Matrix Cookbook

The Matrix Cookbook
[ http://matrixcookbook.com ]
Kaare Brandt Petersen
Michael Syskind Pedersen
Version: November 14, 2008
What is this? These pages are a collection of facts (identities, approximations, inequalities, relations, ...) about matrices and matters relating to them.
It is collected in this form for the convenience of anyone who wants a quick
desktop reference .
Disclaimer: The identities, approximations and relations presented here were
obviously not invented but collected, borrowed and copied from a large amount
of sources. These sources include similar but shorter notes found on the internet
and appendices in books - see the references for a full list.
Errors: Very likely there are errors, typos, and mistakes for which we apologize and would be grateful to receive corrections at cookbook@2302.dk.
Its ongoing: The project of keeping a large repository of relations involving
matrices is naturally ongoing and the version will be apparent from the date in
the header.
Suggestions: Your suggestion for additional content or elaboration of some
topics is most welcome at cookbook@2302.dk.
Keywords: Matrix algebra, matrix relations, matrix identities, derivative of
determinant, derivative of inverse matrix, differentiate a matrix.
Acknowledgements: We would like to thank the following for contributions
and suggestions: Bill Baxter, Brian Templeton, Christian Rishj, Christian
Schr
oppel Douglas L. Theobald, Esben Hoegh-Rasmussen, Glynne Casteel, Jan
Larsen, Jun Bin Gao, J
urgen Struckmeier, Kamil Dedecius, Korbinian Strimmer, Lars Christiansen, Lars Kai Hansen, Leland Wilkinson, Liguo He, Loic
Thibaut, Miguel Bar
ao, Ole Winther, Pavel Sakov, Stephan Hattinger, Vasile
Sima, Vincent Rabaud, Zhaoshui He. We would also like thank The Oticon
Foundation for funding our PhD studies.
1
CONTENTS
CONTENTS
Contents
1 Basics
1.1 Trace and Determinants . . . . . . . . . . . . . . . . . . . . . . .
1.2 The Special Case 2x2 . . . . . . . . . . . . . . . . . . . . . . . . .
2 Derivatives
2.1 Derivatives
2.2 Derivatives
2.3 Derivatives
2.4 Derivatives
2.5 Derivatives
2.6 Derivatives
2.7 Derivatives
2.8 Derivatives
of
of
of
of
of
of
of
of
a Determinant . . . . . . . . . . . .
an Inverse . . . . . . . . . . . . . . .
Eigenvalues . . . . . . . . . . . . . .
Matrices, Vectors and Scalar Forms
Traces . . . . . . . . . . . . . . . . .
vector norms . . . . . . . . . . . . .
matrix norms . . . . . . . . . . . . .
Structured Matrices . . . . . . . . .
3 Inverses
3.1 Basic . . . . . . . . . . .
3.2 Exact Relations . . . . .
3.3 Implication on Inverses .
3.4 Approximations . . . . .
3.5 Generalized Inverse . . .
3.6 Pseudo Inverse . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5
5
5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7
7
8
9
9
11
13
13
14
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
16
16
17
19
19
20
20
4 Complex Matrices
23
4.1 Complex Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2 Higher order and non-linear derivatives . . . . . . . . . . . . . . . 26
4.3 Inverse of complex sum . . . . . . . . . . . . . . . . . . . . . . . 26
5 Solutions and Decompositions
5.1 Solutions to linear equations .
5.2 Eigenvalues and Eigenvectors
5.3 Singular Value Decomposition
5.4 Triangular Decomposition . .
5.5 LU decomposition . . . . . .
5.6 LDM decomposition . . . . .
5.7 LDL decompositions . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
27
27
29
30
32
32
32
32
6 Statistics and Probability

33
6.1 Definition of Moments . . . . . . . . . . . . . . . . . . . . . . . . 33
6.2 Expectation of Linear Combinations . . . . . . . . . . . . . . . . 34
6.3 Weighted Scalar Variable . . . . . . . . . . . . . . . . . . . . . . 35
Petersen & Pedersen, The Matrix Cookbook, Version: November 14, 2008, Page 2
CONTENTS
CONTENTS
7 Multivariate Distributions
7.1 Cauchy . . . . . . . . . .
7.2 Dirichlet . . . . . . . . . .
7.3 Normal . . . . . . . . . .
7.4 Normal-Inverse Gamma .
7.5 Gaussian . . . . . . . . . .
7.6 Multinomial . . . . . . . .
7.7 Students t . . . . . . . .
7.8 Wishart . . . . . . . . . .
7.9 Wishart, Inverse . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
36
36
36
36
36
36
36
36
37
38
8 Gaussians
8.1 Basics . . . . . . . .
8.2 Moments . . . . . .
8.3 Miscellaneous . . . .
8.4 Mixture of Gaussians
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
39
39
41
43
44
9 Special Matrices
9.1 Block matrices . . . . . . . . . . . . . . . .
9.2 Discrete Fourier Transform Matrix, The . .
9.3 Hermitian Matrices and skew-Hermitian . .
9.4 Idempotent Matrices . . . . . . . . . . . . .
9.5 Orthogonal matrices . . . . . . . . . . . . .
9.6 Positive Definite and Semi-definite Matrices
9.7 Singleentry Matrix, The . . . . . . . . . . .
9.8 Symmetric, Skew-symmetric/Antisymmetric
9.9 Toeplitz Matrices . . . . . . . . . . . . . . .
9.10 Transition matrices . . . . . . . . . . . . . .
9.11 Units, Permutation and Shift . . . . . . . .
9.12 Vandermonde Matrices . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
45
45
46
47
48
48
50
51
53
54
55
56
57
10 Functions and Operators

10.1 Functions and Series . . . . .
10.2 Kronecker and Vec Operator
10.3 Vector Norms . . . . . . . . .
10.4 Matrix Norms . . . . . . . . .
10.5 Rank . . . . . . . . . . . . . .
10.6 Integral Involving Dirac Delta
10.7 Miscellaneous . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
58
58
59
61
61
62
62
63
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
Functions
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
A One-dimensional Results
64
A.1 Gaussian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
A.2 One Dimensional Mixture of Gaussians . . . . . . . . . . . . . . . 65
B Proofs and Details
67
B.1 Misc Proofs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
CONTENTS
CONTENTS
Notation and Nomenclature

A
Aij
Ai
Aij
An
A1
A+
A1/2
(A)ij
Aij
[A]ij
a
ai
ai
a
<z
<z
<Z
=z
=z
=Z
Matrix
Matrix indexed for some purpose
Matrix indexed for some purpose or
The n.th power of a square matrix
The inverse matrix of the matrix A
The pseudo inverse matrix of the matrix A (see Sec. 3.6)
The square root of a matrix (if unique), not elementwise
The (i, j).th entry of the matrix A
The (i, j).th entry of the matrix A
The ij-submatrix, i.e. A with i.th row and j.th column deleted
Vector
Vector indexed for some purpose
The i.th element of the vector a
Scalar
Real part of a scalar
Real part of a vector
Real part of a matrix
Imaginary part of a scalar
Imaginary part of a vector
Imaginary part of a matrix
det(A)
Tr(A)
diag(A)
eig(A)
vec(A)
sup
||A||
AT
AT
A
AH
Determinant of A
Trace of the matrix A
Diagonal matrix of the matrix A, i.e. (diag(A))ij = ij Aij
Eigenvalues of the matrix A
The vector-version of the matrix A (see Sec. 10.2.2)
Supremum of a set
Matrix norm (subscript if any denotes what norm)
Transposed matrix
The inverse of the transposed and vice versa, AT = (A1 )T = (AT )1 .
Complex conjugated matrix
Transposed and complex conjugated matrix (Hermitian)
AB
AB
Hadamard (elementwise) product

Kronecker product
0
I
Jij
The null matrix. Zero in all entries.

The identity matrix
The single-entry matrix, 1 at (i, j) and zero elsewhere
A positive definite matrix
A diagonal matrix
Basics
(AB)1
(ABC...)1
(AT )1
(A + B)T
(AB)T
(ABC...)T
(AH )1
(A + B)H
(AB)H
(ABC...)H
1.1
B1 A1
...C1 B1 A1
(A1 )T
A T + BT
BT A T
...CT BT AT
(A1 )H
A H + BH
BH A H
...CH BH AH
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
P
Aii
Pi
i = eig(A)
i i ,
Tr(AT )
Tr(BA)
Tr(A) + Tr(B)
Tr(BCA) = Tr(CAB)
Q
i = eig(A)
i i
n
c det(A),
if A Rnn
det(A)
det(A) det(B)
1/ det(A)
det(A)n
1 + uT v
1 + Tr(A),
small
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
=
=
=
=
=
=
=
=
=
=
Trace and Determinants

Tr(A)
Tr(A)
Tr(A)
Tr(AB)
Tr(A + B)
Tr(ABC)
det(A)
det(cA)
det(AT )
det(AB)
det(A1 )
det(An )
det(I + uvT )
det(I + A)
1.2
BASICS
=
=
=
=
=
=
=
=
=
=
=
=
=
The Special Case 2x2
Consider the matrix A

A=
A11
A21
A12
A22
Determinant and trace

det(A) = A11 A22 A12 A21
(25)
Tr(A) = A11 + A22
(26)
1.2
The Special Case 2x2
BASICS
Eigenvalues
2 Tr(A) + det(A) = 0
1 =
Tr(A) +
p
Tr(A)2 4 det(A)
2
1 + 2 = Tr(A)
p
Tr(A)2 4 det(A)
2 =
2
1 2 = det(A)
Tr(A)
Eigenvectors

v1
A12
1 A11
Inverse
A1 =
1
det(A)

v2

A22
A21
A12
2 A11
A12
A11

(27)
DERIVATIVES
Derivatives
This section is covering differentiation of a number of expressions with respect to

a matrix X. Note that it is always assumed that X has no special structure, i.e.
that the elements of X are independent (e.g. not symmetric, Toeplitz, positive
definite). See section 2.8 for differentiation of structured matrices. The basic
assumptions can be written in a formula as
Xkl
= ik lj
Xij
that is for e.g. vector forms,

xi
x
=
y i
y
x
y

i
x
=
yi
(28)
x
y

=
ij
xi
yj
The following rules are general and very useful when deriving the differential of
an expression ([19]):
A
(X)
(X + Y)
(Tr(X))
(XY)
(X Y)
(X Y)
(X1 )
(det(X))
(ln(det(X)))
XT
XH
2.1
2.1.1
=
=
=
=
=
=
=
=
=
=
=
=
0
(A is a constant)
X
X + Y
Tr(X)
(X)Y + X(Y)
(X) Y + X (Y)
(X) Y + X (Y)
X1 (X)X1
det(X)Tr(X1 X)
Tr(X1 X)
(X)T
(X)H
(29)
(30)
(31)
(32)
(33)
(34)
(35)
(36)
(37)
(38)
(39)
(40)
Derivatives of a Determinant
General form
det(Y)
x
det(Y)
x
x

1 Y
= det(Y)Tr Y
x
" "
#
Y
1 x
= det(Y) Tr Y
x

Y
Y
+Tr Y1
Tr Y1
x
x

#
1 Y
1 Y
Tr Y
Y
x
x
(41)
(42)
2.2
Derivatives of an Inverse
2.1.2
DERIVATIVES
Linear forms
det(X)
X
X det(X)
Xjk
Xik
det(X)(X1 )T
= ij det(X)
(43)
(44)
det(AXB)
X
2.1.3
det(AXB)(X1 )T = det(AXB)(XT )1
(45)
Square forms
If X is square and invertible, then

det(XT AX)
= 2 det(XT AX)XT
X
(46)
If X is not square but A is symmetric, then

det(XT AX)
= 2 det(XT AX)AX(XT AX)1
X
(47)
If X is not square and A is not symmetric, then

det(XT AX)
= det(XT AX)(AX(XT AX)1 + AT X(XT AT X)1 )
X
2.1.4
(48)
Other nonlinear forms
Some special cases are (See [9, 7])

ln det(XT X)|
X
ln det(XT X)
X+
ln | det(X)|
X
det(Xk )
X
2.2
2(X+ )T
= 2XT
=
(X1 )T = (XT )1
= k det(Xk )XT
(49)
(50)
(51)
(52)
Derivatives of an Inverse
From [27] we have the basic identity

Y1
Y 1
= Y1
Y
x
x
(53)
2.3
Derivatives of Eigenvalues
DERIVATIVES
from which it follows

(X1 )kl
Xij
aT X1 b
X
det(X1 )
X
Tr(AX1 B)
X
Tr((X + A)1 )
X
= (X1 )ki (X1 )jl
(54)
= XT abT XT
(55)
= det(X1 )(X1 )T
(56)
= (X1 BAX1 )T
(57)
= ((X + A)1 (X + A)1 )T
(58)
From [32] we have the following result: Let A be an n n invertible square

matrix, W be the inverse of A, and J(A) is an n n -variate and differentiable
function with respect to A, then the partial differentials of J with respect to A
and W satisfy
J
J T
= AT
A
A
W
2.3
Derivatives of Eigenvalues
X
eig(X) =
X
Y
eig(X) =
X
2.4
2.4.1
Tr(X) = I
X
det(X) = det(X)XT
X
(59)
(60)
Derivatives of Matrices, Vectors and Scalar Forms

