Pseudoinverse Calculator

Enter a matrix of any shape to find its pseudoinverse.

Matrix A
Rows
2
Columns
2

Write rows separated by semicolons, like 1, 2; 3, 4.

Try

Result

You entered(1224)+\begin{pmatrix} 1 & 2 \\ 2 & 4 \end{pmatrix}^{+}

A+=(125225225425)\displaystyle A^{+} = \begin{pmatrix} \frac{1}{25} & \frac{2}{25} \\ \frac{2}{25} & \frac{4}{25} \end{pmatrix}

Rank

1\displaystyle 1
Show the work
  1. The Moore–Penrose pseudoinverse

    The pseudoinverse is the one matrix satisfying these four conditions. It exists for every matrix and is the inverse when A is invertible; Ax = b solved with it gives the least-squares solution of smallest length.

    Using AA+A=A, A+AA+=A+, (AA+)T=AA+, (A+A)T=A+AAA^{+}A = A,\ A^{+}AA^{+} = A^{+},\ (AA^{+})^{T} = AA^{+},\ (A^{+}A)^{T} = A^{+}A

    A=(1224),rank⁡A=1A = \begin{pmatrix} 1 & 2 \\ 2 & 4 \end{pmatrix},\quad \operatorname{rank} A = 1
  2. Rank factorisation

    A is neither full column nor full row rank. Split it as C (its pivot columns, independent) times R (the non-zero rows of its reduced form, independent).

    Using A=CRA = CR

    C=(12),R=(12)C = \begin{pmatrix} 1 \\ 2 \end{pmatrix},\quad R = \begin{pmatrix} 1 & 2 \end{pmatrix}
  3. Pseudoinverse from the factors

    Each factor has full rank, so each of the two inverses exists.

    Using A+=RT(RRT)−1(CTC)−1CTA^{+} = R^{T}(RR^{T})^{-1}(C^{T}C)^{-1}C^{T}

    (RRT)−1=(15),(CTC)−1=(15)(RR^{T})^{-1} = \begin{pmatrix} \frac{1}{5} \end{pmatrix},\quad (C^{T}C)^{-1} = \begin{pmatrix} \frac{1}{5} \end{pmatrix}
  4. Check: the four conditions

    All four Penrose conditions hold exactly, so this is the pseudoinverse.

    AA+A=A,A+AA+=A+,AA+=(15252545) (symmetric)AA^{+}A = A,\quad A^{+}AA^{+} = A^{+},\quad AA^{+} = \begin{pmatrix} \frac{1}{5} & \frac{2}{5} \\ \frac{2}{5} & \frac{4}{5} \end{pmatrix}\ \text{(symmetric)}
  5. Result

    The pseudoinverse.

    A+=(125225225425)A^{+} = \begin{pmatrix} \frac{1}{25} & \frac{2}{25} \\ \frac{2}{25} & \frac{4}{25} \end{pmatrix}

Worked examples