What Is a Matrix Square Root?

In mathematics, the square root of a matrix extends the notion of square root from numbers to matrices. A matrix is said to be a square root of if the matrix product is equal to .

Some authors use the name square root or the notation only for the specific case when is positive semidefinite, to denote the unique matrix that is positive semidefinite and such that (for real-valued matrices, where is the transpose of).

Less frequently, the name square root may be used for any factorization of a positive semidefinite matrix as, as in the Cholesky factorization, even if . This distinct meaning is discussed in .

В математике , то квадратный корень из матрицы расширяет понятие квадратного корня из чисел в матрицы . Матрица называется квадратный корень из , если произведение матриц равна .

Некоторые авторы используют квадратный корень имени или обозначение только для конкретного случая, когда является положительно полуопределенным , чтобы обозначить уникальную матрицу которая является положительно полуопределенной и такая, что BB = B T B = A (для действительных значений матрицы, где является транспонированной из ).

Менее часто, название квадратный корень может быть использован для любого разложения положительной полуопределенной матрицы , как B T B = A , как и в разложении Холецкого , даже если BB ≠ A . Это особое значение обсуждается в Положительно определенной матрице § Разложение .

Syntax

X = sqrtm(A) returns a
matrix X, such that X2
= A and the eigenvalues of X are the square roots
of the eigenvalues of A.

Examples

Compute the square root of this matrix. Because these numbers are not
symbolic objects, you get floating-point results.

X =
1.3333 -0.6667 0.0000
-0.3333 1.6667 -0.0000
-0.0572 0.5286 1.4142

Now, convert this matrix to a symbolic object, and compute its square root again:

Check the correctness of the result:

ans =
3×3 logical array
1 1 1
1 1 1
1 1 1

Return Residual of Matrix Square Root

Use the syntax with two output arguments to return the square root of
a matrix and the residual:

Input Arguments

Input, specified as a symbolic matrix.

Output Arguments

Matrix square root, returned as a symbolic matrix such that
X2 = A.

Resnorm — Residual

Residual, returned as a symbolic expression. The residual is computed as
norm(A-X^2,’fro’)/norm(A,’fro’).

Tips

Introduced in R2013a

I’m trying to take the square root of a matrix. That is find the matrix B so B*B=A. None of the methods I’ve found around gives a working result.

First I found this formula on Wikipedia:

Set Y_0 = A and Z_0 = I then the iteration:

Then Y should converge to B.

However implementing the algorithm in python (using numpy for inverse matrices), gave me rubbish results:

As you can see, iterating 100 times, gives worse results than iterating three times, and none of the results get within a 40% error margin.

Then I tried the scipy sqrtm method, but that was even worse:

I don’t know a lot about matrix square rooting, but I figure there must be algorithms that perform better than the above?

Как правило, матрица может иметь несколько квадратных корней. В частности, если и тогда .

Единичная матрица 2 × 2 имеет бесконечно много квадратных корней. Они даны

где любые числа (действительные или комплексные) такие, что . В частности, if является любой т. Е. Любым набором положительных целых чисел, таким, что , то
есть матрица квадратного корня, симметричная и имеющая рациональные элементы. Таким образом

Минус идентичность имеет квадратный корень, например:

который может использоваться для представления мнимой единицы и, следовательно, всех комплексных чисел с использованием вещественных матриц 2 × 2, см. Матричное представление комплексных чисел .

Как и в случае с действительными числами , реальная матрица может не иметь действительного квадратного корня, но иметь квадратный корень с элементами со сложными значениями . Некоторые матрицы не имеют квадратного корня. Примером может служить матрица

Хотя квадратный корень неотрицательного целого числа снова является целым или иррациональным числом , в отличие от целочисленной матрицы, она может иметь квадратный корень, элементы которого рациональны, но не целочислены, как в примерах выше.

X = sqrtm(A) returns
the principal square root of the matrix A, that
is, X*X = A.

X is the unique square root for which every
eigenvalue has nonnegative real part. If A has
any eigenvalues with negative real parts, then a complex result is
produced. If A is singular, then A might
not have a square root. If exact singularity is detected, a warning
is printed.