First Order
xT a
x
aT Xb
X
aT XT b
X
aT Xa
X
X
Xij
(XA)ij
Xmn
(XT A)ij
Xmn
aT x
x
(61)
= abT
(62)
= baT
(63)
aT XT a
X
aaT
= Jij
(64)
(65)
= im (A)nj
(Jmn A)ij
(66)
= in (A)mj
(Jnm A)ij
(67)
2.4
Derivatives of Matrices, Vectors and Scalar Forms
2.4.2
DERIVATIVES
Second Order
X
Xkl Xmn
Xij
klmn
bT XT Xc
X
(Bx + b)T C(Dx + d)
x
(XT BX)kl
Xij
(XT BX)
Xij
Xkl
(68)
kl
= X(bcT + cbT )
(69)
= BT C(Dx + d) + DT CT (Bx + b)
(70)
= lj (XT B)ki + kj (BX)il
(71)
= XT BJij + Jji BX
(Jij )kl = ik jl (72)
See Sec 9.7 for useful properties of the Single-entry matrix Jij
xT Bx
x
bT XT DXc
X
(Xb + c)T D(Xb + c)

X
(B + BT )x
= DT XbcT + DXcbT
=
(D + DT )(Xb + c)bT
(73)
(74)
(75)
Assume W is symmetric, then
(x As)T W(x As) = 2AT W(x As)

s
(x s)T W(x s) = 2W(x s)

x
(x s)T W(x s) = 2W(x s)

s
(x As)T W(x As) = 2W(x As)

x
(x As)T W(x As) = 2W(x As)sT

A
(76)
(77)
(78)
(79)
(80)
As a case with complex values the following holds

(a xH b)2
x
= 2b(a xH b)
(81)
This formula is also known from the LMS algorithm [14]

2.4.3
Higher order and non-linear

n1
X
(Xn )kl
=
(Xr Jij Xn1r )kl
Xij
r=0
(82)
2.5
Derivatives of Traces
DERIVATIVES
For proof of the above, see B.1.1.

n1
X
T n
a X b=
(Xr )T abT (Xn1r )T
X
r=0
n1
Xh
T n T n
a (X ) X b =
X
(83)
Xn1r abT (Xn )T Xr
r=0
+(Xr )T Xn abT (Xn1r )T
(84)
See B.1.1 for a proof.

Assume s and r are functions of x, i.e. s = s(x), r = r(x), and that A is a
constant, then
T
s Ar =
x
(Ax)T (Ax)
x (Bx)T (Bx)
=
=
2.4.4
s
x
T

Ar +
r
x
T
AT s
xT AT Ax
x xT BT Bx
xT AT AxBT Bx
AT Ax
2
2 T
x BBx
(xT BT Bx)2
(85)
(86)
(87)
Gradient and Hessian
Using the above we have for the gradient and the Hessian
f
f
x f =
x
2f
xxT
2.5
= xT Ax + bT x
=
(A + AT )x + b
= A + AT
(88)
(89)
(90)
Assume F (X) to be a differentiable function of each of the elements of X. It

then holds that
Tr(F (X))
= f (X)T
X
where f () is the scalar derivative of F ().
2.5
2.5.1
DERIVATIVES
First Order
Tr(X) = I
X
Tr(XA) = AT
X
Tr(AXB) = AT BT
X
Tr(AXT B) = BA
X
Tr(XT A) = A
X
Tr(AXT ) = A
X
Tr(A X) = Tr(A)I
X
2.5.2
(91)
(92)
(93)
(94)
(95)
(96)
(97)
Second Order
Tr(X2 )
X
Tr(X2 B)
X
Tr(XT BX)
X
Tr(XBXT )
X
Tr(AXBX)
X
Tr(XT X)
X
Tr(BXXT )
X
Tr(BT XT CXB)
X

Tr XT BXC
X
Tr(AXBXT C)
X
h
i
Tr (AXB + C)(AXC + C)T

X
Tr(X X)
X
2XT
(98)
(XB + BX)T
(99)
= BX + BT X
(100)
= XBT + XB
(101)
= A T X T BT + BT X T A T
(102)
2X
(103)
(B + BT )X
(104)
= CT XBBT + CXBBT
(105)
= BXC + BT XCT
(106)
= AT CT XBT + CAXB
(107)
2AT (AXB + C)BT
(108)
Tr(X)Tr(X) = 2Tr(X)I(109)
X
See [7].
2.6
Derivatives of vector norms
2.5.3
DERIVATIVES
Higher Order
Tr(Xk )
X
= k(Xk1 )T
k1
X
Tr(AXk ) =
(Xr AXkr1 )T
X
r=0
T T

T
= CXXT CXBBT
X Tr B X CXX CXB
+CT XBBT XT CT X
+CXBBT XT CX
+CT XXT CT XBBT
(110)
(111)
(112)
2.5.4
Other
Tr(AX1 B) = (X1 BAX1 )T = XT AT BT XT

(113)
X
Assume B and C to be symmetric, then
h
i
Tr (XT CX)1 A = (CX(XT CX)1 )(A + AT )(XT CX)1 (114)

X
h
i
Tr (XT CX)1 (XT BX) = 2CX(XT CX)1 XT BX(XT CX)1

X
+2BX(XT CX)1
(115)
h
i
Tr (A + XT CX)1 (XT BX) = 2CX(A + XT CX)1 XT BX(A + XT CX)1

X
+2BX(A + XT CX)1
(116)
See [7].
Tr(sin(X))
X
2.6
2.6.1
2.7
cos(X)T
(117)
Derivatives of vector norms

Two-norm
xa
||x a||2 =
x
||x a||2
(118)
xa
I
(x a)(x a)T
=
x kx ak2
kx ak2
kx ak32
(119)
||x||22
||xT x||2
=
= 2x
x
x
(120)
Derivatives of matrix norms
For more on matrix norms, see Sec. 10.4.

2.8
Derivatives of Structured Matrices
DERIVATIVES
2.7.1
Frobenius norm
||X||2F =
Tr(XXH ) = 2X
(121)
X
X
See (227). Note that this is also a special case of the result in equation 108.
2.8
Assume that the matrix A has some structure, i.e. symmetric, toeplitz, etc.
In that case the derivatives of the previous section does not apply in general.
Instead, consider the following general rule for differentiating a scalar function
f (A)
"
#
T
X f Akl
f
A
df
(122)
=
= Tr
dAij
Akl Aij
A
Aij
kl
The matrix differentiated with respect to itself is in this document referred to

as the structure matrix of A and is defined simply by
A
= Sij
Aij
(123)
If A has no special structure we have simply Sij = Jij , that is, the structure
matrix is simply the singleentry matrix. Many structures have a representation
in singleentry matrices, see Sec. 9.7.6 for more examples of structure matrices.
2.8.1
The Chain Rule
Sometimes the objective is to find the derivative of a matrix which is a function

of another matrix. Let U = f (X), the goal is to find the derivative of the
function g(U) with respect to X:
g(f (X))
g(U)
=
X
X
(124)
Then the Chain Rule can then be written the following way:
M
g(U)
g(U) X X g(U) ukl
=
=
X
xij
ukl xij
(125)
k=1 l=1
Using matrix notation, this can be written as:

h g(U)
g(U)
U i
= Tr (
)T
.
Xij
U
Xij
(126)
2.8
2.8.2
DERIVATIVES
Symmetric
If A is symmetric, then Sij = Jij + Jji Jij Jij and therefore

T

df
f
f
f
diag
=
+
dA
A
A
A
(127)
That is, e.g., ([5]):

Tr(AX)
X
det(X)
X
ln det(X)
X
2.8.3
= A + AT (A I), see (131)
(128)
det(X)(2X1 (X1 I))
(129)
2X1 (X1 I)
(130)
Diagonal
If X is diagonal, then ([19]):

Tr(AX)
X
2.8.4
= AI
(131)
Toeplitz
Like symmetric matrices and diagonal matrices also Toeplitz matrices has a
special structure which should be taken into account when the derivative with
respect to a matrix with Toeplitz structure.
Tr(AT)
T
Tr(TA)
=
T
(132)
Tr(A)
Tr([AT ]n1 )
Tr([AT ]1n ))
Tr(A)
Tr([[AT ]1n ]n1,2 )

.
.
Tr([[AT ]1n ]2,n1 )
.
.
.
A1n
.
.
.
.
.
.
.
.
.
.
Tr([[AT ]1n ]2,n1 )
.
.
An1
.
.
.
.
.
Tr([[AT ]1n ]n1,2 )
Tr([AT ]n1 )
Tr(A)
Tr([AT ]1n ))
(A)
As it can be seen, the derivative (A) also has a Toeplitz structure. Each value
in the diagonal is the sum of all the diagonal valued in A, the values in the
diagonals next to the main diagonal equal the sum of the diagonal next to the
main diagonal in AT . This result is only valid for the unconstrained Toeplitz
matrix. If the Toeplitz matrix also is symmetric, the same derivative yields
Tr(AT)
Tr(TA)
=
= (A) + (A)T (A) I
T
T
(133)
3
3.1
3.1.1
INVERSES
Inverses
Basic
Definition
The inverse A1 of a matrix A Cnn is defined such that

AA1 = A1 A = I,
(134)
where I is the n n identity matrix. If A1 exists, A is said to be nonsingular.

Otherwise, A is said to be singular (see e.g. [12]).
3.1.2
Cofactors and Adjoint
The submatrix of a matrix A, denoted by [A]ij is a (n 1) (n 1) matrix

obtained by deleting the ith row and the jth column of A. The (i, j) cofactor
of a matrix is defined as
cof(A, i, j) = (1)i+j det([A]ij ),
The matrix of cofactors can be created from the cofactors
cof(A, 1, 1)
cof(A, 1, n)
.
.
..
.
cof(A) =
cof(A, i, j)
.
cof(A, n, 1)
cof(A, n, n)
(135)
(136)
The adjoint matrix is the transpose of the cofactor matrix

adj(A) = (cof(A))T ,
3.1.3
(137)
Determinant
The determinant of a matrix A Cnn is defined as (see [12])

det(A)
n
X
j=1
n
X
(1)j+1 A1j det ([A]1j )
(138)
A1j cof(A, 1, j).
(139)
j=1
3.1.4
Construction
The inverse matrix can be constructed, using the adjoint matrix, by

A1 =
1
adj(A)
det(A)
(140)
For the case of 2 2 matrices, see section 1.2.

3.2
Exact Relations
3.1.5
INVERSES
Condition number
The condition number of a matrix c(A) is the ratio between the largest and the
smallest singular value of a matrix (see Section 5.3 on singular values),
c(A) =
d+
d
(141)
The condition number can be used to measure how singular a matrix is. If the
condition number is large, it indicates that the matrix is nearly singular. The
condition number can also be estimated from the matrix norms. Here
c(A) = kAk kA1 k,
(142)
where k k is a norm such as e.g the 1-norm, the 2-norm, the -norm or the
Frobenius norm (see Sec 10.4p
for more on matrix norms).
The 2-norm of A equals (max(eig(AH A))) [12, p.57]. For a symmetric
matrix, this reduces to ||A||2 = max(|eig(A)|) [12, p.394]. If the matrix ia
symmetric and positive definite, ||A||2 = max(eig(A)). The condition number
based on the 2-norm thus reduces to
kAk2 kA1 k2 = max(eig(A)) max(eig(A1 )) =
3.2
3.2.1
max(eig(A))
.
min(eig(A))
Exact Relations
Basic
(AB)1 = B1 A1
3.2.2
(143)
(144)
The Woodbury identity
The Woodbury identity comes in many variants. The latter of the two can be
found in [12]
(A + CBCT )1
(A + UBV)1
= A1 A1 C(B1 + CT A1 C)1 CT A1 (145)

= A1 A1 U(B1 + VA1 U)1 VA1
(146)
If P, R are positive definite, then (see [30])

(P1 + BT R1 B)1 BT R1 = PBT (BPBT + R)1
3.2.3
(147)
The Kailath Variant

(A + BC)1 = A1 A1 B(I + CA1 B)1 CA1
(148)
See [4, page 153].
3.2
Exact Relations
3.2.4
INVERSES
The Searle Set of Identities
The following set of identities, can be found in [25, page 151],

(I + A1 )1 = A(A + I)1
(A + BBT )1 B = A1 B(I + BT A1 B)1
(A1 + B1 )1 = A(A + B)1 B = B(A + B)1 A
A A(A + B)1 A = B B(A + B)1 B
A1 + B1 = A1 (A + B)B1
(I + AB)1 = I A(I + BA)1 B
(I + AB)1 A = A(I + BA)1
3.2.5
(149)
(150)
(151)
(152)
(153)
(154)
(155)
Rank-1 update of Moore-Penrose Inverse
The following is a rank-1 update for the Moore-Penrose pseudo-inverse of real

valued matrices and proof can be found in [18]. The matrix G is defined below:
(A + cdT )+ = A+ + G
(156)
1 + dT A+ c
A+ c
(A+ )T d
(I AA+ )c
(I A+ A)T d
(157)
(158)
(159)
(160)
(161)
Using the the notation
v
n
w
m
=
=
=
=
=
the solution is given as six different cases, depending on the entities ||w||,
||m||, and . Please note, that for any (column) vector v it holds that v+ =
vT
vT (vT v)1 = ||v||
2 . The solution is:
Case 1 of 6: If ||w|| 6= 0 and ||m|| 6= 0. Then

