Closed-form expressions of the eigen decomposition of 2

x 2 and 3 x 3 Hermitian matrices

Charles-Alban Deledalle, Loic Denis, Sonia Tabti, Florence Tupin

To cite this version:

Charles-Alban Deledalle, Loic Denis, Sonia Tabti, Florence Tupin. Closed-form expressions of the
eigen decomposition of 2 x 2 and 3 x 3 Hermitian matrices. [Research Report] Université de Lyon.
2017. �hal-01501221�

HAL Id: hal-01501221

[Link]
Submitted on 4 Apr 2017

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Closed-form expressions of the eigen decomposition
of 2 × 2 and 3 × 3 Hermitian matrices
Charles-Alban Deledalle, Loı̈c Denis, Sonia Tabti, Florence Tupin
April 3, 2017

Abstract
The eigen decomposition of covariance matrices is at the core of many data analysis tech-
niques. The study of 2-components or 3-components vector fields typically requires comput-
ing numerous eigen decompositions of 2 × 2 or 3 × 3 matrices. This is, for example, the
case in the analysis of interferometric or polarimetric SAR images, see MuLoG algorithm
([Link] The closed-form expression of eigen-
values and eigenvectors then provides a way to derive faster data processing algorithms. This
note gives these expressions in the general case (special cases where some coefficients are zero,
or the eigenvalues are not separated may not be covered and then require either to introduce
a small perturbation of the initial matrix or to derive other expressions).

1 Formulæ for 2 × 2 Hermitian matrices

We consider the Hermitian matrix C defined by:
 ∗ ∗
a c∗
    
v1,1 v2,1 λ1 0 v1,1 v1,2
C= = ∗ ∗ (1)
c b v1,2 v2,2 0 λ2 v2,1 v2,2

where a and b are real valued, c is complex valued and c∗ is the complex conjugate of c.

Eigenvalues: The eigenvalues of C are given by


λ1 = (a + b − δ)/2
(2)
λ2 = (a + b + δ)/2
p
with δ = 4|c|2 + (a − b)2 .

Eigenvectors: The eigenvectors of C are given by



 v1,1 = (a − b + δ)/(2c) = (λ2 − b)/c
 v1,2 = 1


(3)
v2,1 = (a − b − δ)/(2c) = (λ1 − b)/c




v2,2 = 1


Note that if c = 0, the matrix C is already diagonal so the eigenvalues are a and b and the
corresponding eigenvectors are v1 = (1, 0)t and v2 = (0, 1)t .

1
Changing the eigenvalues: Let’s apply a function F : R → R on the eigenvalues of C. We
get a matrix C̃ defined by:

ã c̃∗
     ∗ ∗

v v2,1 F(λ1 ) 0 v1,1 v1,2
C̃ ≡ = 1,1 ∗ ∗ . (4)
c̃ b̃ v1,2 v2,2 0 F(λ2 ) v2,1 v2,2

For example, when F : x 7→ log(x), C̃ is called the matrix logarithm of matrix C, or when
F : x 7→ exp(x), C̃ is called the exponential of matrix C.
Based on the previous results, we can derive the expression of C̃:
 h i


 ã = (a − b + δ)λ̃ 2 − (a − b − δ) λ̃1 /(2δ) ,
h i
b̃ = (b − a + δ)λ̃ 2 − (b − a − δ) λ̃1 /(2δ) , (5)


c̃ = c(λ̃2 − λ̃1 )/δ ,


with λ̃1 = F(λ1 ) and λ̃2 = F(λ2 ).

2 Formulæ for 3 × 3 Hermitian matrices

We now consider the 3 × 3 Hermitian matrix C defined by:
 ∗ ∗ ∗
a d∗ f ∗
    
v1,1 v2,1 v3,1 λ1 0 0 v1,1 v1,2 v1,3
∗ ∗ ∗ 
C =  d b e∗  = v1,2 v2,2 v3,2   0 λ2 0  v2,1 v2,2 v2,3 (6)
∗ ∗ ∗
f e c v1,3 v2,3 v3,3 0 0 λ3 v3,1 v3,2 v3,3

where a, b and c are real-valued, d, e and f are complex-valued.

