On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

Math · Algebra Fundamentals · Grade 9-12

Inverse Matrix

The inverse matrix A−1 undoes A: AA−1=A−1A=I.

For 2×2: [abcd]−1=1ad−bc[d−b−ca], provided ad−bc≠0.

Quick Answer

The inverse matrix A−1 undoes A: AA−1=A−1A=I. Use it to undo a transformation, or to solve a square system Ax=b in one move as x=A−1b. It plays the role that dividing plays for numbers, since matrices cannot be divided directly. An inverse exists only for a square matrix with a nonzero determinant; for a 2×2 matrix, swap the diagonal entries, negate the other two, and divide everything by ad−bc.

Why This Matters

With an inverse, a system Ax=b is solved by one multiplication, and the same A−1 solves it again for every new right-hand side b, which is how a fixed recipe or network gets re-solved for new totals. Inverses also decode messages scrambled by a matrix and reverse transformations in graphics. Skip the determinant check and you can spend a page trying to invert a matrix that has no inverse.

Intuitive Explanation

For numbers, 5 and 15 undo each other: 5⋅15=1, and multiplying by 1 changes nothing. For matrices the role of 1 is played by the identity I=[1001], since IA=AI=A for every 2×2 matrix A. The inverse A−1 is the matrix with AA−1=A−1A=I.

Take A=[3243]. Its determinant is 3⋅3−2⋅4=1. Swap the diagonal entries and negate the other two to get [3−2−43], then divide by the determinant, which here is 1. Check by multiplying: row 1 of AA−1 is (9−8, −6+6)=(1,0) and row 2 is (12−12, −8+9)=(0,1). The product is I.

The division by ad−bc is why the determinant decides everything. If det⁡A=0, the formula would divide by zero. Geometrically, A has flattened the plane onto a line: different points have landed on the same spot, so no matrix can send each one back. Such a matrix is called singular.

Two look-alikes are not inverses. The transpose flips rows and columns, turning [3243] into [3423], and their product is [13181825], not I. Taking the reciprocal of each entry fails too. Only the product test AA−1=I decides.

Think of it like this. A matrix and its inverse are like a scrambling code and its unscrambling code: encode a message with A, decode it with A−1, and you are back where you began. A matrix with determinant 0 is a shredder rather than a code: it sends different messages to the same output, so no unscrambling code can exist.

Core idea. A−1 is the unique matrix with AA−1=I, and it exists only when det⁡A≠0.


When to Use

Use it when you need to undo a matrix: solve Ax=b for a square system, reverse a transformation, or decode a message encoded by multiplication. The signal is a square matrix and a question that dividing would answer, if matrices could be divided. Check det⁡A≠0 first. If the determinant is 0 there is no inverse; row-reduce the system to see whether it has no solutions or infinitely many. For a one-off system, row reduction is often quicker than finding A−1.

Tip. Ask: Is the matrix square with nonzero determinant, so an undo-matrix A−1 exists?

How to Recognize It

Before finding an inverse, check whether the question really needs the undoing matrix itself or only a fact about the matrix.

  1. Does the question want the matrix that reverses A, not just a yes-or-no answer about it?

    Yes means compute A−1. If you only need to know whether A can be reversed, or whether a system has one solution, the determinant alone answers it.

  2. Would you divide by A if A were an ordinary number?

    Solving AX=B, or recovering a message encoded by multiplying by A, is a division question in disguise, and the inverse does the dividing. Sums and scalar multiples of matrices never need it.

  3. Are you being asked to flip the matrix across its main diagonal?

    Turning rows into columns is the transpose, written AT, not the inverse. The two agree only for special matrices such as rotations; for most matrices AT does not undo A.

  4. Will you solve the system only once, with three or more unknowns?

    Then row reduction of the augmented matrix is usually less work than finding A−1 first. The inverse pays off when the same A is used with many different right-hand sides.

Inverse Matrix compared with Determinant, Transpose, and Reciprocal of a number

All four rows are about reversing or reshaping a matrix or a number, so their names get swapped. Ask what you need: the matrix that multiplies A back to I, one number saying whether that matrix exists, a mirror image across the diagonal, or the everyday 1x.