G
= vw+ (m+ )T nT + (m+ )T w+

1
1
=
vwT
mnT +
mwT
||w||2
||m||2
||m||2 ||w||2
(162)
(163)
Case 2 of 6: If ||w|| = 0 and ||m|| 6= 0 and = 0. Then

G
= vv+ A+ (m+ )T nT
1
1
=
vvT A+
mnT
2
||v||
||m||2
(164)
(165)
3.3
Implication on Inverses
INVERSES
Case 3 of 6: If ||w|| = 0 and 6= 0. Then

T
||v||2
1
||m||2 + T
T +
G = mv A
m+v
(A ) v + n
||v||2 ||m||2 + ||2
(166)
Case 4 of 6: If ||w|| 6= 0 and ||m|| = 0 and = 0. Then
G
= A+ nn+ vw+
1
1
=
A+ nnT
vwT
||n||2
||w||2
(167)
(168)
Case 5 of 6: If ||m|| = 0 and 6= 0. Then

T

1
||w||2 +
||n||2
G = A+ nwT
(169)
A
w
+
n
n
+
v
||n||2 ||w||2 + ||2
Case 6 of 6: If ||w|| = 0 and ||m|| = 0 and = 0. Then

G
3.3
= vv+ A+ A+ nn+ + v+ A+ nvn+

1
1
v T A+ n
=
vvT A+
A+ nnT +
vnT
2
2
||v||
||n||
||v||2 ||n||2
(170)
(171)
Implication on Inverses
(A + B)1 = A1 + B1
If
then
AB1 A = BA1 B
(172)
See [25].
3.3.1
A PosDef identity
Assume P, R to be positive definite and invertible, then

(P1 + BT R1 B)1 BT R1 = PBT (BPBT + R)1
(173)
See [30].
3.4
Approximations
The following is a Taylor expansion if An 0 when n ,

(I + A)1 = I A + A2 A3 + ...
Note the following variant can be useful

1 1
1
1 2
1 3
1
(I + A) =
I A + 2 A 3 A + ...
c c
c
c
c
(174)
(175)
The following approximation is from [22] and holds when A large and symmetric
A A(I + A)1 A
= I A1
(176)
If 2 is small compared to Q and M then

(Q + 2 M)1
= Q1 2 Q1 MQ1
(177)
3.5
Generalized Inverse
3.5
3.5.1
INVERSES
Generalized Inverse
Definition
A generalized inverse matrix of the matrix A is any matrix A such that (see
[26])
AA A = A
(178)
The matrix A is not unique.
3.6
3.6.1
Pseudo Inverse
Definition
The pseudo inverse (or Moore-Penrose inverse) of a matrix A is the matrix A+

that fulfils
I
II
III
IV
AA+ A = A
A+ AA+ = A+
AA+ symmetric
A+ A symmetric
The matrix A+ is unique and does always exist. Note that in case of complex matrices, the symmetric condition is substituted by a condition of being
Hermitian.
3.6
Pseudo Inverse
3.6.2
INVERSES
Properties
Assume A+ to be the pseudo-inverse of A, then (See [3] for some of them)

(A+ )+
(AT )+
(AH )+
(A )+
(A+ A)AH
(A+ A)AT
(cA)+
A+
A+
(AT A)+
(AAT )+
A+
A+
(AH A)+
(AAH )+
(AB)+
f (AH A) f (0)I
f (AAH ) f (0)I
=
=
=
=
=
6
=
=
=
=
=
=
=
=
=
=
=
=
=
A
(A+ )T
(A+ )H
(A+ )
AH
AT
(1/c)A+
(AT A)+ AT
AT (AAT )+
A+ (AT )+
(AT )+ A+
(AH A)+ AH
AH (AAH )+
A+ (AH )+
(AH )+ A+
(A+ AB)+ (ABB+ )+
A+ [f (AAH ) f (0)I]A
A[f (AH A) f (0)I]A+
(179)
(180)
(181)
(182)
(183)
(184)
(185)
(186)
(187)
(188)
(189)
(190)
(191)
(192)
(193)
(194)
(195)
(196)
where A Cnm .
Assume A to have full rank, then
(AA+ )(AA+ ) =
(A+ A)(A+ A) =
Tr(AA+ ) =
Tr(A+ A) =
AA+
A+ A
rank(AA+ )
rank(A+ A)
(See [26])
(See [26])
(197)
(198)
(199)
(200)
For two matrices it hold that

(AB)+
(A B)+
3.6.3
= (A+ AB)+ (ABB+ )+

= A + B+
(201)
(202)
Construction
Assume that A has full rank, then

A nn
A nm
A nm
Square
Broad
Tall
rank(A) = n
rank(A) = n
rank(A) = m
A+ = A1
A+ = AT (AAT )1
A+ = (AT A)1 AT
3.6
Pseudo Inverse
INVERSES
Assume A does not have full rank, i.e. A is nm and rank(A) = r < min(n, m).
The pseudo inverse A+ can be constructed from the singular value decomposition A = UDVT , by
T
A+ = Vr D1
(203)
r Ur
where Ur , Dr , and Vr are the matrices with the degenerated rows and columns
deleted. A different way is this: There do always exist two matrices C n r
and D r m of rank r, such that A = CD. Using these matrices it holds that
A+ = DT (DDT )1 (CT C)1 CT
(204)
See [3].
COMPLEX MATRICES
Complex Matrices
4.1
Complex Derivatives
In order to differentiate an expression f (z) with respect to a complex z, the

Cauchy-Riemann equations have to be satisfied ([7]):
df (z)
<(f (z))
=(f (z))
=
+i
dz
<z
<z
(205)
and
df (z)
<(f (z)) =(f (z))
= i
+
dz
=z
=z
or in a more compact form:
f (z)
f (z)
=i
.
=z
<z
(206)
(207)
A complex function that satisfies the Cauchy-Riemann equations for points in a

region R is said yo be analytic in this region R. In general, expressions involving
complex conjugate or conjugate transpose do not satisfy the Cauchy-Riemann
equations. In order to avoid this problem, a more generalized definition of
complex derivative is used ([24], [6]):
Generalized Complex Derivative:
df (z)
1 f (z)
f (z)
=
i
.
dz
2 <z
=z
(208)
Conjugate Complex Derivative

1 f (z)
f (z)
df (z)
=
+
i
.
dz
2 <z
=z
(209)
The Generalized Complex Derivative equals the normal derivative, when f is an

analytic function. For a non-analytic function such as f (z) = z , the derivative
equals zero. The Conjugate Complex Derivative equals zero, when f is an
analytic function. The Conjugate Complex Derivative has e.g been used by [21]
when deriving a complex gradient.
Notice:
f (z)
f (z)
df (z)
6=
+i
.
(210)
dz
<z
=z
Complex Gradient Vector: If f is a real function of a complex vector z,
then the complex gradient vector is given by ([14, p. 798])
f (z)
=
=
df (z)
dz
f (z)
f (z)
+i
.
<z
=z
(211)
4.1
Complex Derivatives
COMPLEX MATRICES
Complex Gradient Matrix: If f is a real function of a complex matrix Z,

then the complex gradient matrix is given by ([2])
f (Z)
=
=
df (Z)
dZ
f (Z)
f (Z)
+i
.
<Z
=Z
(212)
These expressions can be used for gradient descent algorithms.

4.1.1
The Chain Rule for complex numbers
The chain rule is a little more complicated when the function of a complex
u = f (x) is non-analytic. For a non-analytic function, the following chain rule
can be applied ([7])
g(u)
x
=
=
g u
g u
+
u x u x
g u g u
+
u x
u
x
(213)
Notice, if the function is analytic, the second term reduces to zero, and the function is reduced to the normal well-known chain rule. For the matrix derivative
of a scalar function g(U), the chain rule can be written the following way:
T
T
Tr(( g(U)
Tr(( g(U)
g(U)
U ) U)
U ) U )
=
+
.
X
X
X
4.1.2
(214)
Complex Derivatives of Traces
If the derivatives involve complex numbers, the conjugate transpose is often involved. The most useful way to show complex derivative is to show the derivative
with respect to the real and the imaginary part separately. An easy example is:
Tr(X )
Tr(XH )
=
<X
<X
Tr(XH )
Tr(X )
=i
i
=X
=X
= I
(215)
= I
(216)
Since the two results have the same sign, the conjugate complex derivative (209)
should be used.
Tr(X)
Tr(XT )
=
<X
<X
Tr(X)
Tr(XT )
i
=i
=X
=X
= I
(217)
= I
(218)
Here, the two results have different signs, and the generalized complex derivative
(208) should be used. Hereby, it can be seen that (92) holds even if X is a
4.1
Complex Derivatives
COMPLEX MATRICES
complex number.
Tr(AXH )
<X
Tr(AXH )
i
=X
Tr(AX )
<X
Tr(AX )
i
=X
= A
(219)
= A
(220)
= AT
(221)
= AT
(222)
Tr(XXH )
Tr(XH X)
=
<X
<X
H
Tr(XX )
Tr(XH X)
i
=i
=X
=X
2<X
= i2=X
(223)
(224)
By inserting (223) and (224) in (208) and (209), it can be seen that
Tr(XXH )
= X
X
Tr(XXH )
=X
X
(225)
(226)
Since the function Tr(XXH ) is a real function of the complex matrix X, the
complex gradient matrix (212) is given by
Tr(XXH ) = 2
4.1.3
Tr(XXH )
= 2X
X
(227)
Complex Derivative Involving Determinants
Here, a calculation example is provided. The objective is to find the derivative of

det(XH AX) with respect to X Cmn . The derivative is found with respect to
the real part and the imaginary part of X, by use of (37) and (33), det(XH AX)
can be calculated as (see App. B.1.2 for details)
det(XH AX)
X
=
=
1 det(XH AX)
det(XH AX)
i
2
<X
=X
T
H
H
1 H
det(X AX) (X AX) X A
(228)
and the complex conjugate derivative yields

det(XH AX)
X
=
=
1 det(XH AX)
det(XH AX)
+i
2
<X
=X
det(XH AX)AX(XH AX)1
(229)
4.2
4.2
Higher order and non-linear derivatives
COMPLEX MATRICES
Higher order and non-linear derivatives

(Ax)H (Ax)
x (Bx)H (Bx)
=
=
4.3
xH AH Ax
x xH BH Bx
AH Ax
xH AH AxBH Bx
2 H
2
x BBx
(xH BH Bx)2
(230)
(231)
Inverse of complex sum
Given real matrices A, B find the inverse of the complex sum A + iB. Form
the auxiliary matrices
E = A + tB
F = B tA,
(232)
(233)
and find a value of t such that E1 exists. Then

(A + iB)1
=
=
=
=
=
(1 it)(E + iF)1
(234)
1
1
1
1
1
(1 it)((E + FE F) i(E + FE F) FE )(235)
(1 it)(E + FE1 F)1 (I iFE1 )
(236)
1
1
1
1
(E + FE F) ((I tFE ) i(tI + FE ))
(237)
1
1
1
(E + FE F) (I tFE )
i(E + FE1 F)1 (tI + FE1 )
(238)
5
5.1
5.1.1
SOLUTIONS AND DECOMPOSITIONS
Solutions and Decompositions

Solutions to linear equations
Simple Linear Regression
Assume we have data (xn , yn ) for n = 1, ..., N and are seeking the parameters
a, b R such that yi
= axi + b. With a least squares error function, the optimal
values for a, b can be expressed using the notation
x = (x1 , ..., xN )T
1 = (1, ..., 1)T RN 1
y = (y1 , ..., yN )T
and
Rxx = xT x Rx1 = xT 1
Ryx = yT x Ry1 = yT 1
as

5.1.2
a
b

=
Rxx
Rx1
Rx1
R11
R11 = 1T 1
1
Rx,y
Ry1

(239)
Existence in Linear Systems
Assume A is n m and consider the linear system

Ax = b
(240)
Construct the augmented matrix B = [A b] then

Condition
rank(A) = rank(B) = m
rank(A) = rank(B) < m
rank(A) < rank(B)
5.1.3
Solution
Unique solution x
Many solutions x
No solutions x
Standard Square
Assume A is square and invertible, then

Ax = b
5.1.4
x = A1 b
(241)
Degenerated Square
Assume A is n n but of rank r < n. In that case, the system Ax = b is solved

by
x = A+ b
where A+ is the pseudo-inverse of the rank-deficient matrix, constructed as
described in section 3.6.3.
5.1
Solutions to linear equations5
5.1.5
Cramers rule
The equation
Ax = b,
(242)
where A is square has exactly one solution x if the ith element in x can be
found as
det B
xi =
,
(243)
det A
where B equals A, but the ith column in A has been substituted by b.
5.1.6
Over-determined Rectangular
Assume A to be n m, n > m (tall) and rank(A) = m, then

Ax = b
x = (AT A)1 AT b = A+ b
(244)
that is if there exists a solution x at all! If there is no solution the following

can be useful:
Ax = b
xmin = A+ b
(245)
Now xmin is the vector x which minimizes ||Ax b||2 , i.e. the vector which is
least wrong. The matrix A+ is the pseudo-inverse of A. See [3].
5.1.7
Under-determined Rectangular
Assume A is n m and n < m (broad) and rank(A) = n.