Create a matrix representation of the fourth difference operator, A. This matrix is symmetric and positive definite.

A =

5 -4 1 0 0
-4 6 -4 1 0
1 -4 6 -4 1
0 1 -4 6 -4
0 0 1 -4 6

Calculate the unique positive definite square root of A using sqrtm. X is the matrix representation of the second difference operator.

X =

2 -1 0 0 0
-1 2 -1 0 0
0 -1 2 -1 0
0 0 -1 2 -1
0 0 0 -1 2

Matrix with Multiple Squareroots

Consider a matrix that has four squareroots, A.

Two of the squareroots of A are given by Y1 and Y2:

Confirm that Y1 and Y2 are squareroots of matrix A.

ans =
10-3 ×

-0.1258 -0.1997
-0.2995 -0.4254

Calculate the squareroot of A with sqrtm. The sqrtm function chooses the positive square roots and produces Y1, even though Y2 seems to be a more natural result.

Y =

1.5667 1.7408
2.6112 4.1779

Input matrix, specified as a square matrix.

Usage notes and limitations:

Thread-Based Environment Run code in the background using MATLAB® backgroundPool or accelerate code with Parallel Computing Toolbox™ ThreadPool.

This function fully supports thread-based environments. For
more information, see Run MATLAB Functions in Thread-Based Environment.

Version History

Introduced before R2006a

По итерации Денмана – Биверса

Другой способ найти квадратный корень из матрицы размера — это итерация квадратного корня Денмана – Биверса.

Пусть Y 0 = A и Z 0 = I , где I — единичная матрица размера . Итерация определяется как

Поскольку здесь используется пара последовательностей обратных матриц, более поздние элементы которых меняются сравнительно мало, только первые элементы имеют высокие вычислительные затраты, поскольку остаток может быть вычислен из более ранних элементов всего за несколько проходов варианта метода Ньютона для вычисления обратных значений ,

При этом для более поздних значений можно установить, а затем использовать для некоторого небольшого n (возможно, всего 1), и аналогично для

Сходимость не гарантируется даже для матриц, у которых есть квадратные корни, но если процесс сходится, матрица сходится квадратично к квадратному корню , в то время как сходится к своему обратному корню

По вавилонскому методу

Еще один итерационный метод получается, если взять известную формулу вавилонского метода вычисления квадратного корня из действительного числа и применить ее к матрицам. Пусть X 0 = I , где I — единичная матрица . Итерация определяется как

Опять же, сходимость не гарантируется, но если процесс сходится, матрица квадратично сходится к квадратному корню . По сравнению с итерацией Денмана – Биверса преимущество вавилонского метода состоит в том, что на каждом шаге итерации необходимо вычислять только одну . С другой стороны, поскольку итерация Денмана – Биверса использует пару последовательностей обратных матриц, более поздние элементы которых меняются сравнительно мало, только первые элементы имеют высокие вычислительные затраты, поскольку остаток может быть вычислен из более ранних элементов всего за несколько проходов вариант метода итерацию Денмана – Биверса выше); конечно, тот же подход можно использовать для получения единственной последовательности обратных чисел, необходимой для вавилонского метода. Однако, в отличие от итерации Денмана – Биверса, вавилонский метод численно нестабилен и с большей вероятностью не сможет сойтись.

Вавилонский метод следует из метода Ньютона для уравнения и использования для

If you give an array as the argument to a mathematical function in a programming language or problem solving environment you are likely to receive the result of applying the function to each component of the array. For example, here MATLAB exponentiates the integers from 0 to 3:

When the array represents a matrix this componentwise evaluation is not useful, as it does not obey the rules of matrix algebra. To see what properties we might require, consider the matrix square root. This function is notable historically because Cayley considered it in the 1858 paper in which he introduced matrix algebra.

