Home

noninvertibility

Noninvertibility is the property of lacking an inverse. In mathematics, an inverse function f^{-1} or a matrix A^{-1} does not exist when the operation is not bijective or the transformation collapses information. In physics and other fields, noninvertibility means the original state cannot be uniquely recovered from its image.

In linear algebra, a square matrix A is noninvertible if det(A) = 0; equivalently, rank(A) < n. Such

Noninvertibility also occurs for non-square matrices. A matrix with full row rank has a left inverse, and

For practical purposes, the Moore–Penrose pseudoinverse A+ provides a best approximate inverse for noninvertible matrices, and

In dynamics and information theory, noninvertible maps cause information loss: repeated application can merge distinct states,

Examples include the linear map f(x, y) = (x, 0) represented by the matrix [[1, 0], [0, 0]]

a
matrix
has
no
two-sided
inverse.
Consequences
include
that
the
linear
system
Ax
=
b
may
have
no
solution
or
infinitely
many
solutions;
the
null
space
is
nontrivial.
a
matrix
with
full
column
rank
has
a
right
inverse.
Only
when
the
matrix
is
square
and
has
full
rank
(det(A)
≠
0)
does
a
two-sided
inverse
exist.
is
widely
used
in
least-squares
solutions,
data
fitting,
and
signal
processing.
leaving
indistinguishable
outcomes.
Beyond
mathematics,
noninvertibility
underpins
certain
one-way
functions
in
computer
science,
where
deriving
the
input
from
the
output
is
easy
in
one
direction
but
hard
in
reverse.
(determinant
zero)
and
the
polynomial
function
f(x)
=
x^2
on
the
real
numbers,
which
is
not
globally
invertible.