Ax = b
xmin = AT (AAT )1 b
(246)
The equation have many solutions x. But xmin is the solution which minimizes
||Ax b||2 and also the solution with the smallest norm ||x||2 . The same holds
for a matrix version: Assume A is n m, X is m n and B is n n, then
AX = B
Xmin = A+ B
(247)
The equation have many solutions X. But Xmin is the solution which minimizes
||AX B||2 and also the solution with the smallest norm ||X||2 . See [3].
Similar but different: Assume A is square n n and the matrices B0 , B1
are n N , where N > n, then if B0 has maximal rank
AB0 = B1
Amin = B1 BT0 (B0 BT0 )1
(248)
where Amin denotes the matrix which is optimal in a least square sense. An
interpretation is that A is the linear approximation which maps the columns
vectors of B0 into the columns vectors of B1 .
5.1.8
Linear form and zeros

Ax = 0,
A=0
(249)
5.2
Eigenvalues and Eigenvectors5
5.1.9
Square form and zeros
If A is symmetric, then
xT Ax = 0,
5.1.10
A=0
(250)
The Lyapunov Equation

AX + XB = C
vec(X) = (I A + BT I)1 vec(C)
(251)
(252)
Sec 10.2.1 and 10.2.2 for details on the Kronecker product and the vec operator.
5.1.11
Encapsulating Sum
P
n An XBn
= C
1
P T
vec(X) =
vec(C)
n Bn A n
(253)
(254)
See Sec 10.2.1 and 10.2.2 for details on the Kronecker product and the vec
operator.
5.2
5.2.1
Eigenvalues and Eigenvectors

Definition
The eigenvectors v and eigenvalues are the ones satisfying

Avi = i vi
AV = VD,
(D)ij = ij i ,
(255)
(256)
where the columns of V are the vectors vi

5.2.2
General Properties
Assume that A Rnm and B Rmn ,

eig(AB) = eig(BA)
A is n m At most min(n, m) distinct i
rank(A) = r At most r non-zero i
(257)
(258)
(259)
5.3
Singular Value Decomposition

5 SOLUTIONS AND DECOMPOSITIONS
5.2.3
Symmetric
Assume A is symmetric, then

VVT
i
Tr(Ap )
eig(I + cA)
eig(A cI)
eig(A1 )
=
=
=
=
I
R
P
(i.e. V is orthogonal)
(i.e. i is real)
p
i i
1 + ci
i c
1
i
(260)
(261)
(262)
(263)
(264)
(265)
For a symmetric, positive matrix A,

eig(AT A) = eig(AAT ) = eig(A) eig(A)
5.2.4
(266)
Characteristic polynomial
The characteristic polynomial for the matrix A is

0
= det(A I)
= n g1 n1 + g2 n2 ... + (1)n gn
(267)
(268)
Note that the coefficients gj for j = 1, ..., n are the n invariants under rotation
of A. Thus, gj is the sum of the determinants of all the sub-matrices of A taken
j rows and columns at a time. That is, g1 is the trace of A, and g2 is the sum
of the determinants of the n(n 1)/2 sub-matrices that can be formed from A
by deleting all but two rows and columns, and so on see [17].
5.3
Any n m matrix A can be written as

A = UDVT ,
where
5.3.1
U = p
eigenvectors of AAT
D =
diag(eig(AAT ))
V = eigenvectors of AT A
(269)
nn
nm
mm
(270)
Symmetric Square decomposed into squares
Assume A to be n n and symmetric. Then

T
A = V
D
V
,
(271)
where D is diagonal with the eigenvalues of A, and V is orthogonal and the

eigenvectors of A.
5.3

5 SOLUTIONS AND DECOMPOSITIONS
5.3.2
Square decomposed into squares
Assume A Rnn . Then

UT
(272)
where D is diagonal with the square root of the eigenvalues of AAT , V is the
eigenvectors of AAT and UT is the eigenvectors of AT A.
5.3.3
Square decomposed into rectangular
Assume V D UT = 0 then we can expand the SVD of A into

T

D 0
U
A = V V
,
0 D
UT
(273)
where the SVD of A is A = VDUT .

5.3.4
Rectangular decomposition I
Assume A is n m, V is n n, D is n n, UT is n m

A
D
UT
= V
,
(274)
5.3.5
Rectangular decomposition II
Assume A is n m, V is n m, D is m m, UT is m m

D UT
A
V
=
5.3.6
(275)
Rectangular decomposition III
Assume A is n m, V is n n, D is n m, UT is m m

UT
,
A
D
= V
(276)
5.4
Triangular Decomposition
5.4
Triangular Decomposition
5.5
LU decomposition
Assume A is a square matrix with non-zero leading principal minors, then

A = LU
(277)
where L is a unique unit lower triangular matrix and U is a unique upper

triangular matrix.
5.5.1
Cholesky-decomposition
Assume A is a symmetric positive definite square matrix, then

A = UT U = LLT ,
(278)
where U is a unique upper triangular matrix and L is a unique lower triangular

matrix.
5.6
LDM decomposition
Assume A is a square matrix with non-zero leading principal minors1 , then

A = LDMT
(279)
where L, M are unique unit lower triangular matrices and D is a unique diagonal
matrix.
5.7
LDL decompositions
The LDL decomposition are special cases of the LDM decomposition. Assume
A is a non-singular symmetric definite square matrix, then
A = LDLT = LT DL
(280)
where L is a unit lower triangular matrix and D is a diagonal matrix. If A is

also positive definite, then D has strictly positive diagonal entries.
1 If the matrix that corresponds to a principal minor is a quadratic upper-left part of the
larger matrix (i.e., it consists of matrix elements in rows and columns from 1 to k), then the
principal minor is called a leading principal minor. For an n times n square matrix, there are
n leading principal minors. [31]
STATISTICS AND PROBABILITY
Statistics and Probability
6.1
Definition of Moments
Assume x Rn1 is a random variable

6.1.1
Mean
The vector of means, m, is defined by

(m)i = hxi i
6.1.2
(281)
Covariance
The matrix of covariance M is defined by

(M)ij = h(xi hxi i)(xj hxj i)i
(282)
M = h(x m)(x m)T i
(283)
or alternatively as
6.1.3
Third moments
The matrix of third centralized moments in some contexts referred to as

coskewness is defined using the notation
(3)
mijk = h(xi hxi i)(xj hxj i)(xk hxk i)i
(284)
h
i
(3) (3)
M3 = m::1 m::2 ...m(3)
::n
(285)
as
where : denotes all elements within the given index. M3 can alternatively be
expressed as
M3 = h(x m)(x m)T (x m)T i
(286)
6.1.4
Fourth moments
The matrix of fourth centralized moments in some contexts referred to as

cokurtosis is defined using the notation
(4)
mijkl = h(xi hxi i)(xj hxj i)(xk hxk i)(xl hxl i)i
(287)
as
i
h
(4)
(4)
(4)
(4)
(4)
(4)
(4)
(4)
M4 = m::11 m::21 ...m::n1 |m::12 m::22 ...m::n2 |...|m::1n m::2n ...m(4)
::nn
(288)
or alternatively as
M4 = h(x m)(x m)T (x m)T (x m)T i
(289)
6.2
Expectation of Linear Combinations

6 STATISTICS AND PROBABILITY
6.2
Expectation of Linear Combinations
6.2.1
Linear Forms
Assume X and x to be a matrix and a vector of random variables. Then (see

See [26])
E[AXB + C] = AE[X]B + C
Var[Ax] = AVar[x]AT
Cov[Ax, By] = ACov[x, y]BT
(290)
(291)
(292)
Assume x to be a stochastic vector with mean m, then (see [7])

E[Ax + b] = Am + b
E[Ax] = Am
E[x + b] = m + b
6.2.2
(293)
(294)
(295)
Quadratic Forms
Assume A is symmetric, c = E[x] and = Var[x]. Assume also that all

coordinates xi are independent, have the same central moments 1 , 2 , 3 , 4
and denote a = diag(A). Then (See [26])
E[xT Ax] = Tr(A) + cT Ac
(296)
T
2
2
T 2
T
2 T
Var[x Ax] = 22 Tr(A ) + 42 c A c + 43 c Aa + (4 32 )a a (297)
Also, assume x to be a stochastic vector with mean m, and covariance M. Then
(see [7])
E[(Ax + a)(Bx + b)T ]
E[xxT ]
E[xaT x]
E[xT axT ]
E[(Ax)(Ax)T ]
E[(x + a)(x + a)T ]
E[(Ax + a)T (Bx + b)]
E[xT x]
E[xT Ax]
E[(Ax)T (Ax)]
E[(x + a)T (x + a)]
=
=
=
=
=
=
=
=
=
=
=
AMBT + (Am + a)(Bm + b)T

M + mmT
(M + mmT )a
aT (M + mmT )
A(M + mmT )AT
M + (m + a)(m + a)T
(298)
(299)
(300)
(301)
(302)
(303)
Tr(AMBT ) + (Am + a)T (Bm + b)

Tr(M) + mT m
Tr(AM) + mT Am
Tr(AMAT ) + (Am)T (Am)
Tr(M) + (m + a)T (m + a)
(304)
(305)
(306)
(307)
(308)
See [7].
6.3
Weighted Scalar Variable
6.2.3
STATISTICS AND PROBABILITY
Cubic Forms
Assume x to be a stochastic vector with independent coordinates, mean m,

covariance M and central moments v3 = E[(x m)3 ]. Then (see [7])
E[(Ax + a)(Bx + b)T (Cx + c)]
= Adiag(BT C)v3
+Tr(BMCT )(Am + a)
+AMCT (Bm + b)
+(AMBT + (Am + a)(Bm + b)T )(Cm + c)
E[xxT x] = v3 + 2Mm + (Tr(M) + mT m)m
E[(Ax + a)(Ax + a)T (Ax + a)] = Adiag(AT A)v3
+[2AMAT + (Ax + a)(Ax + a)T ](Am + a)
+Tr(AMAT )(Am + a)
E[(Ax + a)bT (Cx + c)(Dx + d)T ] = (Ax + a)bT (CMDT + (Cm + c)(Dm + d)T )
+(AMCT + (Am + a)(Cm + c)T )b(Dm + d)T
+bT (Cm + c)(AMDT (Am + a)(Dm + d)T )
6.3
Weighted Scalar Variable
Assume x Rn1 is a random variable, w Rn1 is a vector of constants and

y is the linear combination y = wT x. Assume further that m, M2 , M3 , M4
denotes the mean, covariance, and central third and fourth moment matrix of
the variable x. Then it holds that
hyi
h(y hyi)2 i
h(y hyi)3 i
h(y hyi)4 i
=
=
=
=
wT m
wT M2 w
wT M3 w w
wT M4 w w w
(309)
(310)
(311)
(312)
MULTIVARIATE DISTRIBUTIONS
Multivariate Distributions
7.1
Cauchy
The density function for a Cauchy distributed vector t RP 1 , is given by

p(t|, ) = P/2
( 1+P
det()1/2
2 )
(1/2) 1 + (t )T 1 (t )(1+P )/2
(313)
where is the location, is positive definite, and denotes the gamma function. The Cauchy distribution is a special case of the Student-t distribution.
7.2
Dirichlet
The Dirichlet distribution is a kind of inverse distribution compared to the

multinomial distribution on the bounded continuous variate x = [x1 , . . . , xP ]
[16, p. 44]
P

P
P
p Y
p
p 1
x
p(x|) = QP
p
(
)
p
p
p
7.3
Normal
The normal distribution is also known as a Gaussian distribution. See sec. 8.
7.4
Normal-Inverse Gamma
7.5
Gaussian
See sec. 8.
7.6
Multinomial
If the vector n contains counts, i.e. (n)i 0, 1, 2, ..., then the discrete multinomial disitrbution for n is given by
d
P (n|a, n) =
where ai are probabilities, i.e. 0 ai 1 and
7.7
d
X
Y
n!
ani ,
n1 ! . . . n d ! i i
ni = n
(314)
ai = 1.
Students t
The density of a Student-t distributed vector t RP 1 , is given by

p(t|, , ) = ()P/2
( +P
det()1/2
2 )

(/2) 1 + 1 (t )T 1 (t )(+P )/2
(315)
7.8
Wishart
where is the location, the scale matrix is symmetric, positive definite,

is the degrees of freedom, and denotes the gamma function. For = 1, the
Student-t distribution becomes the Cauchy distribution (see sec 7.1).
7.7.1
Mean
E(t) = ,
7.7.2
(316)
Variance
cov(t) =
7.7.3
>1
,
2
>2
(317)
Mode
The notion mode meaning the position of the most probable value
mode(t) =
7.7.4
(318)
Full Matrix Version
If instead of a vector t RP 1 one has a matrix T RP N , then the Student-t

distribution for T is
p(T|M, , , )
= N P/2
P
Y
[( + P p + 1)/2]
[( p + 1)/2]
p=1
det()/2 det()N/2