Дополнительно:  Почему греется ноутбук и что с этим делать - Лайфхакер

A square root of an

is a matrix

. Any square root

is a square root of

(here, the superscript

Furthermore, for any nonsingular matrix

If we choose

as a matrix that takes

to its Jordan canonical form

then we have

, so that

, or in other words

can be expressed as

Generalizing from these properties of the matrix square root it is reasonable to ask that a function

of a square matrix

(These are of course not the only requirements; indeed

Many definitions of

Power Series

has the derivatives of

along its diagonals in the upper triangle. Write

is zero apart from a superdiagonal of 1s. The formula for

The series is finite because

is a diagonalizable matrix the definition simplifies considerably: if

Cauchy Integral Formula

For a function

analytic on and inside a closed contour

that encloses the spectrum of

This formula is mathematically elegant and can be used to provide short proofs of certain theoretical results.

Equivalence of Definitions

These and several other definitions turn out to be equivalent under suitable conditions. This was recognized by Rinehart, who wrote in 1955

Computing a Matrix Function in MATLAB

In MATLAB, matrix functions are distinguished from componentwise array evaluation by the postfix “m” on a function name. Thus expm, logm, and sqrtm are matrix function counterparts of exp, log, and sqrt. Compare the example at the start of this article with

The MATLAB function funm calculates

for general matrix functions

References

This is a minimal set of references, which contain further useful references within.

  • Nicholas J. Higham, Functions of Matrices: Theory and Computation, Society for Industrial and Applied Mathematics, Philadelphia, PA, USA, 2008.
  • Roger Horn and Charles Johnson, Topics in Matrix Analysis, Cambridge University Press, 1991.

is any matrix

For a scalar

), there are two square roots (which are equal if

), and they are real if and only if

is real and nonnegative. For

is easily seen to have no square roots. The matrix

has four square roots,

has infinitely many square roots (namely the

involutory matrices), including

and any symmetric orthogonal matrix, such as

(which is a Householder matrix). Clearly, a square root of a diagonal matrix need not be diagonal.

The matrix square root of most practical interest is the one whose eigenvalues lie in the right half-plane, which is called the principal square root, written

is nonsingular and has no eigenvalues on the negative real axis then

has a unique principal square root. For the diagonal matrix

A symmetric positive definite matrix has a unique symmetric positive definite square root. Indeed if

is symmetric positive definite then it has a spectral decomposition

is orthogonal and

is diagonal with positive diagonal elements, and then

is nonsingular then it has at least

square roots, where

In some contexts involving symmetric positive definite matrices

, such as Kalman filtering, a matrix

has structure one can ask whether a square root having the same structure, or some other related structure, exists. Results are known for (for example)

An important distinction is between square roots of

that can be expressed as a polynomial in

(primary square roots) and those that cannot. Square roots of the latter type arise when

which represents a rotation through an angle

radians clockwise. The natural square root of

The matrix square root arises in many applications, often in connection with other matrix problems such as the polar decomposition, matrix geometric means, Markov chains (roots of transition matrices), quadratic matrix equations, and generalized eigenvalue problems. Most often the matrix is symmetric positive definite, but square roots of nonsymmetric matrices are also needed. Among modern applications, the matrix square root can be found in recent papers on machine learning.

Матрица с различными ненулевыми собственными значениями имеет 2 n квадратных корня. Такая матрица имеет собственное разложение VDV − 1, где — матрица, столбцы которой являются собственными векторами матрицы а — диагональная матрица, диагональные элементы которой являются соответствующими собственными значениями . Таким образом, квадратные корни из задаются формулой VD 1/2 V −1 , где — любая матрица квадратного корня из , которая для различных собственных значений должна быть диагональной с диагональными элементами, равными квадратным корням из диагональных элементов. из ; поскольку есть два возможных выбора для квадратного корня из каждого диагонального элемента , есть 2 n вариантов для матрицы .

Это также приводит к доказательству вышеупомянутого наблюдения, что положительно определенная матрица имеет ровно один положительно определенный квадратный корень: положительно определенная матрица имеет только положительные собственные значения, и каждое из этих собственных значений имеет только один положительный квадратный корень; и поскольку собственные значения матрицы квадратного корня являются диагональными элементами , для того, чтобы сама матрица квадратного корня была положительно определенной, необходимо использовать только уникальные положительные квадратные корни исходных собственных значений.