Eigenvalues: The eigenvalues of C are given by

  √ 
 λ1 = a + b + c − 2 √x1 cos(ϕ/3) /3 ,
λ2 = a + b + c + 2 x1 cos[(ϕ − π)/3] /3 , (7)
√
λ3 = a + b + c + 2 x1 cos[(ϕ + π)/3] /3 ,


with

 x1 = a2 + b2 + c2 − ab − ac − bc + 3(|d|2 + |f |2 + |e|2 )
x2 = −(2a − b − c)(2b − a − c)(2c − a − b) (8)
+9 (2c − a − b)|d|2 + (2b − a − c)|f |2 + (2a − b − c)|e|2 − 54 <(d∗ e∗ f )
 

and
 √ 
4x31 −x22
atan

 x2 if x2 > 0


ϕ = π/2 if x2 = 0 (9)
 √ 
4x31 −x22


atan +π if x2 < 0
x2


Eigenvectors: The eigenvectors of C are given by:

     
(λ1 − c − e · m1 )/f (λ2 − c − e · m2 )/f (λ3 − c − e · m3 )/f
v1 =  m1  , v2 =  m2  and v3 =  m3 ,
1 1 1
(10)

2
with
 d(c−λ1 )−e∗ f
 m1 =
 f (b−λ1 )−d e ,
d(c−λ2 )−e∗ f
m2 = f (b−λ2 )−d e , (11)
d(c−λ3 )−e∗ f

m3 = .

f (b−λ3 )−d e

(special attention should be paid to cases where f = 0, or f (b − λ1 ) − d e = 0, f (b − λ2 ) − d e = 0

or f (b − λ3 ) − d e = 0)

Changing the eigenvalues: Let’s apply a function F : R → R on the eigenvalues of C. We

get a matrix C̃ defined by:

ã d˜∗ f˜∗
     ∗ ∗ ∗

v1,1 v2,1 v3,1 F(λ1 ) 0 0 v1,1 v1,2 v1,3
C̃ =  d˜ b̃ ẽ∗  = v1,2 v2,2 v3,2   0 F(λ2 ) ∗
0  v2,1 ∗
v2,2 ∗ 
v2,3 . (12)
∗ ∗ ∗
f˜ ẽ c̃ v1,3 v2,3 v3,3 0 0 F(λ3 ) v3,1 v3,2 v3,3

Based on the previous results, we can derive the expression of C̃:




 ã = λ̃1 |λ1 − c − e m1 |2 + λ̃2 |λ2 − c − e m2 |2 + λ̃3 |λ3 − c − e m3 |2 /|f |2 ,
b̃ = λ̃1 |m1 |2 + λ̃2 |m2 |2 + λ̃3 |m3 |2 ,





c̃ = λ̃
 1 + λ̃2 + λ̃3 ,

˜ = λ̃1 m1 (λ1 − c − e m1 )∗ + λ̃2 m2 (λ2 − c − e m2 )∗ + λ̃3 m3 (λ3 − c − e m3 )∗ /f ∗ ,
 (13)

 d
∗ ∗ ∗

 ẽ = λ̃1 m1 + λ̃2 m2 + λ̃3∗ m3 ,



f˜ = λ̃1 (λ1 − c − e m1 ) + λ̃2 (λ2 − c − e m2 )∗ + λ̃3 (λ3 − c − e m3 )∗ /f ∗ ,
 

with λ̃1 = F(λ1 )/n1 and λ̃2 = F(λ2 )/n2 and λ̃3 = F(λ3 )/n3 and

 n1 = 1 + |m1 |2 + |λ1 − c − e m1 |2 /|f |2 ,
n2 = 1 + |m2 |2 + |λ2 − c − e m2 |2 /|f |2 , (14)
n3 = 1 + |m3 |2 + |λ3 − c − e m3 |2 /|f |2 .