(+P )/2
det 1 + (T M)1 (T M)T
(319)
where M is the location, is the rescaling matrix, is positive definite, is
the degrees of freedom, and denotes the gamma function.
7.8
Wishart
The central Wishart distribution for M RP P , M is positive definite, where

m can be regarded as a degree of freedom parameter [16, equation 3.8.1] [8,
section 2.5],[11]
p(M|, m)
2mP/2 P (P 1)/4
1
QP
p
[ 12 (m + 1 p)]
det()m/2 det(M)(mP 1)/2

1
1
exp Tr( M)
2
7.8.1
(320)
Mean
E(M) = m
(321)
7.9
Wishart, Inverse
7.9
Wishart, Inverse
The (normal) Inverse Wishart distribution for M RP P , M is positive definite, where m can be regarded as a degree of freedom parameter [11]
p(M|, m)
2mP/2 P (P 1)/4
1
QP
p
[ 12 (m + 1 p)]
det()m/2 det(M)(mP 1)/2

1
exp Tr(M1 )
2
7.9.1
(322)
Mean
E(M) =
1
mP 1
(323)
8
8.1
8.1.1
GAUSSIANS
Gaussians
Basics
Density and normalization
The density of x N (m, ) is

p(x) = p

1
exp (x m)T 1 (x m)
2
det(2)
1
Note that if x is d-dimensional, then

Integration and normalization

Z
1
T 1
exp (x m) (x m) dx
2

Z
1 T 1
T 1
exp x x + m x dx
2

Z
1 T
T
exp x Ax + c x dx
2
(324)
det(2) = (2)d det().
p
det(2)

1 T 1
=
det(2) exp m m
2

p
1 T T
1
=
det(2A ) exp c A c
2
p
If X = [x1 x2 ...xn ] and C = [c1 c2 ...cn ], then

Z
p
n
1
1
T
T
T 1
1
exp Tr(X AX) + Tr(C X) dX = det(2A ) exp Tr(C A C)
2
2
The derivatives of the density are
p(x)
x
2p
xxT
8.1.2
= p(x)1 (x m)

= p(x) 1 (x m)(x m)T 1 1
(325)
(326)
Marginal Distribution
Assume x Nx (, ) where

xa
a
x=
=
xb
b

=
a
Tc
c
b

(327)
then
p(xa )
p(xb )
= Nxa (a , a )
= Nxb (b , b )
(328)
(329)
8.1
Basics
8.1.3
GAUSSIANS
Conditional Distribution
Assume x Nx (, ) where

xa
a
x=
=
xb
b

=
a
Tc
c
b

(330)
then
a)
p(xa |xb ) = Nxa (
a ,
b)
p(xb |xa ) = Nxb (
b ,
n
a = a + c 1
b (xb b ) (331)
a = a c 1 T
c
b
n
T
1
b = b + c a (xa a )
(332)
b = b T 1 c
c
a
Note, that the covariance matrices are the Schur complement of the block matrix, see 9.1.5 for details.
8.1.4
Linear combination
Assume x N (mx , x ) and y N (my , y ) then

Ax + By + c N (Amx + Bmy + c, Ax AT + By BT )
Rearranging Means
p
det(2(AT 1 A)1 )
p
Nx [A1 m, (AT 1 A)1 ]
NAx [m, ] =
det(2)
(333)
8.1.5
8.1.6
(334)
Rearranging into squared form
If A is symmetric, then
1
1
1
xT Ax + bT x = (x A1 b)T A(x A1 b) + bT A1 b
2
2
2
1
1
1
T
T
1
T
1
Tr(X AX) + Tr(B X) = Tr[(X A B) A(X A B)] + Tr(BT A1 B)
2
2
2
8.1.7
Sum of two squared forms
In vector formulation (assuming 1 , 2 are symmetric)

1
(x m1 )T 1
1 (x m1 )
2
1
(x m2 )T 1
2 (x m2 )
2
1
= (x mc )T 1
c (x mc ) + C
2
(335)
(336)
(337)
8.2
Moments
1
c
mc
C
GAUSSIANS
1
= 1
(338)
1 + 2
1
1 1
1
1
= (1 + 2 ) (1 m1 + 2 m2 )
(339)
1 T 1
1
1 1
1
(m + mT2 1
(1
=
2 )(1 + 2 )
1 m1 + 2 m2 )(340)
2 1 1

1
T 1
m
+
m
m
(341)
mT1 1
1
2
2
1
2
2
In a trace formulation (assuming 1 , 2 are symmetric)

1
Tr((X M1 )T 1
1 (X M1 ))
2
1
Tr((X M2 )T 1
2 (X M2 ))
2
1
= Tr[(X Mc )T 1
c (X Mc )] + C
2
1
c
Mc
C
8.1.8
(342)
(343)
(344)
1
= 1
(345)
1 + 2
1
1 1
1
1
= (1 + 2 ) (1 M1 + 2 M2 )
(346)
i
1 h 1
1
1 1
1
T
=
Tr (1 M1 + 1
(1
2 M2 ) (1 + 2 )
1 M1 + 2 M2 )
2
1
T 1
Tr(MT1 1
(347)
1 M1 + M2 2 M2 )
2
Product of gaussian densities
Let Nx (m, ) denote a density of x, then

Nx (m1 , 1 ) Nx (m2 , 2 ) = cc Nx (mc , c )
cc
mc
c
(348)
= Nm1 (m2 , (1 + 2 ))

1
1
= p
exp (m1 m2 )T (1 + 2 )1 (m1 m2 )
2
det(2(1 + 2 ))
1 1
1
= (1
(1
1 + 2 )
1 m1 + 2 m2 )
1 1
= (1
1 + 2 )
but note that the product is not normalized as a density of x.
8.2
8.2.1
Moments
Mean and covariance of linear forms
First and second moments. Assume x N (m, )

E(x) = m
(349)
8.2
Moments
GAUSSIANS
Cov(x, x) = Var(x) = = E(xxT ) E(x)E(xT ) = E(xxT ) mmT
(350)
As for any other distribution is holds for gaussians that

E[Ax] = AE[x]
Var[Ax] = AVar[x]AT
Cov[Ax, By] = ACov[x, y]BT
8.2.2
(351)
(352)
(353)
Mean and variance of square forms
Mean and variance of square forms: Assume x N (m, )

E(xxT ) = + mmT
(354)
E[xT Ax] = Tr(A) + mT Am
(355)
Var(xT Ax) = Tr[A(A + AT )] + ...
+mT (A + AT )(A + AT )m
(356)
0 T
0
0 T
0
E[(x m ) A(x m )] = (m m ) A(m m ) + Tr(A) (357)
If = 2 I and A is symmetric, then
Var(xT Ax)
2 4 Tr(A2 ) + 4 2 mT A2 m
(358)
Assume x N (0, 2 I) and A and B to be symmetric, then

Cov(xT Ax, xT Bx) = 2 4 Tr(AB)
8.2.3
Cubic forms
E[xbT xxT ]
8.2.4
(359)
= mbT (M + mmT ) + (M + mmT )bmT

+bT m(M mmT )
(360)
Mean of Quartic Forms

E[xxT xxT ]
= 2( + mmT )2 + mT m( mmT )
+Tr()( + mmT )
E[xxT AxxT ] = ( + mmT )(A + AT )( + mmT )
+mT Am( mmT ) + Tr[A]( + mmT )
T
T
E[x xx x] = 2Tr(2 ) + 4mT m + (Tr() + mT m)2
E[xT AxxT Bx] = Tr[A(B + BT )] + mT (A + AT )(B + BT )m
+(Tr(A) + mT Am)(Tr(B) + mT Bm)
8.3
Miscellaneous
GAUSSIANS
E[aT xbT xcT xdT x]

= (aT ( + mmT )b)(cT ( + mmT )d)
+(aT ( + mmT )c)(bT ( + mmT )d)
+(aT ( + mmT )d)(bT ( + mmT )c) 2aT mbT mcT mdT m
E[(Ax + a)(Bx + b)T (Cx + c)(Dx + d)T ]
= [ABT + (Am + a)(Bm + b)T ][CDT + (Cm + c)(Dm + d)T ]
+[ACT + (Am + a)(Cm + c)T ][BDT + (Bm + b)(Dm + d)T ]
+(Bm + b)T (Cm + c)[ADT (Am + a)(Dm + d)T ]
+Tr(BCT )[ADT + (Am + a)(Dm + d)T ]
E[(Ax + a)T (Bx + b)(Cx + c)T (Dx + d)]
= Tr[A(CT D + DT C)BT ]
+[(Am + a)T B + (Bm + b)T A][CT (Dm + d) + DT (Cm + c)]
+[Tr(ABT ) + (Am + a)T (Bm + b)][Tr(CDT ) + (Cm + c)T (Dm + d)]
See [7].
8.2.5
Moments
E[x]
k m k
(361)
Cov(x)
XX
8.3
8.3.1
k k0 (k + mk mTk mk mTk0 )
(362)
k0
Miscellaneous
Whitening
Assume x N (m, ) then

z = 1/2 (x m) N (0, I)
(363)
Conversely having z N (0, I) one can generate data x N (m, ) by setting

x = 1/2 z + m N (m, )
1/2
1/2
Note that
means the matrix which fulfils
and is unique since is positive definite.
8.3.2
1/2
(364)
= , and that it exists
The Chi-Square connection
Assume x N (m, ) and x to be n dimensional, then

z = (x m)T 1 (x m) 2n
(365)
where 2n denotes the Chi square distribution with n degrees of freedom.

8.4
Mixture of Gaussians
8.3.3
GAUSSIANS
Entropy
Entropy of a D-dimensional gaussian

Z
p
D
H(x) = N (m, ) ln N (m, )dx = ln det(2) +
2
8.4
8.4.1
(366)
Mixture of Gaussians
Density
The variable x is distributed as a mixture of gaussians if it has the density

p(x) =
K
X
k=1

1
exp (x mk )T 1
(x
m
)
k
k
2
det(2k )
k p
(367)
where k sum to 1 and the k all are positive definite.

8.4.2
Derivatives
P
Defining p(s) = k k Ns (k , k ) one get
ln p(s)
j
=
=
ln p(s)
j
=
=
ln p(s)
j
=
=
j Ns (j , j )
P
ln[j Ns (j , j )]
(368)
N
(
,
)
j
k
s
k
k
k
j Ns (j , j ) 1
P
(369)
k k Ns (k , k ) j
j Ns (j , j )
P
ln[j Ns (j , j )]
(370)
k k Ns (k , k ) j

j Ns (j , j ) 1
P
j (s j )
(371)
k k Ns (k , k )
j Ns (j , j )
P
ln[j Ns (j , j )]
(372)
N
(
,
k
j
k
k k s

j Ns (j , j ) 1
1
T 1
P
1
j + j (s j )(s j ) j (373)
2
N
(
,
)
k
k
k k s
But k and k needs to be constrained.
9
9.1
SPECIAL MATRICES
Special Matrices
Block matrices
Let Aij denote the ijth block of A.

9.1.1
Multiplication
Assuming the dimensions of the blocks matches we have

A11 A12
B11 B12
A11 B11 + A12 B21 A11 B12 + A12 B22
=
A21 A22
B21 B22
A21 B11 + A22 B21 A21 B12 + A22 B22
9.1.2
The Determinant
The determinant can be expressed as by the use of

= A11 A12 A1
22 A21
= A22 A21 A1
11 A12
C1
C2
as

det
9.1.3

A12
A22
A11
A21
(374)
(375)
= det(A22 ) det(C1 ) = det(A11 ) det(C2 )
The Inverse
The inverse can be expressed as by the use of

= A11 A12 A1
22 A21
= A22 A21 A1
11 A12
C1
C2
as

=
9.1.4
A12
A22
A11
A21
1

=
C1
1
1
C2 A21 A1
11
1
1
1
A1
11 + A11 A12 C2 A21 A11
1
1
A22 A21 C1
(376)
(377)
1
A1
11 A12 C2
1
C2
1
C1
1 A12 A22
1
1
1
A22 + A22 A21 C1 A12 A1
22
Block diagonal
For block diagonal matrices we have

det
A11
0
A11
0
0
A22
0
A22
1

=
(A11 )1
0
0
(A22 )1

(378)

=
det(A11 ) det(A22 )
(379)
9.2
Discrete Fourier Transform Matrix, The
9.1.5
SPECIAL MATRICES
Schur complement
The Schur complement of the matrix

A11
A21
A12
A22
is the matrix
A11 A12 A1
22 A21
that is, what is denoted C2 above. Using the Schur complement, one can rewrite
the inverse of a block matrix
1

A11 A12
A21 A22

1
I
0
0
(A11 A12 A1
I A12 A1
22 A21 )
22
=
A1
I
0
I
0
A1
22 A21
22
The Schur complement is useful when solving linear systems of the form

A11 A12
x1
b1
=
A21 A22
x2
b2
which has the following equation for x1
1
(A11 A12 A1
22 A21 )x1 = b1 A12 A22 b2
When the appropriate inverses exists, this can be solved for x1 which can then
be inserted in the equation for x2 to solve for x2 .
9.2
Discrete Fourier Transform Matrix, The
The DFT matrix is an N N symmetric matrix WN , where the k, nth element

is given by
j2kn
WNkn = e N
(380)
Thus the discrete Fourier transform (DFT) can be expressed as
X(k) =
N
1
X
x(n)WNkn .
(381)
n=0
Likewise the inverse discrete Fourier transform (IDFT) can be expressed as

x(n) =
N 1
1 X
X(k)WNkn .
N
(382)
k=0
The DFT of the vector x = [x(0), x(1), , x(N 1)]T can be written in matrix
form as
X = WN x,
(383)
9.3
Hermitian Matrices and skew-Hermitian
SPECIAL MATRICES
where X = [X(0), X(1), , x(N 1)]T . The IDFT is similarly given as

x = W1
N X.
(384)
Some properties of WN exist:

W1
N
WN WN
WN
If WN = e
j2
N
1
W
N N
= NI
= WH
N
(385)
(386)
(387)
, then [23]
m+N/2
WN
= WNm
(388)
Notice, the DFT matrix is a Vandermonde Matrix.