Положительные полуопределенные матрицы

Симметричная вещественная матрица размера n × n называется положительно полуопределенной, если для всех (здесь обозначает , преобразование вектора-столбца в вектор-строку). Квадратная вещественная матрица неотрицательно тогда и только тогда , когда для некоторой матрицы . Таких матриц может быть много разных . Положительно полуопределенная матрица может также иметь много таких матриц , что . Однако всегда имеет ровно один квадратный корень который является положительно полуопределенным (и, следовательно, симметричным). В частности, поскольку должен быть симметричным , поэтому два условия или эквивалентны.

Для комплексных матриц вместо этого используется сопряженное транспонирование , а положительные полуопределенные матрицы являются , что означает .

Теорема Пусть быть неотрицательна матрица (действительной или комплексной). Тогда существует ровно одна положительно полуопределенная матрица

Эта уникальная матрица называется основным , неотрицательным , или положительный квадратный корень (последний в случае положительно определенных матриц ).

Главный квадратный корень вещественной положительно полуопределенной матрицы вещественен. Главный квадратный корень положительно определенной матрицы положительно определен; в более общем плане , ранг главного квадратного корня из так же , как в ранг .

Операция извлечения главного квадратного корня непрерывна на этом наборе матриц. Эти свойства являются следствием применения голоморфного функционального исчисления к матрицам. Существование и единственность главного квадратного корня можно вывести непосредственно из жордановой нормальной формы (см. Ниже).

Решения в закрытом виде

Если матрица идемпотентна , то есть , по определению, одним из ее квадратных корней является сама матрица.

Диагональные и треугольные матрицы

Если — диагональная матрица размера n × n , то некоторые из ее квадратных корней являются диагональными матрицами , где . Если диагональные элементы являются реальными и неотрицательным то неотрицательно, а если квадратные корни берутся без отрицательного знака, результирующая матрица является главным корнем . Диагональная матрица может иметь дополнительные недиагональные корни, если некоторые элементы на диагонали равны, как показано выше на единичной матрице.

Если — верхнетреугольная матрица (то есть ее элементы предназначены для ) и не более одного из ее диагональных элементов равно нулю, то одно верхнетреугольное решение уравнения может быть найдено следующим образом. Поскольку уравнение должно выполняться, пусть будет главным квадратным корнем комплексного числа . По предположению , это гарантирует, что для всех (потому что главные квадратные корни комплексных чисел все лежат на одной половине комплексной плоскости). Из уравнения

мы делаем вывод, что можно вычислить рекурсивно для увеличения от 1 до

Если является верхним треугольником, но имеет несколько нулей на диагонали, то квадратный корень может не существовать, как показано на примере . Обратите внимание, что диагональные элементы треугольной матрицы — это в точности ее Свойства треугольной матрицы

П × п матрица является , если существует матрица и диагональная матрица , такие , что . Это происходит тогда и только тогда, когда составляющих основу для . В этом случае можно выбрать как матрицу с собственными векторами в качестве столбцов, и, таким образом, квадратный корень из

где является любой квадратный корень из . Действительно,

Например, матрицу можно диагонализовать как

имеет главный квадратный корень

давая квадратный корень

Когда является симметричным, диагонализирующая матрица может быть сделана ортогональной матрицей путем подходящего выбора собственных векторов (см. Спектральную теорему ). Тогда обратное к — это просто транспонирование, так что

По разложению Шура

Каждая -комплекснозначная квадратная матрица , независимо от диагонализируемости, имеет , данное , где находится верхняя треугольным и является диагональные элементы ; если не более одного из них равно нулю, то следующий квадратный корень

где квадратный корень из верхней треугольной матрицы можно найти, как описано выше.

Если положительно определено, то все собственные значения являются положительными действительными числами, поэтому выбранная диагональ также состоит из положительных действительных чисел. Следовательно, собственные значения являются положительными действительными числами, что означает, что полученная матрица является главным корнем .

