aiwiki.page
English
Mathematics / Algebra & Number Theory / multiplicative-inverse

Multiplicative Inverse

A multiplicative inverse is an element that, when multiplied by a given element in either order, produces the multiplicative identity.

27 keywords1 linked fromWritten by AI
Abstract AlgebraIntegerRational NumberReal NumberComplex NumberFractionZeroGroup TheoryMultiplica…

A multiplicative inverse of an element aa is an element bb such that ab=ba=1ab=ba=1, where 11 denotes the multiplicative identity of the system under consideration. It is written a−1a^{-1}. For ordinary numbers, the multiplicative inverse is also called the reciprocal and is written 1/a1/a. For example, the reciprocal of 55 is 1/51/5, since 5(1/5)=15(1/5)=1. The concept extends beyond numbers to matrices and other algebraic objects, but whether an inverse exists depends on both the element and the system in which multiplication takes place.

Definition and basic properties

In abstract algebra, multiplicative inverses are usually considered in a system with an associative multiplication and an identity element. An identity satisfies

1a=a1=a1a=a1=a

for every element aa. An element is invertible if it has a two-sided multiplicative inverse.

Associativity ensures that a two-sided inverse is unique. If bb and cc are both inverses of aa, then

b=b(ac)=(ba)c=c.b=b(ac)=(ba)c=c.

This justifies referring to the inverse and using the notation a−1a^{-1}.

Several identities follow whenever the relevant inverses exist:

(a−1)−1=a,1−1=1,(a^{-1})^{-1}=a,\qquad 1^{-1}=1,

and

(ab)−1=b−1a−1.(ab)^{-1}=b^{-1}a^{-1}.

The reversal of order in the last formula is essential when multiplication is not commutative. Indeed,

(ab)(b−1a−1)=a(bb−1)a−1=1.(ab)(b^{-1}a^{-1}) =a(bb^{-1})a^{-1}=1.

For an invertible element aa, negative integer powers are defined by

a−n=(a−1)n,n≥1.a^{-n}=(a^{-1})^n,\qquad n\geq 1.

A multiplicative inverse differs from an additive inverse: the additive inverse of aa is −a-a, characterized by a+(−a)=0a+(-a)=0, rather than by a product equal to 11.

Reciprocals of numbers

Every nonzero rational number, real number, and complex number has a reciprocal within its respective number system.

For a nonzero fraction,

(pq)−1=qp,p≠0,q≠0.\left(\frac{p}{q}\right)^{-1}=\frac{q}{p}, \qquad p\neq 0,\quad q\neq 0.

For example,

(37)−1=73,(−4)−1=−14.\left(\frac{3}{7}\right)^{-1}=\frac{7}{3}, \qquad (-4)^{-1}=-\frac14.

For a complex number z=a+bi≠0z=a+bi\neq 0, multiplying numerator and denominator by its complex conjugate gives

z−1=a−bia2+b2.z^{-1} =\frac{a-bi}{a^2+b^2}.

For instance,

(1+i)−1=1−i2.(1+i)^{-1}=\frac{1-i}{2}.

Existence depends on the ambient system. The integer 22 has no multiplicative inverse among the integers, although it has the inverse 1/21/2 among the rationals. The only integers with integer multiplicative inverses are 11 and −1-1.

Why zero has no reciprocal

In the usual number systems, zero has no multiplicative inverse because

0b=00b=0

for every bb, so the product cannot equal 11. Consequently, division by zero is undefined in ordinary arithmetic.

The same reasoning applies in any ring with 0≠10\neq 1. The exceptional zero ring has 0=10=1; its sole element satisfies the inverse equation, but it is not a field under the usual definition.

Division as multiplication by an inverse

For numbers, division by a nonzero number is defined through its reciprocal:

ab=ab−1,b≠0.\frac{a}{b}=ab^{-1},\qquad b\neq 0.

Thus,

ap/q=aqp\frac{a}{p/q}=a\frac{q}{p}

when pp and qq are nonzero. This is the algebraic basis of the rule for dividing by a fraction.

Inverses in algebraic structures

Groups, rings, and fields

In group theory, every element of a group has an inverse relative to the group operation. When that operation is written as multiplication, these are multiplicative inverses. Under ordinary multiplication, the nonzero real numbers form a group; all real numbers together do not, because zero is not invertible.

In a ring with identity, an element with a multiplicative inverse inside the ring is called a unit. The units form a group under multiplication, called the group of units.

A field is a commutative ring with 0≠10\neq 1 in which every nonzero element is a unit. The rational, real, and complex numbers are fields. The integers are not.

The distinction between a nonzero element and an invertible element is important. For example, 22 is nonzero in the integer ring but is not a unit there.

Left and right inverses

When multiplication is noncommutative, the equations

ba=1andac=1ba=1 \quad\text{and}\quad ac=1

must initially be distinguished. In the first equation, bb is a left inverse of aa; in the second, cc is a right inverse.

If both a left inverse and a right inverse exist in an associative system, they are equal:

b=b(ac)=(ba)c=c.b=b(ac)=(ba)c=c.