The following important relation between the circulant matrix and the discrete Fourier transform (DFT) exists
TC = W1
N (I (WN t))WN ,
(389)
where t = [t0 , t1 , , tn1 ] is the first row of TC .
9.3
Hermitian Matrices and skew-Hermitian
A matrix A Cmn is called Hermitian if

AH = A
For real valued matrices, Hermitian and symmetric matrices are equivalent.
A is Hermitian xH Ax R,
A is Hermitian eig(A) R
x Cn1
(390)
(391)
Note that
A = B + iC
where B, C are hermitian, then
B=
9.3.1
A + AH
,
2
C=
A AH
2i
Skew-Hermitian
A matrix A is called skew-hermitian if

A = AH
For real valued matrices, skew-Hermitian and skew-symmetric matrices are
equivalent.
A Hermitian iA is skew-hermitian
A skew-Hermitian xH Ay = xH AH y, x, y
A skew-Hermitian eig(A) = i, R
(392)
(393)
(394)
9.4
Idempotent Matrices
9.4
SPECIAL MATRICES
Idempotent Matrices
A matrix A is idempotent if
AA = A
Idempotent matrices A and B, have the following properties
An = A,
forn = 1, 2, 3, ...
IA
is idempotent
AH
is idempotent
H
IA
is idempotent
If AB = BA AB is idempotent
rank(A) = Tr(A)
A(I A) = 0
(I A)A = 0
A+ = A
f (sI + tA) = (I A)f (s) + Af (s + t)
(395)
(396)
(397)
(398)
(399)
(400)
(401)
(402)
(403)
(404)
Note that A I is not necessarily idempotent.

9.4.1
Nilpotent
A matrix A is nilpotent if
A2 = 0
A nilpotent matrix has the following property:
f (sI + tA)
9.4.2
= If (s) + tAf 0 (s)
(405)
Unipotent
A matrix A is unipotent if
AA = I
A unipotent matrix has the following property:
f (sI + tA)
9.5
[(I + A)f (s + t) + (I A)f (s t)]/2
(406)
Orthogonal matrices
If a square matrix Q is orthogonal, if and only if,

QT Q = QQT = I
and then Q has the following properties
Its eigenvalues are placed on the unit circle.
9.5
Orthogonal matrices
SPECIAL MATRICES
Its eigenvectors are unitary, i.e. have length one.

The inverse of an orthogonal matrix is orthogonal too.
Basic properties for the orthogonal matrix Q
Q1
QT
QQT
QT Q
det(Q)
9.5.1
=
=
=
=
=
QT
Q
I
I
1
Ortho-Sym
A matrix Q+ which simultaneously is orthogonal and symmetric is called an

ortho-sym matrix [20]. Hereby
QT+ Q+
Q+
= I
=
QT+
(407)
(408)
The powers of an ortho-sym matrix are given by the following rule

Qk+
=
=
9.5.2
1 + (1)k
1 + (1)k+1
I+
Q+
2
2
1 + cos(k)
1 cos(k)
I+
Q+
2
2
(409)
(410)
Ortho-Skew
A matrix which simultaneously is orthogonal and antisymmetric is called an

ortho-skew matrix [20]. Hereby
QH
Q
Q
= I
=
QH
(411)
(412)
The powers of an ortho-skew matrix are given by the following rule

Qk
=
=
9.5.3
ik + (i)k
ik (i)k
Ii
Q
2
2
cos(k )I + sin(k )Q
2
2
(413)
(414)
Decomposition
A square matrix A can always be written as a sum of a symmetric A+ and an

antisymmetric matrix A
A = A + + A
(415)
9.6
Positive Definite and Semi-definite Matrices
9.6
9.6.1
SPECIAL MATRICES
Positive Definite and Semi-definite Matrices

Definitions
A matrix A is positive definite if and only if

xT Ax > 0,
x 6= 0
(416)
A matrix A is positive semi-definite if and only if

xT Ax 0,
(417)
Note that if A is positive definite, then A is also positive semi-definite.

9.6.2
Eigenvalues
The following holds with respect to the eigenvalues:

A pos. def.
A pos. semi-def.
9.6.3
eig( A+A
)>0
2
A+AH
eig( 2 ) 0
(418)
Trace
The following holds with respect to the trace:

A pos. def.
A pos. semi-def.
9.6.4
Tr(A) > 0
Tr(A) 0
(419)
Inverse
If A is positive definite, then A is invertible and A1 is also positive definite.

9.6.5
Diagonal
If A is positive definite, then Aii > 0, i

9.6.6
Decomposition I
The matrix A is positive semi-definite of rank r there exists a matrix B of

rank r such that A = BBT
The matrix A is positive definite there exists an invertible matrix B such
that A = BBT
9.6.7
Decomposition II
Assume A is an n n positive semi-definite, then there exists an n r matrix

B of rank r such that BT AB = I.
9.7
Singleentry Matrix, The
9.6.8
Equation with zeros
Assume A is positive semi-definite, then XT AX = 0

9.6.9
SPECIAL MATRICES
AX = 0
Rank of product
Assume A is positive definite, then rank(BABT ) = rank(B)

9.6.10
Positive definite property
If A is n n positive definite and B is r n of rank r, then BABT is positive

definite.
9.6.11
Outer Product
If X is n r, where n r and rank(X) = n, then XXT is positive definite.

9.6.12
Small pertubations
If A is positive definite and B is symmetric, then A tB is positive definite for

sufficiently small t.
9.6.13
Hadamard inequality
If A is a positive definite or semi-definite matrix, then

Y
det(A)
Aii
i
See [15, pp.477]

9.6.14
Hadamard product relation
Assume that P = AAT and Q = BBT are semi positive definite matrices, it
then holds that
P Q = RRT
where the columns of R are constructed as follows: ri+(j1)NA = ai bj , for
i = 1, 2, ..., NA and j = 1, 2, ..., NB . The result is unpublished, but reported by
Pavel Sakov and Craig Bishop.
9.7
9.7.1

Definition
The single-entry matrix Jij Rnn is defined as the matrix which is zero
everywhere except in the entry (i, j) in which it is 1. In a 4 4 example one
9.7
SPECIAL MATRICES
might have
J23
0
0
=
0
0
0
0
0
0
0
1
0
0
0
0
0
0
(420)
The single-entry matrix is very useful when working with derivatives of expressions involving matrices.
9.7.2
Swap and Zeros
Assume A to be n m and Jij to be m p

AJij = 0 0 . . . Ai
... 0
(421)
i.e. an n p matrix of zeros with the i.th column of A in place of the j.th
column. Assume A to be n m and Jij to be p n
0
..
.
(422)
Jij A =
Aj
0
.
..
0
i.e. an p m matrix of zeros with the j.th row of A in the placed of the i.th
row.
9.7.3
Rewriting product of elements

Aki Bjl = (Aei eTj B)kl
(AJij B)kl
(423)
Aik Blj = (AT ei eTj BT )kl
(AT Jij BT )kl
(424)
Aik Bjl =
Aki Blj =
9.7.4
(A ei eTj B)kl
(Aei eTj BT )kl
=
=
ij
(A J B)kl
ij
(AJ B )kl
(425)
(426)
Properties of the Singleentry Matrix
If i = j
Jij Jij = Jij
Jij (Jij )T = Jij
(Jij )T (Jij )T = Jij

(Jij )T Jij = Jij
If i 6= j
Jij Jij = 0
Jij (Jij )T = Jii
(Jij )T (Jij )T = 0
(Jij )T Jij = Jjj
9.8
Symmetric, Skew-symmetric/Antisymmetric
9.7.5
SPECIAL MATRICES
The Singleentry Matrix in Scalar Expressions
Assume A is n m and J is m n, then

Tr(AJij ) = Tr(Jij A) = (AT )ij
(427)
Assume A is n n, J is n m and B is m n, then

Tr(AJij B) = (AT BT )ij
Tr(AJji B) = (BA)ij
Tr(AJij Jij B) = diag(AT BT )ij
(428)
(429)
(430)
Assume A is n n, Jij is n m B is m n, then

xT AJij Bx =
x AJij Jij Bx =
T
9.7.6
(AT xxT BT )ij

diag(AT xxT BT )ij
(431)
(432)
Structure Matrices
The structure matrix is defined by

A
= Sij
Aij
(433)
Sij = Jij
(434)
Sij = Jij + Jji Jij Jij
(435)
If A has no special structure then
If A is symmetric then
9.8
9.8.1
Symmetric, Skew-symmetric/Antisymmetric
Symmetric
The matrix A is said to be symmetric if

A = AT
(436)
Symmetric matrices have many important properties, e.g. that their eigenvalues
are real and eigenvectors orthogonal.
9.8.2
Skew-symmetric/Antisymmetric
The antisymmetric matrix is also known as the skew symmetric matrix. It has
the following property from which it is defined
A = AT
(437)
9.9
Toeplitz Matrices
SPECIAL MATRICES
Hereby, it can be seen that the antisymmetric matrices always have a zero
diagonal. The n n antisymmetric matrices also have the following properties.
det(AT )
det(A)
=
=
det(A) = (1)n det(A)

det(A) = 0, if n is odd
(438)
(439)
The eigenvalues of an antisymmetric matrix are placed on the imaginary axis

and the eigenvectors are unitary.
9.8.3
Decomposition
A square matrix A can always be written as a sum of a symmetric A+ and an

antisymmetric matrix A
A = A+ + A
(440)
Such a decomposition could e.g. be
A=
9.9
A AT
A + AT
+
= A+ + A
2
2
(441)
Toeplitz Matrices
A Toeplitz matrix T is a matrix where the elements of each diagonal is the

same. In the n n square case, it has the following structure:
t0
t1 tn1
t11 t12 t1n
..
..
..
..
t1
t21 . . . . . .
.
.
.
.
(442)
T= .
=
.
.
.
.
.
..
..
..
.. t
..
..
t1
12
t(n1) t1
t0
tn1 t21 t11
A Toeplitz matrix is persymmetric. If a matrix is persymmetric (or orthosymmetric), it means that the matrix is symmetric about its northeast-southwest
diagonal (anti-diagonal) [12]. Persymmetric matrices is a larger class of matrices, since a persymmetric matrix not necessarily has a Toeplitz structure. There
are some special cases of Toeplitz matrices. The symmetric Toeplitz matrix is
given by:
t0
t1 tn1
..
.. ..
.
.
t1
.
T=
(443)
..
.
.
.
.
.
.
.
t1
t(n1) t1
t0
The circular Toeplitz matrix:
t0
TC =
tn
..
.
t1
t1
..
.
..
..
.
..
.
.
tn1
tn1
..
.
t1
t0
(444)
9.10
Transition matrices
The upper triangular Toeplitz matrix:
t0 t1 tn1
..
0 ... ...
.
TU = .
.
.
.. ..
..
t1
0 0
t0
and the lower triangular Toeplitz matrix:
t0
0
.
.. ...
t1
TL =
..
..
..
.
.
.
t(n1) t1
9.9.1
SPECIAL MATRICES
(445)
0
..
.
0
t0
(446)
Properties of Toeplitz Matrices
The Toeplitz matrix has some computational advantages. The addition of two
Toeplitz matrices can be done with O(n) flops, multiplication of two Toeplitz
matrices can be done in O(n ln n) flops. Toeplitz equation systems can be solved
in O(n2 ) flops. The inverse of a positive definite Toeplitz matrix can be found
in O(n2 ) flops too. The inverse of a Toeplitz matrix is persymmetric. The
product of two lower triangular Toeplitz matrices is a Toeplitz matrix. More
information on Toeplitz matrices and circulant matrices can be found in [13, 7].
9.10
Transition matrices
A square matrix P is a transition matrix, also known as stochastic matrix or

probability matrix, if
X
0 (P)ij 1,
(P)ij = 1
j
The transition matrix usually describes the probability of moving from state i
to j in one step and is closely related to markov processes. Transition matrices
have the following properties
Prob[i j in 1 step] = (P)ij
Prob[i j in 2 steps] = (P2 )ij
Prob[i j in k steps] = (Pk )ij
If all rows are identical Pn = P
P = ,
is called invariant
where is a so-called stationary probability vector, i.e., 0 i 1 and
1.
(447)
(448)
(449)
(450)
(451)
P
i =
9.11
Units, Permutation and Shift
9.11
SPECIAL MATRICES
Units, Permutation and Shift
9.11.1
Unit vector
Let ei Rn1 be the ith unit vector, i.e. the vector which is zero in all entries
except the ith at which it is 1.
9.11.2
Rows and Columns