По разложению Жордана

Точно так же , как и для разложения Шуры, каждая квадратная матрица может быть разложена , где нормальной форме Жордана

Достаточно проверить единственность для жордановой клетки с λ = 1. Построенный выше квадрат имеет вид S = I + L, где L — многочлен от N без постоянного члена. Любой другой корень квадратный Т с положительными собственными значениями имеет вид T = I + M с нильпогентных, коммутирующий с N и , следовательно , L . Но тогда 0 = S 2 — T 2 = 2 ( L — M ) ( I + ( L + M ) / 2) . Поскольку и коммутируют, матрица нильпотентна, а I + ( L + M ) / 2 обратима с обратным, заданным рядом Неймана . Следовательно , .

Дополнительно:  Программа для удаления системных приложений на андроид без root прав скачать бесплатно на русском

Если — матрица с положительными собственными значениями и минимальным многочленом p ( t ) , то разложение Жордана на обобщенные собственные подпространства можно вывести из разложения на частичные дроби p ( t ) −1 . Соответствующие проекции на обобщенные собственные подпространства задаются вещественных полиномов в . На каждом собственном подпространстве имеет вид λ ( I + N ), как указано выше. Выражение степенного ряда для квадратного корня в собственном подпространстве показывает, что главный квадратный корень из имеет вид q ( A ), где q ( t ) — многочлен с действительными коэффициентами.

Вспомните формальный степенной ряд , который сходится при условии (поскольку коэффициенты степенного ряда суммируемы). Подключение к этому выражению дает

при условии, что . В силу это условие эквивалентно требованию, чтобы спектр содержался внутри диска . Этот метод определения или вычисления особенно полезен в случае положительного полуопределенного. В этом случае мы имеем и, следовательно , так что выражение определяет квадратный корень, из которого, кроме того, оказывается единственный положительный полуопределенный корень. Этот метод остается в силе для определения квадратных корней операторов в бесконечномерных банаховых или гильбертовых пространствах или некоторых элементах (C *) банаховых алгебр.

Квадратные корни из положительных операторов

В линейной алгебре и теории операторов , учитывая ограниченный положительно полуопределенный оператор (неотрицательный оператор) T в комплексном гильбертовом пространстве, B является квадратным корнем из T, если T = B * B , где B * обозначает эрмитово сопряженное к B . Согласно спектральной теореме , то непрерывное функциональное исчисление может быть применено для получения оператора T 1/2 таким образом, что
Т 1/2 сам по себе является положительной и ( T 1/2 ) 2 = T . Оператор Т 1/2 является уникальный неотрицательное корень квадратный из Т .

Ограниченный неотрицательный оператор в комплексном гильбертовом пространстве самосопряжен по определению. Итак, T = ( T 1/2 ) * T 1/2 . Наоборот, очевидно, что каждый оператор вида B * B неотрицателен. Следовательно, оператор T неотрицателен тогда и только тогда, когда T = B * B для некоторого B (эквивалентно, T = CC * для некоторого C ).

Факторизация Холецкая предоставляет другой конкретный пример квадратного корня, который не следует путать с уникальным неотрицательным квадратным корнем.

Унитарная свобода квадратных корней

Если T — неотрицательный оператор в конечномерном гильбертовом пространстве, то все квадратные корни из T связаны унитарными преобразованиями. Точнее, если T = A * A = B * B , то существует унитарный U такой, что A = UB .

Действительно, возьмем B = T
быть единственным неотрицательным корень квадратный из Т . Если T строго положительно, то B обратимо, и поэтому U = AB −1 унитарно:

Если T неотрицательно, но не строго положительно, то обратное к B не может быть определено, но псевдообратное B + Мура – ​​Пенроуза может быть определено. В этом случае оператор является частичной изометрией , то есть унитарным оператором из диапазона T в себя. Это может быть расширен до унитарного оператора U на всем пространстве, установив его равным идентичности на ядре Т . В более общем смысле это верно для бесконечномерного гильбертова пространства, если, кроме того, T имеет замкнутый диапазон значений . В общем случае, если A , B — замкнутые и плотно определенные операторы в гильбертовом пространстве H , и A * A = B * B , то A = UB, где U — частичная изометрия.

Квадратные корни и унитарная свобода квадратных корней находят применение в функциональном анализе и линейной алгебре.