Inverse Matrix

Meaning
Use this when you need to reverse a square matrix's effect or solve Ax=b, and its determinant is nonzero.
Key test
Is the matrix square with nonzero determinant, so an undo-matrix A−1 exists?
Formula
For 2×2: [abcd]−1=1ad−bc[d−b−ca], provided ad−bc≠0.
Example
Find the inverse of A=[4332].

Determinant

Meaning
The scalar test for whether an inverse even exists.
Key test
Use first; if det⁡=0 there is no inverse to compute.
Formula
det⁡=ad−bc
Example
det⁡=0 means singular

Transpose

Meaning
Flips rows and columns; does NOT undo A.
Key test
Use when reflecting a matrix across its diagonal, not inverting it.
Formula
(AT)ij=aji
Example
swap rows and columns

Reciprocal of a number

Meaning
The scalar analogy: 5−1=15.
Key test
Use as the intuition for what 'undo by multiplying' means.
Formula
x⋅x−1=1
Example
5⋅15=1

Formula & Notation

For 2×2: [abcd]−1=1ad−bc[d−b−ca], provided ad−bc≠0.
For A∈Rn×n, A−1 exists iff det⁡(A)≠0, and satisfies AA−1=A−1A=In. For n=2: [abcd]−1=1ad−bc[d−b−ca]. In general, A−1=1det⁡(A)adj(A).

How to read it: A−1 denotes the inverse. I is the identity matrix (1s on diagonal, 0s elsewhere). A matrix with no inverse is called singular.

Worked Examples

Example 1 — Invert a 2x2

Easy

Find the inverse of A=[4332].
Show the steps and the answer
  1. Compute the determinant: 4⋅2−3⋅3=8−9=−1.

    It is not zero, so the inverse exists, and it is the number you will divide by.

  2. Swap the diagonal entries and negate the others: [2−3−34].

    This is the [d−b−ca] part of the formula.

  3. Divide every entry by −1: A−1=[−233−4].

    Dividing by a negative determinant flips every sign.

  4. Multiply back: the rows of AA−1 are (4(−2)+3⋅3, 4⋅3+3(−4))=(1,0) and (3(−2)+2⋅3, 3⋅3+2(−4))=(0,1).

    The product is the identity, so this really is the inverse.

Answer: [−233−4]

Takeaway: Swap, negate, divide by the determinant, then multiply back to confirm you get I.

Example 2 — Singular matrix

Standard

Find the inverse of [2412].
Show the steps and the answer
  1. Compute the determinant first: 2⋅2−4⋅1=0.

    The formula divides by the determinant, so a zero stops it before it starts.

  2. Look at the columns: (4,2) is twice (2,1).

    Parallel columns mean the matrix flattens the plane onto a line.

  3. Show two inputs colliding: the matrix sends (2,−1) to (2⋅2+4(−1), 1⋅2+2(−1))=(0,0), the same output as (0,0) itself.

    When two different points land on the same spot, no matrix can send each back to where it came from.

  4. Conclude: the matrix is singular.

    A zero determinant is exactly the test for having no inverse.

Answer: No inverse (singular)

Takeaway: Compute the determinant before anything else; if it is zero, stop, because there is no inverse.

Example 3 — Find the mistake: the missing 1/det

Application

Find the inverse of N=[4213]. Leo writes: "Swap the 4 and 3 and negate the 2 and 1, so N−1=[3−2−14]." Find the mistake and correct it.
Show the steps and the answer
  1. Test Leo's answer by multiplying: N[3−2−14]=[12−2−8+83−3−2+12]=[100010].

    An inverse must give I; this gives 10I, ten times too big.

  2. Find the determinant: 4⋅3−2⋅1=12−2=10.

    The factor of 10 in the product is exactly the determinant Leo never divided by.

  3. Divide every entry by 10: N−1=110[3−2−14]=[0.3−0.2−0.10.4].

    Scaling by 110 turns 10I into I.

Answer: N−1=110[3−2−14]

Takeaway: The swapped-and-negated matrix becomes the inverse only after dividing by the determinant; multiplying back exposes a missing factor.