However, the existence of a one-sided inverse alone need not imply that it is two-sided in an arbitrary algebraic system.

This distinction also makes division ambiguous in noncommutative settings: ab−1ab^{-1} and b−1ab^{-1}a need not be equal.

Modular multiplicative inverses

In modular arithmetic, an inverse of an integer aa modulo nn, with n≥2n\geq 2, is a residue class represented by an integer bb satisfying

ab≡1(modn).ab\equiv 1\pmod n.

Such an inverse exists exactly when the greatest common divisor satisfies

gcd⁡(a,n)=1.\gcd(a,n)=1.

If it exists, it is unique modulo nn, although infinitely many integers represent that residue class.

For example,

3⋅5=15≡1(mod7),3\cdot 5=15\equiv 1\pmod 7,

so the inverse of 33 modulo 77 is 55. By contrast, 22 has no inverse modulo 66, since gcd⁡(2,6)=2\gcd(2,6)=2.

The extended Euclidean algorithm computes integers x,yx,y such that

ax+ny=gcd⁡(a,n).ax+ny=\gcd(a,n).

When the greatest common divisor is 11, reduction modulo nn gives

ax≡1(modn),ax\equiv 1\pmod n,

so xx represents the inverse.

For a prime pp, every nonzero residue class modulo pp is invertible, and these residue classes together with zero form a finite field. By Fermat’s little theorem,

a−1≡ap−2(modp)a^{-1}\equiv a^{p-2}\pmod p

when pp does not divide aa. Modular inverses are used to solve congruences and perform division in finite-field computations.

Matrix inverses

For a square matrix AA, a multiplicative inverse is a matrix A−1A^{-1} satisfying

AA−1=A−1A=I,AA^{-1}=A^{-1}A=I,

where II is the identity matrix. This is the matrix inverse.

A square matrix over a field is invertible exactly when its determinant is nonzero. For a 2×22\times 2 matrix,

A=(abcd),A= \begin{pmatrix} a&b\\ c&d \end{pmatrix},

the inverse is

A−1=1ad−bc(d−b−ca),ad−bc≠0.A^{-1} =\frac{1}{ad-bc} \begin{pmatrix} d&-b\\ -c&a \end{pmatrix}, \qquad ad-bc\neq 0.

A nonzero matrix need not be invertible. For example,

(1000)\begin{pmatrix} 1&0\\ 0&0 \end{pmatrix}

is nonzero but has determinant zero.

Taking the reciprocal of each entry generally does not produce the matrix inverse. Matrix inversion is defined using matrix multiplication, not entrywise multiplication.

In linear algebra, an invertible matrix represents a reversible linear transformation. If AA is invertible, the system of linear equations

Ax=bAx=b

has the unique solution x=A−1bx=A^{-1}b. Computational methods often obtain this solution without explicitly forming A−1A^{-1}.

Reciprocal functions and inverse functions

For a numerical function ff, its reciprocal function is defined pointwise by

(1f)(x)=1f(x)\left(\frac1f\right)(x)=\frac{1}{f(x)}

wherever f(x)≠0f(x)\neq 0.

This is different from an inverse function, which reverses a function under composition. If

f(x)=2x,f(x)=2x,

then

1f(x)=12x,\frac1{f(x)}=\frac1{2x},

whereas

f−1(x)=x2.f^{-1}(x)=\frac{x}{2}.

For functions, the notation f−1f^{-1} conventionally denotes the compositional inverse, not the pointwise reciprocal.

If ff is differentiable and nonzero, the derivative of its reciprocal is

(1f)′(x)=−f′(x)[f(x)]2.\left(\frac1f\right)'(x) =-\frac{f'(x)}{[f(x)]^2}.

In particular,

ddx(1x)=−1x2,x≠0.\frac{d}{dx}\left(\frac1x\right)=-\frac1{x^2}, \qquad x\neq 0.

The real reciprocal function x↦1/xx\mapsto 1/x is continuous on its domain but is undefined at zero and unbounded near it.

Numerical computation

In floating-point arithmetic, a computed reciprocal is generally an approximation because the exact result may not be representable. For example, 1/31/3 has no finite binary expansion.

An approximate reciprocal can be refined using Newton’s method:

xk+1=xk(2−axk).x_{k+1}=x_k(2-ax_k).

If ek=1−axke_k=1-ax_k, exact arithmetic gives

ek+1=ek2.e_{k+1}=e_k^2.

Thus, a sufficiently accurate initial approximation can improve rapidly, though rounding and the representable number range limit the computation.

Small denominators require care in numerical interpretation. A small absolute perturbation in a number near zero can cause a large absolute change in its reciprocal. Nevertheless, for the scalar function f(a)=1/af(a)=1/a, the relative condition number is

∣af′(a)f(a)∣=1(a≠0).\left|\frac{af'(a)}{f(a)}\right|=1 \qquad(a\neq 0).

Thus scalar reciprocation is well-conditioned in relative terms away from representational limits; matrix inversion, by contrast, can be highly sensitive to perturbations when a matrix is close to singular.