Если A — обратимый оператор в конечномерном гильбертовом пространстве, то существуют единственный унитарный оператор U и положительный оператор P такие, что

это полярное разложение А . Положительный оператор P является единственным положительным квадратным корнем из положительного оператора A ∗ A , а U определяется как U = AP −1 .

Если A необратимо, то он все еще имеет полярную композицию, в которой P определен таким же образом (и уникален). Унитарный оператор U не единственен. Скорее, можно определить «естественный» унитарный оператор следующим образом: AP + — это унитарный оператор из диапазона A в себя, который может быть расширен тождеством на ядре A ∗ . В результате унитарный оператор U , то получается полярное разложение А .

По результату Чоя линейное отображение

полностью положительна тогда и только тогда, когда она имеет вид

называется матрицей Чоя матрицы Φ. Операторы Kraus соответствуют к, не обязательно квадратному, квадратные корни из М Ф : Для любого квадратного корня B из M Ф , можно получить семейство операторов Крауса V I , отменяя операцию Vec к каждому колонку Ь I из B . Таким образом, все наборы операторов Крауса связаны частичными изометриями.

В квантовой физике матрица плотности для n -уровневой квантовой системы представляет собой комплексную матрицу n × n, которая является положительно полуопределенной со следом 1. Если можно выразить как

где и Σ = 1, множество

Условие трассировки 1 означает

и v i — нормализованный a i . Мы видим, что

дает смешанное состояние ρ .

Фильтр Калмана без запаха

В нецентрированном фильтре Калмана (UKF) квадратный корень из ковариационной матрицы ошибок состояния требуется для нецентрированного преобразования, которое является используемым методом статистической линеаризации. Было представлено сравнение различных методов вычисления квадратного корня матрицы в приложении UKF для объединения датчиков GPS / INS, которое показало, что метод разложения Холецкого лучше всего подходит для приложений UKF.

Iterative solutions

Another way to find the square root of an matrix A is the Denman–Beavers square root iteration.

Let Y0 = A and Z0 = I, where I is the identity matrix. The iteration is defined by

As this uses a pair of sequences of matrix inverses whose later elements change comparatively little, only the first elements have a high computational cost since the remainder can be computed from earlier elements with only a few passes of a variant of Newton’s method for computing inverses,

With this, for later values of one would set

and then use

for some small n (perhaps just 1), and similarly for

Convergence is not guaranteed, even for matrices that do have square roots, but if the process converges, the matrix

converges quadratically to a square root 1/2, while

converges to its inverse, −1/2.

By the Babylonian method

Yet another iterative method is obtained by taking the well-known formula of the Babylonian method for computing the square root of a real number, and applying it to matrices. Let X0 = I, where I is the identity matrix. The iteration is defined by

Again, convergence is not guaranteed, but if the process converges, the matrix

converges quadratically to a square root A1/2. Compared to Denman–Beavers iteration, an advantage of the Babylonian method is that only one matrix inverse need be computed per iteration step. On the other hand, as Denman–Beavers iteration uses a pair of sequences of matrix inverses whose later elements change comparatively little, only the first elements have a high computational cost since the remainder can be computed from earlier elements with only a few passes of a variant of Newton’s method for computing inverses (see Denman–Beavers iteration above); of course, the same approach can be used to get the single sequence of inverses needed for the Babylonian method. However, unlike Denman–Beavers iteration, the Babylonian method is numerically unstable and more likely to fail to converge.

Square roots of positive operators

In linear algebra and operator theory, given a bounded positive semidefinite operator (a non-negative operator) T on a complex Hilbert space, B is a square root of T if T = B* B, where B* denotes the Hermitian adjoint of B. According to the spectral theorem, the continuous functional calculus can be applied to obtain an operator T1/2 such thatT1/2 is itself positive and (T1/2)2 = T. The operator T1/2 is the unique non-negative square root of T.

A bounded non-negative operator on a complex Hilbert space is self adjoint by definition. So Conversely, it is trivially true that every operator of the form B* B is non-negative. Therefore, an operator T is non-negative if and only if T = B* B for some B (equivalently, T = CC* for some C).