Where You'll See It

  • Decoding a scrambled message

    Letters become numbers (A = 1, B = 2, ...) and each pair is encoded by multiplying by [3243], so "HI" =(8,9) becomes (42,59). The receiver multiplies by the inverse [3−2−43]: (3⋅42−2⋅59, −4⋅42+3⋅59)=(8,9), which reads "HI" again.

  • Re-solving a bakery's supply problem

    Cake 1 uses 2 cups of flour and 1 of sugar; cake 2 uses 3 and 2. So A=[2312] and A−1=[2−3−12]. With 31 cups of flour and 18 of sugar, A−1[3118]=[85]: 8 and 5 cakes use everything. Next week's supplies reuse the same A−1.

  • The undo button in a drawing program

    A program stretches a shape with S=[2003], sending the corner (2,3) to (4,9). Undo applies S−1=[120013], which halves every x and divides every y by 3, putting (4,9) back at (2,3).

Common Mistakes

  1. The slip: Forgetting the 1ad−bc factor

    Instead: the 2×2 inverse scales the adjugate by 1/det⁡.

  2. The slip: Mis-swapping entries

    Instead: for 2×2, swap a and d, negate b and c: [d−b−ca].

  3. The slip: Attempting to invert when det⁡=0

    Instead: a singular matrix has no inverse; check the determinant first.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. Find the inverse of [3152].

    Hint: Find the determinant first, then swap and negate.

  2. Find the inverse of [2143].

    Hint: This determinant is not 1, so remember to scale.

  3. Does [6342] have an inverse? Explain using its determinant.

    Hint: Compute ad−bc.

  4. Check that [1−201] is the inverse of [1201] by multiplying them.

    Hint: The product of a matrix and its inverse is the identity.

  5. To undo multiplication by A=[1234], a student plans to take the reciprocal of each entry. Inverse matrix or reciprocals of numbers?

    Hint: Would entry-by-entry reciprocals multiply with A to give the identity?

  6. A student finds the inverse of [4232] as [2−2−34]. What did the student forget?

    Hint: Compute the determinant. Is it 1?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Is the inverse of AB equal to A−1B−1?

No, the order flips: (AB)−1=B−1A−1(AB)^{-1}=B^{-1}A^{-1}. Think of putting on socks and then shoes; to undo it you take off the shoes first and then the socks. Applied to a vector, ABAB does BB first and then AA, so undoing it means undoing AA first and then BB. Check: (AB)(B−1A−1)=A(BB−1)A−1=AA−1=I(AB)(B^{-1}A^{-1})=A(BB^{-1})A^{-1}=AA^{-1}=I.

Why can't I just divide one matrix by another?

Because order matters. For numbers, ba\frac{b}{a} is the same whether you think of it as a−1ba^{-1}b or ba−1ba^{-1}. For matrices, A−1BA^{-1}B and BA−1BA^{-1} are usually different, so a fraction bar would not say which one you meant. Writing the inverse on the correct side keeps it clear: AX=BAX=B gives X=A−1BX=A^{-1}B, while XA=BXA=B gives X=BA−1X=BA^{-1}.

What is the inverse of the inverse?

It is the original matrix: (A−1)−1=A(A^{-1})^{-1}=A, just as the decoding key for a decoding key is the original code. This also gives a quick check on your work: once you have A−1A^{-1}, multiplying it by AA in either order must give the identity matrix. If even one entry of the product is off, there is an arithmetic slip, usually a sign in the swapped-and-negated matrix or a missing 1ad−bc\frac{1}{ad-bc}.

Do inverse matrices have uses outside school problems?

Yes. Computer graphics stores a camera's position and angle as a matrix and uses its inverse to work out where every object sits relative to the camera. The Hill cipher, an early encryption method, encoded blocks of letters by multiplying by a matrix and decoded them with its inverse. In practice computers seldom compute an inverse in full; faster row-reduction methods give the same answers.

Learning Path

Inverse Matrix

You are here

You need the Determinant (to test whether an inverse exists) and Matrix Multiplication (to check AA−1=I) first. This concept adds the undo-matrix and the 2×2 formula: swap, negate, divide by ad−bc. Next, Solving Systems of Equations with Matrices uses x=A−1b to solve a whole system with one multiplication.

See Also