= eTi A
= Aej
i.th row of A
j.th column of A
9.11.3
(452)
(453)
Permutations
Let P be some permutation
0 1
P= 1 0
0 0
matrix, e.g.
0

0 = e2
1
eT2
= eT1
eT3
e1
e3
(454)
For permutation matrices it holds that

PPT = I
and that
(455)
eT2 A
PA = eT1 A
eT3 A
AP =
Ae2
Ae1
Ae3
(456)
That is, the first is a matrix which has columns of A but in permuted sequence
and the second is a matrix which has the rows of A but in the permuted sequence.
9.11.4
Translation, Shift or Lag Operators
Let L denote the lag (or translation

example by
0
1
L=
0
0
or shift) operator defined on a 4 4

0
0
1
0
0
0
0
1
0
0
0
0
(457)
i.e. a matrix of zeros with one on the sub-diagonal, (L)ij = i,j+1 . With some
signal xt for t = 1, ..., N , the n.th power of the lag operator shifts the indices,
i.e.
n 0
for t = 1, .., n
(Ln x)t =
(458)
xtn for t = n + 1, ..., N
9.12
Vandermonde Matrices
SPECIAL MATRICES
A related but slightly different matrix is the recurrent shifted operator defined
on a 4x4 example by
0 0 0 1
= 1 0 0 0
L
(459)
0 1 0 0
0 0 1 0
ij = i,j+1 + i,1 j,dim(L) . On a signal x it has the
i.e. a matrix defined by (L)
effect
n x)t = xt0 , t0 = [(t n) mod N ] + 1
(L
(460)
is like the shift operator L except that it wraps the signal as if it
That is, L
was periodic and shifted (substituting the zeros with the rear end of the signal).
is invertible and orthogonal, i.e.
Note that L
1 = L
T
L
9.12
(461)
Vandermonde Matrices
A Vandermonde matrix has the form [15]
1 v1 v12 v1n1
1 v2 v22 v n1
2
V= . .
..
..
.
.
. .
.
.
1 vn vn2 vnn1
(462)
The transpose of V is also said to a Vandermonde matrix. The determinant is

given by [29]
Y
det V =
(vi vj )
(463)
i>j
10
10
10.1
10.1.1
FUNCTIONS AND OPERATORS
Functions and Operators

Functions and Series
Finite Series
(Xn I)(X I)1 = I + X + X2 + ... + Xn1
10.1.2
(464)
Taylor Expansion of Scalar Function
Consider some scalar function f (x) which takes the vector x as an argument.
This we can Taylor expand around x0
1
f (x)
= f (x0 ) + g(x0 )T (x x0 ) + (x x0 )T H(x0 )(x x0 )
2
where
g(x0 ) =
10.1.3
f (x)

x x0
H(x0 ) =
(465)
2 f (x)

xxT x0
Matrix Functions by Infinite Series
As for analytical functions in one dimension, one can define a matrix function
for square matrices X by an infinite series
f (X) =
cn Xn
(466)
n=0
P
assuming the limit exists and is finite. If the coefficients cn fulfils n cn xn < ,
then one can prove that the above series exists and is finite, see [1]. Thus for
any analytical function f (x) there exists a corresponding matrix function f (x)
constructed by the Taylor expansion. Using this one can prove the following
results:
1) A matrix A is a zero of its own characteristic polynomium [1]:
X
p() = det(I A) =
cn n
p(A) = 0
(467)
n
2) If A is square it holds that [1]

A = UBU1
f (A) = Uf (B)U1
(468)
3) A useful fact when using power series is that

An 0forn
if
|A| < 1
(469)
10.2
Kronecker and Vec Operator
10.1.4
10
Exponential Matrix Function
In analogy to the ordinary scalar exponential function, one can define exponential and logarithmic matrix functions:
eA
X
1
1 n
A = I + A + A2 + ...
n!
2
n=0
(470)
eA
X
1
1
(1)n An = I A + A2 ...
n!
2
n=0
(471)
etA
X
1
1
(tA)n = I + tA + t2 A2 + ...
n!
2
n=0
(472)
X
(1)n1 n
1
1
A = A A2 + A3 ...
n
2
3
n=1
(473)
ln(I + A)
Some of the properties of the exponential function are [1]

eA eB
(eA )1
d tA
e
dt
d
Tr(etA )
dt
det(eA )
10.1.5
= eA+B
= eA
if
AB = BA
= AetA = etA A,
=
tR
Tr(AetA )
= eTr(A)
(478)
Trigonometric Functions
cos(A)
10.2.1
(476)
(477)
X
(1)n A2n+1
1
1
= A A3 + A5 ...
sin(A)
(2n
+
1)!
3!
5!
n=0
10.2
(474)
(475)
X
(1)n A2n
1
1
= I A2 + A4 ...
(2n)!
2!
4!
n=0
(479)
(480)

The Kronecker Product
The Kronecker product of an m n matrix A and an r q matrix B, is an

mr nq matrix, A B defined as
A11 B A12 B ... A1n B

A21 B A22 B ... A2n B
AB=
(481)
..
..
.
.
Am1 B Am2 B
... Amn B
10.2
10
The Kronecker product has the following properties (see [19])

A (B + C)
AB
A (B C)
(A A B B)
(A B)T
(A B)(C D)
(A B)1
(A B)+
rank(A B)
Tr(A B)
det(A B)
{eig(A B)}
{eig(A B)}
eig(A B)
AB+AC
BA
in general
(A B) C
A B (A B)
A T BT
AC BD
A1 B1
A + B+
rank(A)rank(B)
Tr(A)Tr(B) = Tr(A B )
= det(A)rank(B) det(B)rank(A)
= {eig(B A)}
if A, B are square
T
= {eig(A)eig(B) }
if A, B are symmetric and square
= eig(A) eig(B)
=
6=
=
=
=
=
=
=
=
=
(482)
(483)
(484)
(485)
(486)
(487)
(488)
(489)
(490)
(491)
(492)
(493)
(494)
(495)
Where {i } denotes the set of values i , that is, the values in no particular
order or structure, and A denotes the diagonal matrix with the eigenvalues of
A.
10.2.2
The Vec Operator
The vec-operator applied on a matrix A stacks the columns into a vector, i.e.
for a 2 2 matrix
A11

A21
A11 A12
A=
vec(A) =
A12
A21 A22
A22
Properties of the vec-operator include (see [19])
vec(AXB) = (BT A)vec(X)
Tr(AT B) = vec(A)T vec(B)
vec(A + B) = vec(A) + vec(B)
vec(A) = vec(A)
(496)
(497)
(498)
(499)
10.3
Vector Norms
10.3
10.3.1
10
Vector Norms
Examples
||x||1
||x||22
||x||p
|xi |
(500)
= x x
"
#1/p
X
p
=
|xi |
(501)
i
H
(502)
||x||
max |xi |
(503)
Further reading in e.g. [12, p. 52]
10.4
10.4.1
Matrix Norms
Definitions
A matrix norm is a mapping which fulfils

||A|| 0
||A|| = 0 A = 0
||cA|| = |c|||A||,
cR
||A + B|| ||A|| + ||B||
10.4.2
(504)
(505)
(506)
(507)
Induced Norm or Operator Norm
An induced norm is a matrix norm induced by a vector norm by the following

||A|| = sup{||Ax||
| ||x|| = 1}
(508)
where || || ont the left side is the induced matrix norm, while || || on the right
side denotes the vector norm. For induced norms it holds that
||I|| = 1
||Ax|| ||A|| ||x||,
||AB|| ||A|| ||B||,
10.4.3
for all A, x
for all A, B
(509)
(510)
(511)
Examples
||A||1
||A||2
||A||p
||A||
X
max
|Aij |
j
i
q
=
max eig(AH A)
= ( max ||Ax||p )1/p
||x||p =1
X
= max
|Aij |
=
(512)
(513)
(514)
(515)
10.5
Rank
10
||A||F
sX
|Aij |2 =
q
Tr(AAH )
(Frobenius)
(516)
ij
||A||max
max |Aij |
||A||KF
= ||sing(A)||1
(517)
ij
(Ky Fan)
(518)
where sing(A) is the vector of singular values of the matrix A.

10.4.4
Inequalities
E. H. Rasmussen has in yet unpublished material derived and collected the

following inequalities. They are collected in a table as below, assuming A is an
m n, and d = rank(A)
||A||max
||A||max
||A||1
||A||
||A||2
||A||F
||A||KF
m
n
mn
mn
mnd
||A||1
1
||A||
1
m
n
n
n
nd
m
md
||A||2
1
m
n
d
d
||A||F
1
m
n
1
||A||KF
1
m
n
1
1
which are to be read as, e.g.

||A||2
m ||A||
(519)
10.4.5
Condition Number
p
The 2-norm of A equals (max(eig(AT A))) [12, p.57]. For a symmetric, positive definite matrix, this reduces to max(eig(A)) The condition number based
on the 2-norm thus reduces to
kAk2 kA1 k2 = max(eig(A)) max(eig(A1 )) =
10.5
10.5.1
max(eig(A))
.
min(eig(A))
(520)
Rank
Sylvesters Inequality
If A is m n and B is n r, then
rank(A) + rank(B) n rank(AB) min{rank(A), rank(B)}
10.6
(521)
Integral Involving Dirac Delta Functions
Assuming A to be square, then

Z
p(s)(x As)ds =
1
p(A1 x)
det(A)
(522)
10.7
Miscellaneous
10
Assuming A to be underdetermined, i.e. tall, then

)
(
Z
1 T p(A+ x) if x = AA+ x
det(A
A)
p(s)(x As)ds =
0
elsewhere
(523)
See [9].
10.7
Miscellaneous
For any A it holds that

rank(A) = rank(AT ) = rank(AAT ) = rank(AT A)
(524)
It holds that
A is positive definite
B invertible, such that A = BBT
(525)
A
A.1
A.1.1
ONE-DIMENSIONAL RESULTS
One-dimensional Results
Gaussian
Density
(x )2
p(x) =
exp
2 2
2 2

A.1.2
A.1.3
Normalization
Z
(s)2
e 22 ds =
2 2
r

2
Z
b 4ac
(ax2 +bx+c)
e
dx =
exp
a
4a
2

r
Z
c1 4c2 c0
c2 x2 +c1 x+c0
e
dx =
exp
c2
4c2
(526)
(527)
(528)
(529)
Derivatives
p(x)
ln p(x)
p(x)
ln p(x)
A.1.4
(x )
2
(x )
=
2

1 (x )2
= p(x)
2

1 (x )2
=
2
= p(x)
(530)
(531)
(532)
(533)
Completing the Squares

c2 x2 + c1 x + c0 = a(x b)2 + w
a = c2
or
b=
1 c1
2 c2
w=
1 c21
+ c0
4 c2
1
(x )2 + d
2 2
1
c2
2 =
d = c0 1
2c2
4c2
c2 x2 + c1 x + c0 =
=
A.1.5
c1
2c2
Moments
If the density is expressed by

1
(s )2
p(x) =
exp
2 2
2 2
or p(x) = C exp(c2 x2 + c1 x)
(534)
then the first few basic moments are

A.2
One Dimensional Mixture of Gaussians

A ONE-DIMENSIONAL RESULTS
hxi
hx2 i
2 + 2
hx3 i
3 2 + 3
hx4 i
4 + 62 2 + 3 4
c1
2c2

2
c1
1
2c2 + h 2c2
i
c21
c1
3
2
(2c )
2c2
2 4
2
c1
c1
+
6
2c2
2c2
1
2c2
+3
1
2c2
2
and the central moments are

h(x )i
2
h(x ) i
h(x )3 i
4
h(x ) i
=
=
=
=
=
2
0h
=
=
1
2c2
1
2c2
i2
A kind of pseudo-moments (un-normalized integrals) can easily be derived as

2
r
Z
c1
2
n
n
exp(c2 x + c1 x)x dx = Zhx i =
exp
hxn i
(535)
c2
4c2
From the un-centralized moments one can derive other entities like
A.2
A.2.1
hx2 i hxi2
hx3 i hx2 ihxi
=
=
2
2 2
=
=
hx4 i hx2 i2
2 4 + 42 2
c2
1 4 2c12

Density and Normalization
p(s) =
K
X
k
A.2.2
1
2c2
2c1
(2c2 )2
2
(2c2 )2

1 (s k )2
p k exp
2
k2
2k2
(536)
Moments
An useful fact of MoG, is that

hxn i =
k hxn ik
(537)
where hik denotes average with respect to the k.th component. We can calculate
the first four moments from the densities

X
1
1 (x k )2
p(x) =
k p
exp
(538)
2
k2
2k2
k
X

p(x) =
k Ck exp ck2 x2 + ck1 x
(539)
k
as
A.2

A ONE-DIMENSIONAL RESULTS
hx3 i
hx4 i
hxi
2
hx i
k k
2
k k (k
2k )
2
3
k k (3k k + k )
P
4
2 2
4
k k (k + 6k k + 3k )
ck1
2ck2
1
2ck2
i
+
ck1
2ck2
2
h
ii
c2k1
ck1
3
2
k
k
(2ck2 )
2ck2