The Cholesky factorization provides another particular example of square root, which should not be confused with the unique non-negative square root.

Unitary freedom of square roots

If T is a non-negative operator on a finite-dimensional Hilbert space, then all square roots of T are related by unitary transformations. More precisely, if T = A*A = B*B, then there exists a unitary U such that A = UB.

Indeed, take B = T to be the unique non-negative square root of T. If T is strictly positive, then B is invertible, and so is unitary:

If T is non-negative without being strictly positive, then the inverse of B cannot be defined, but the Moore–Penrose pseudoinverse B+ can be. In that case, the operator is a partial isometry, that is, a unitary operator from the range of T to itself. This can then be extended to a unitary operator U on the whole space by setting it equal to the identity on the kernel of T. More generally, this is true on an infinite-dimensional Hilbert space if, in addition, T has closed range. In general, if A, B are closed and densely defined operators on a Hilbert space H, and A* A = B* B, then A = UB where U is a partial isometry.

Some applications

Square roots, and the unitary freedom of square roots, have applications throughout functional analysis and linear algebra.

Polar decomposition

See main article: Polar decomposition. If A is an invertible operator on a finite-dimensional Hilbert space, then there is a unique unitary operator U and positive operator P such that

Дополнительно:  Критический процесс умер в Windows 10: исправьте эту ошибку

this is the polar decomposition of A. The positive operator P is the unique positive square root of the positive operator A∗A, and U is defined by .

Kraus operators

See main article: Choi’s theorem on completely positive maps.

By Choi’s result, a linear map

is completely positive if and only if it is of the form

where k ≤ nm. Let ⊂ Cn × n be the n2 elementary matrix units. The positive matrix

is called the Choi matrix of Φ. The Kraus operators correspond to the, not necessarily square, square roots of MΦ: For any square root B of MΦ, one can obtain a family of Kraus operators Vi by undoing the Vec operation to each column bi of B. Thus all sets of Kraus operators are related by partial isometries.

Mixed ensembles

See main article: Density matrix.

In quantum physics, a density matrix for an n-level quantum system is an n × n complex matrix ρ that is positive semidefinite with trace 1. If ρ can be expressed as

and Σ pi = 1, the set

is said to be an ensemble that describes the mixed state ρ. Notice is not required to be orthogonal. Different ensembles describing the state ρ are related by unitary operators, via the square roots of ρ. For instance, suppose

The trace 1 condition means

and vi be the normalized ai. We see that

gives the mixed state ρ.

Unscented Kalman Filter

In the Unscented Kalman Filter (UKF), the square root of the state error covariance matrix is required for the unscented transform which is the statistical linearization method used. A comparison between different matrix square root calculation methods within a UKF application of GPS/INS sensor fusion was presented, which indicated that the Cholesky decomposition method was best suited for UKF applications.

In general, a matrix can have several square roots. In particular, if

has infinitely many square roots. They are given by

are any numbers (real or complex) such that

.In particular if

is any Pythagorean triple—that is, any set of positive integers such that

is a square root matrix of

Minus identity has a square root, for example:

which can be used to represent the imaginary unit and hence all complex numbers using 2×2 real matrices, see Matrix representation of complex numbers.

Just as with the real numbers, a real matrix may fail to have a real square root, but have a square root with complex-valued entries.Some matrices have no square root. An example is the matrix

While the square root of a nonnegative integer is either again an integer or an irrational number, in contrast an integer matrix can have a square root whose entries are rational, yet non-integral, as in examples above.

Matrices with distinct eigenvalues

An matrix with distinct nonzero eigenvalues has 2n square roots. Such a matrix,, has an eigendecomposition where is the matrix whose columns are eigenvectors of and is the diagonal matrix whose diagonal elements are the corresponding eigenvalues . Thus the square roots of are given by, where is any square root matrix of, which, for distinct eigenvalues, must be diagonal with diagonal elements equal to square roots of the diagonal elements of ; since there are two possible choices for a square root of each diagonal element of, there are 2n choices for the matrix .

