+2z=3$, $\ 3x+3y+3z=4$ has $\Delta=\Delta_1=\Delta_2=\Delta_3=0$ but no solution.
- 5 The system $x+2y-z=3$, $\ 3x-y+2z=1$, $\ 2x-3y+\alpha z=\beta$ has infinitely many solutions. Find $\alpha^2+\beta^2$.
- 6 For $A=\begin{bmatrix}1&2&3\\2&3&4\\3&4&5\end{bmatrix}$, show that $|A|=0$ and use a column of $\operatorname{adj}A$ to find a non-trivial solution of $AX=O$.
E1. JEE Main · MCQ $A$ is a $3\times3$ matrix with $|A|=4$. Then $|\operatorname{adj}(\operatorname{adj}A)|$ equals (A) $16$ (B) $64$ (C) $4096$ (D) $256$
E2. JEE Main · Numerical Let $A=\begin{bmatrix}1&0&2\\0&2&1\\1&1&k\end{bmatrix}$ with $|\operatorname{adj}A|=25$. Find the sum of all possible values of $k$. Give the answer to the nearest integer.
E3. JEE Main · MCQ The system $x-2y+z=1$, $\ 2x+y-3z=4$, $\ 3x-y+\lambda z=\mu$ has no solution if (A) $\lambda=-2$, $\mu=5$ (B) $\lambda\ne-2$ (C) $\lambda=-2$, $\mu\ne5$ (D) $\lambda=2$, $\mu\ne5$
E4. JEE Main · Numerical Let $A=\begin{bmatrix}2&1\\3&2\end{bmatrix}$ and suppose $A^{-1}=\alpha A+\beta I$. Find $\alpha+\beta$. Give the answer to the nearest integer.
E5. JEE Main · MCQ $A$ and $B$ are invertible $3\times3$ matrices with $|A|=2$ and $|B|=3$. Then $\big|\operatorname{adj}(B^{-1}A^{-1})\big|$ equals (A) $\tfrac1{36}$ (B) $\tfrac16$ (C) $36$ (D) $\tfrac1{216}$
E6. JEE Main · Numerical Find the sum of the squares of all real $k$ for which the system $kx-y+2z=0$, $\ x+y=0$, $\ 3x-3y+kz=0$ has a non-trivial solution. Give the answer to the nearest integer.
Research question. In 1929 the mathematician Lester Hill designed a cipher that encrypts pairs of letters by multiplying them by a matrix. How does it work, why must the key matrix be invertible "modulo 26", and how can a matrix inverse also break it?
Why it matters. Modern ciphers such as AES still contain a matrix-multiplication step that mixes letters together; understanding why a mixing matrix must be invertible — and why a purely linear cipher is weak — is the first lesson of cryptography.
What to do.
- Number the letters A $=0$, B $=1$, …, Z $=25$. Take the key $K=\begin{bmatrix}5&8\\3&7\end{bmatrix}$. Split a message into pairs, write each pair as a column $P$, and compute $C=KP$, replacing every entry by its remainder on division by $26$. Encrypt "MATH".
- Decryption needs a matrix $K'$ with $K'K\equiv I$ (mod $26$). Adapt Theorem 3.13: $|K|=11$, and $11\cdot19=209=8\cdot26+1$, so "$\tfrac1{11}$" is $19$ modulo $26$. Show that $K'=19\operatorname{adj}K$ (reduced mod $26$) works, and decrypt your ciphertext.
- Why would the key $\begin{bmatrix}2&1\\4&5\end{bmatrix}$ (determinant $6$) be a bad choice? Find two different pairs of letters that it encrypts to the same pair.
- Known-plaintext attack: suppose a spy learns that two plaintext pairs $P_1$, $P_2$ were encrypted to $C_1$, $C_2$. Explain how the matrix equation $K\,[P_1\ P_2]=[C_1\ C_2]$ reveals $K$, and try it on your own example.
Tools: Paper and a calculator, or a spreadsheet with MOD(); optionally a few lines of Python.
Hand in: Your encryption and decryption of a short message, the two colliding pairs for the bad key, and a paragraph explaining why invertibility (mod $26$) is essential and why the attack in step 4 works.
Going further. A key works exactly when $|K|$ shares no factor with $26$. How many of the $26^4$ possible $2\times2$ keys have a determinant that is even? What does that suggest about how many keys are usable?
🧪 Self-check quiz
Q1. The inverse of $\begin{bmatrix}2&1\\5&3\end{bmatrix}$ is
Q2. $A$ is a $3\times3$ matrix with $|A|=3$. Then $|\operatorname{adj}A|$ is
Q3. For invertible matrices $A$, $B$ of the same order, $(AB)^{-1}$ equals
Q4. For a system $AX=B$ with $|A|=0$ and $(\operatorname{adj}A)B\ne O$,
Q5. A homogeneous system $AX=O$ with $|A|\ne0$ has