2
2
P
c2k1
ck1
1
6 2ck2 + 3
k k
2ck2
2ck2
h
If all the gaussians are centered, i.e. k = 0 for all k, then

hxi
hx i
hx3 i
=
=
hx i
0
P
2
k k k
0
P
4
k k 3k
=
=
=
=
0
P
k k
0
P
k k 3
1
2ck2
1
2ck2
i2
From the un-centralized moments one can derive other entities like
2

P
2
hx2 i hxi2
=
0 k k 0 k + k k k 0
k,k
2

P
3
2
2
hx3 i hx2 ihxi =
k,k0 k k0 3k k + k (k + k )k0

P
4
2 2
4
2
2
2
2
hx4 i hx2 i2
=
k,k0 k k0 k + 6k k + 3k (k + k )(k0 + k0 )
A.2.3
Derivatives
P
Defining p(s) = k k Ns (k , k2 ) we get for a parameter j of the j.th component
j Ns (j , j2 ) ln(j Ns (j , j2 ))
ln p(s)
=P
(540)
2
j
j
k k Ns (k , k )
that is,
ln p(s)
j
ln p(s)
j
ln p(s)
j
j Ns (j , j2 ) 1
P
2
k k Ns (k , k ) j
(541)
j Ns (j , j2 ) (s j )
P
2
j2
k k Ns (k , k )
"
#
j Ns (j , j2 ) 1 (s j )2
P
1
2
j2
k k Ns (k , k ) j
(542)
(543)
Note thatP
k must be constrained to be proper ratios. Defining the ratios by
j = erj / k erk , we obtain
ln p(s) X ln p(s) l
=
rj
l rj
l
where
l
= l (lj j )
rj
(544)
B
B.1
B.1.1
PROOFS AND DETAILS
Proofs and Details

Misc Proofs
Proof of Equation 83
Essentially we need to calculate

(Xn )kl
Xij
Xij
Xk,u1 Xu1 ,u2 ...Xun1 ,l
u1 ,...,un1
= k,i u1 ,j Xu1 ,u2 ...Xun1 ,l

+Xk,u1 u1 ,i u2 ,j ...Xun1 ,l
..
.
+Xk,u1 Xu1 ,u2 ...un1 ,i l,j
=
n1
X
(Xr )ki (Xn1r )jl
r=0
n1
X
(Xr Jij Xn1r )kl
r=0
Using the properties of the single entry matrix found in Sec. 9.7.4, the result
follows easily.
B.1.2
Details on Eq. 546
det(XH AX)
= det(XH AX)Tr[(XH AX)1 (XH AX)]

= det(XH AX)Tr[(XH AX)1 ((XH )AX + XH (AX))]
= det(XH AX) Tr[(XH AX)1 (XH )AX]

+Tr[(XH AX)1 XH (AX)]
=
det(XH AX) Tr[AX(XH AX)1 (XH )]

+Tr[(XH AX)1 XH A(X)]
First, the derivative is found with respect to the real part of X

det(XH AX)
<X
Tr[AX(XH AX)1 (XH )]

det(XH AX)
<X
H
1 H
Tr[(X AX) X A(X)]
+
<X

= det(XH AX) AX(XH AX)1 + ((XH AX)1 XH A)T
B.1
Misc Proofs
PROOFS AND DETAILS
Through the calculations, (92) and (219) were used. In addition, by use of (220),
the derivative is found with respect to the imaginary part of X
i
det(XH AX)
=X
Tr[AX(XH AX)1 (XH )]

= i det(XH AX)
=X
Tr[(XH AX)1 XH A(X)]
+
=X

= det(XH AX) AX(XH AX)1 ((XH AX)1 XH A)T
Hence, derivative yields

det(XH AX)
X
=
=
1 det(XH AX)
det(XH AX)
i
2
<X
=X
T
H
H
1 H
det(X AX) (X AX) X A
and the complex conjugate derivative yields

det(XH AX)
X
=
=
1 det(XH AX)
det(XH AX)
+i
2
<X
=X
H
H
1
det(X AX)AX(X AX)
Notice, for real X, A, the sum of (228) and (229) is reduced to (48).
Similar calculations yield
det(XAXH )
X
=
=
1 det(XAXH )
det(XAXH )
i
2
<X
=X

H
H
H 1 T
det(XAX ) AX (XAX )
(545)
1 det(XAXH )
det(XAXH )
+i
2
<X
=X
det(XAXH )(XAXH )1 XA
(546)
and
det(XAXH )
X
=
=
REFERENCES
REFERENCES
References
[1] Karl Gustav Andersson and Lars-Christer Boiers. Ordinaera differentialekvationer. Studenterlitteratur, 1992.
[2] J
orn Anem
uller, Terrence J. Sejnowski, and Scott Makeig. Complex independent component analysis of frequency-domain electroencephalographic
data. Neural Networks, 16(9):13111323, November 2003.
[3] S. Barnet. Matrices. Methods and Applications. Oxford Applied Mathematics and Computin Science Series. Clarendon Press, 1990.
[4] Christopher Bishop. Neural Networks for Pattern Recognition. Oxford
University Press, 1995.
[5] Robert J. Boik. Lecture notes: Statistics 550. Online, April 22 2002. Notes.
[6] D. H. Brandwood. A complex gradient operator and its application in
adaptive array theory. IEE Proceedings, 130(1):1116, February 1983. PTS.
F and H.
[7] M. Brookes. Matrix Reference Manual, 2004. Website May 20, 2004.
[8] Contradsen K., En introduktion til statistik, IMM lecture notes, 1984.
[9] Mads Dyrholm. Some matrix results, 2004. Website August 23, 2004.
[10] Nielsen F. A., Formula, Neuro Research Unit and Technical university of
Denmark, 2002.
[11] Gelman A. B., J. S. Carlin, H. S. Stern, D. B. Rubin, Bayesian Data
Analysis, Chapman and Hall / CRC, 1995.
[12] Gene H. Golub and Charles F. van Loan. Matrix Computations. The Johns
Hopkins University Press, Baltimore, 3rd edition, 1996.
[13] Robert M. Gray. Toeplitz and circulant matrices: A review. Technical
report, Information Systems Laboratory, Department of Electrical Engineering,Stanford University, Stanford, California 94305, August 2002.
[14] Simon Haykin. Adaptive Filter Theory. Prentice Hall, Upper Saddle River,
NJ, 4th edition, 2002.
[15] Roger A. Horn and Charles R. Johnson. Matrix Analysis. Cambridge
University Press, 1985.
[16] Mardia K. V., J.T. Kent and J.M. Bibby, Multivariate Analysis, Academic
Press Ltd., 1979.
[17] Mathpages on Eigenvalue Problems and Matrix Invariants,
REFERENCES
REFERENCES
http://www.mathpages.com/home/kmath128.htm
[18] Carl D. Meyer. Generalized inversion of modified matrices. SIAM Journal
of Applied Mathematics, 24(3):315323, May 1973.
[19] Thomas P. Minka. Old and new matrix algebra useful for statistics, December 2000. Notes.
[20] Daniele Mortari OrthoSkew and OrthoSym Matrix Trigonometry John
Lee Junkins Astrodynamics Symposium, AAS 03265, May 2003. Texas
A&M University, College Station, TX
[21] L. Parra and C. Spence. Convolutive blind separation of non-stationary
sources. In IEEE Transactions Speech and Audio Processing, pages 320
327, May 2000.
[22] Kaare Brandt Petersen, Jiucang Hao, and Te-Won Lee. Generative and
filtering approaches for overcomplete representations. Neural Information
Processing - Letters and Reviews, vol. 8(1), 2005.
[23] John G. Proakis and Dimitris G. Manolakis. Digital Signal Processing.
Prentice-Hall, 1996.
[24] Laurent Schwartz. Cours dAnalyse, volume II. Hermann, Paris, 1967. As
referenced in [14].
[25] Shayle R. Searle. Matrix Algebra Useful for Statistics. John Wiley and
Sons, 1982.
[26] G. Seber and A. Lee. Linear Regression Analysis. John Wiley and Sons,
2002.
[27] S. M. Selby. Standard Mathematical Tables. CRC Press, 1974.
[28] Inna Stainvas. Matrix algebra in differential calculus. Neural Computing
Research Group, Information Engeneering, Aston University, UK, August
2002. Notes.
[29] P. P. Vaidyanathan. Multirate Systems and Filter Banks. Prentice Hall,
1993.
[30] Max Welling. The Kalman Filter. Lecture Note.
[31] Wikipedia on minors: Minor (linear algebra),
http://en.wikipedia.org/wiki/Minor_(linear_algebra)
[32] Zhaoshui He, Shengli Xie, et al, Convolutive blind source separation in
frequency domain based on sparse representation, IEEE Transactions on
Audio, Speech and Language Processing, vol.15(5):1551-1563, July 2007.
Index
Anti-symmetric, 49
Block matrix, 41
Chain rule, 14
Cholesky-decomposition, 28
Co-kurtosis, 29
Co-skewness, 29
Condition number, 61
Cramers Rule, 57
Derivative of a complex matrix, 22
Derivative of a determinant, 7
Derivative of a trace, 11
Derivative of an inverse, 8
Derivative of symmetric matrix, 15
Derivatives of Toeplitz matrix, 15
Dirichlet distribution, 33
Eigenvalues, 26
Eigenvectors, 26
Exponential Matrix Function, 55
Nilpotent, 44
Norm of a matrix, 59
Norm of a vector, 59
Normal-Inverse Gamma distribution, 33
Normal-Inverse Wishart distribution, 34
Orthogonal, 44
Power series of matrices, 54
Probability matrix, 51
Pseudo-inverse, 20
Schur complement, 36, 42
Single entry matrix, 47
Singular Valued Decomposition (SVD),
26
Skew-Hermitian, 43
Skew-symmetric, 49
Stochastic matrix, 51
Student-t, 32
Sylvesters Inequality, 61
Symmetric, 49
Gaussian, conditional, 36
Gaussian, entropy, 40
Gaussian, linear combination, 36
Gaussian, marginal, 35
Gaussian, product of densities, 37
Generalized inverse, 20
Taylor expansion, 54
Toeplitz matrix, 50
Transition matrix, 51
Trigonometric functions, 55
Hermitian, 43
Vandermonde matrix, 53
Vec operator, 55, 56
Unipotent, 44
Idempotent, 44
Kronecker product, 55
Wishart distribution, 33
Woodbury identity, 17
LDL decomposition, 28
LDM-decomposition, 28
Linear regression, 56
LU decomposition, 28
Lyapunov Equation, 59
Moore-Penrose inverse, 20
Multinomial distribution, 33
71

Matrix Cookbook

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Matrix Cookbook

Hochgeladen von

Copyright:

Verfügbare Formate

The Matrix Cookbook

6 Statistics and Probability

10 Functions and Operators

Notation and Nomenclature

Hadamard (elementwise) product

The null matrix. Zero in all entries.

Trace and Determinants

The Special Case 2x2

Consider the matrix A

Determinant and trace

Tr(A) = A11 + A22

The Special Case 2x2

This section is covering differentiation of a number of expressions with respect to

If X is square and invertible, then

If X is not square but A is symmetric, then

If X is not square and A is not symmetric, then

Other nonlinear forms

Some special cases are (See [9, 7])

From [27] we have the basic identity

from which it follows

= (X1 )ki (X1 )jl

= ((X + A)1 (X + A)1 )T

From [32] we have the following result: Let A be an n n invertible square

Derivatives of Matrices, Vectors and Scalar Forms

Derivatives of Matrices, Vectors and Scalar Forms

= lj (XT B)ki + kj (BX)il

(Jij )kl = ik jl (72)

(Xb + c)T D(Xb + c)

Assume W is symmetric, then

(x As)T W(x As) = 2AT W(x As)

(x s)T W(x s) = 2W(x s)

(x s)T W(x s) = 2W(x s)

(x As)T W(x As) = 2W(x As)

(x As)T W(x As) = 2W(x As)sT

As a case with complex values the following holds

This formula is also known from the LMS algorithm [14]

Higher order and non-linear

For proof of the above, see B.1.1.

Xn1r abT (Xn )T Xr

+(Xr )T Xn abT (Xn1r )T

See B.1.1 for a proof.

Gradient and Hessian

Assume F (X) to be a differentiable function of each of the elements of X. It

Tr (AXB + C)(AXC + C)T

2AT (AXB + C)BT

Derivatives of vector norms

Tr(AX1 B) = (X1 BAX1 )T = XT AT BT XT

Tr (XT CX)1 A = (CX(XT CX)1 )(A + AT )(XT CX)1 (114)

Tr (XT CX)1 (XT BX) = 2CX(XT CX)1 XT BX(XT CX)1

Tr (A + XT CX)1 (XT BX) = 2CX(A + XT CX)1 XT BX(A + XT CX)1

Derivatives of vector norms

Derivatives of matrix norms

For more on matrix norms, see Sec. 10.4.

Derivatives of Structured Matrices

Derivatives of Structured Matrices

The matrix differentiated with respect to itself is in this document referred to

The Chain Rule

Sometimes the objective is to find the derivative of a matrix which is a function

Using matrix notation, this can be written as:

Derivatives of Structured Matrices

If A is symmetric, then Sij = Jij + Jji Jij Jij and therefore

That is, e.g., ([5]):

= A + AT (A I), see (131)

det(X)(2X1 (X1 I))