This also leads to a proof of the above observation, that a positive-definite matrix has precisely one positive-definite square root: a positive definite matrix has only positive eigenvalues, and each of these eigenvalues has only one positive square root; and since the eigenvalues of the square root matrix are the diagonal elements of, for the square root matrix to be itself positive definite necessitates the use of only the unique positive square roots of the original eigenvalues.

Solutions in closed form

See also: Square root of a 2 by 2 matrix.

If a matrix is idempotent, meaning

, then by definition one of its square roots is the matrix itself.

Diagonal and triangular matrices

If is a diagonal n × n matrix

,then some of its square roots are diagonal matrices

.If the diagonal elements of are real and non-negative then it is positive semidefinite, and if the square roots are taken with non-negative sign, the resulting matrix is the principal root of .A diagonal matrix may have additional non-diagonal roots if some entries on the diagonal are equal, as exemplified by the identity matrix above.

If is an upper triangular matrix (meaning its entries are

) and at most one of its diagonal entries is zero, then one upper triangular solution of the equation

should be satisfied, let

be the principal square root of the complex number

.By the assumption

, this guarantees that

for all (because the principal square roots of complex numbers all lie on one half of the complex plane).From the equation

we deduce that

can be computed recursively for

increasing from 1 to n-1 as:

If is upper triangular but has multiple zeroes on the diagonal, then a square root might not exist, as exemplified by

.Note the diagonal entries of a triangular matrix are precisely its eigenvalues (see Triangular matrix#Properties).

By diagonalization

An n × n matrix is diagonalizable if there is a matrix and a diagonal matrix such that . This happens if and only if has n eigenvectors which constitute a basis for . In this case, can be chosen to be the matrix with the n eigenvectors as columns, and thus a square root of is

where is any square root of . Indeed,

For example, the matrix

can be diagonalized as, where

has principal square root

giving the square root

When is symmetric, the diagonalizing matrix can be made an orthogonal matrix by suitably choosing the eigenvectors (see spectral theorem). Then the inverse of is simply the transpose, so that

By Schur decomposition

Every complex-valued square matrix

, regardless of diagonalizability, has a Schur decomposition given by

is upper triangular and

is unitary (meaning The eigenvalues of

are exactly the diagonal entries of

where a square root

of the upper triangular matrix

can be found as described above.

is positive definite, then the eigenvalues are all positive reals, so the chosen diagonal of

also consists of positive reals.Hence the eigenvalues of

are positive reals, which means the resulting matrix is the principal root of

By Jordan decomposition

Similarly as for the Schur decomposition, every square matrix

can be decomposed as

where is invertible and is in Jordan normal form.

If is a matrix with positive eigenvalues and minimal polynomial, then the Jordan decomposition into generalized eigenspaces of can be deduced from the partial fraction expansion of . The corresponding projections onto the generalized eigenspaces are given by real polynomials in . On each eigenspace, has the form as above. The power series expression for the square root on the eigenspace show that the principal square root of has the form q(A) where q(t) is a polynomial with real coefficients.

(since the coefficients of the power series are summable). Plugging in

into this expression yields

is contained within the disk

. This method of defining or computing

is especially useful in the case where

is positive semi-definite. In that case, we have

, so that the expression

defines a square root of

which moreover turns out to be the unique positive semi-definite root. This method remains valid to define square roots of operators on infinite-dimensional Banach or Hilbert spaces or certain elements of (C*) Banach algebras.

Positive semidefinite matrices

A symmetric real n × n matrix is called positive semidefinite if

denotes the transpose, changing a column vector into a row vector).A square real matrix is positive semidefinite if and only if

for some matrix .There can be many different such matrices .A positive semidefinite matrix can also have many matrices such that

.However, always has precisely one square root that is positive semidefinite (and hence symmetric).In particular, since is required to be symmetric,

, so the two conditions

is used instead and positive semidefinite matrices are Hermitian, meaning

This unique matrix is called the principal, non-negative, or positive square root (the latter in the case of positive definite matrices).

The principal square root of a real positive semidefinite matrix is real.The principal square root of a positive definite matrix is positive definite; more generally, the rank of the principal square root of is the same as the rank of .

Оцените статью
Master Hi-technology