Koszul/Souriau Fisher Metric Spaces & Optimization by Maximum - - PowerPoint PPT Presentation

koszul souriau fisher metric spaces optimization by
SMART_READER_LITE
LIVE PREVIEW

Koszul/Souriau Fisher Metric Spaces & Optimization by Maximum - - PowerPoint PPT Presentation

Koszul/Souriau Fisher Metric Spaces & Optimization by Maximum Entropy: Hessian Information Geometry, Lie Group Thermodynamics & Poincar'Marle'Souriau Equation Frdric


slide-1
SLIDE 1
  • Thales Air Systems Date

Koszul/Souriau Fisher Metric Spaces & Optimization by Maximum Entropy: Hessian Information Geometry, Lie Group Thermodynamics & Poincaré'Marle'Souriau Equation

Frédéric BARBARESCO, THALES AIR SYSTEMS Senior Scientist & Advanced Studies Manager Advanced Radar Concepts Dept. 21/11/2014

slide-2
SLIDE 2

2 / 2 /

Preamble

  • Problem 1: How to define density of probability for covariance

matrices of stationary time series (THPD matrix: Toeplitz Hermitian Positive Definite matrix)

Problem 2: How to define « Ordered Statistics » for covariance

matrices, knowing that there is no « total orders » for these matrices

Thales Air Systems Date

( )

Toeplitz Definite Positive , det Hermitian ?? ) / ( ξ ξ ξ ξ ξ ξ ξ > = =

+

p

n

ξ ξ ξ ξ ξ ξ ξ ≤ ≤ ≤ > − ⇒ < .... : Order Global no But Definite Positive : Order Local

2 1 1 2 2 1

slide-3
SLIDE 3

3 / 3 /

Preamble

Solution to Problem 1: Koszul/Souriau Solution of Maximum Entropy Entropy as Legendre Transform of Generalized Characteristic Function (Laplace

Transform on Convex Cone with Inner Product given by Cartan-Killing form)

Density of Probability as Souriau Covariant Solution of Maximum Entropy Solution to Problem 2: Frechet Median Barycenter Metric given by Koszul Hessian Geometry (hessian of Entropy) & Souriau Lie Group

Thermodynamics (metric defined by symplectic cocycle and Geometric temperature)

Geometric Median by Fréchet barycenter in Metric space, solved by Karcher Flow

Thales Air Systems Date

( ) ( )

− −

Θ − Θ −

=

*

  • ξ,

ξ,

dξ e e p

ξ ξ ξ ξ

1 1

) (

) (

1 ξ −

Θ = x dx x d x ) ( ) ( Φ = Θ = ξ ξ ξ ξ ξ

ξ

d p

=

*

) ( .

Ω −

− = Φ

*

,

log ) ( ξ

ξ d

e x

x

( )

) (

,

y xad

ad Tr y x

θ

=

=

=

n i i median

d Min Arg

1

) , ( ξ ξ ξ

ξ 2 , 2 2 2

*

log ) ( log ) ( x d e x p E x I

x x

∂ ∂ =       ∂ ∂ − =

Ω −

ξ ξ

ξ ξ

slide-4
SLIDE 4

4 / 4 /

Thales Air Systems Date

Koszul'Vinberg Characteristic Function

François Massieu in 1869 demonstrated that some thermal

properties of physical systems could be derived from “characteristic functions”.

This idea was developed by Gibbs and Duhem with the notion of

potentials in thermodynamics, and introduced by Poincaré in probability.

We will study generalization of this concept by Jean-Louis Koszul in Mathematics Jean-Marie Souriau in Statistical Physics. The Koszul-Vinberg Characteristic Function (KVCF) on convex

cones will be presented as cornerstone of “Information Geometry” theory:

defining Koszul Entropy as Legendre transform of minus the logarithm of KVCF

(their gradients defining mutually inverse diffeomorphisms)

Fisher Information Metrics as hessian of these dual functions. Koszul proved that these metrics are invariant by all

automorphisms of the convex cones.

slide-5
SLIDE 5

5 / 5 /

Paul LEVY (general use of characteristic function in Probability) Henri POINCARE (Introduction of characteristic function Ψ in Probability)

ψ φ

φ

log

  • r

= = e ψ

1869 François MASSIEU (introduction of characteristic function in Thermodynamic: Gibbs-Duhem Potentials)

( )

T T S / 1 . 1 ∂ ∂ − = φ φ

« je montre, dans ce mémoire, que toutes les propriétés d’un corps peuvent se déduire d’une fonction unique, que j’appelle la fonction caractéristique de ce corps»

Koszul Characteristic: Massieu/Poincaré/Levy/Balian

Roger BALIAN (metric for quantum states by hessian metric from Von- Neumann Entropy)

Roger Balian, 1986 DISSIPATION IN MANY-BODY SYSTEMS: A GEOMETRIC APPROACH BASED ON INFORMATION THEORY

X D X F D S ˆ , ˆ ) ˆ ( ) ˆ ( − = X Tr X F ˆ exp ln ) ˆ ( =

[ ]

D d D d Tr S d ds ˆ ln . ˆ

2 2

= − =

slide-6
SLIDE 6

6 / 6 /

Thermodynamic Duhem'Massieu Potentials

  • Duhem P., « Sur les équations générales de la

Thermodynamique », Annales Scientifiques de l’Ecole Normale Supérieure, 3e série, tome VIII, p. 231, 1891

“Nous avons fait de la Dynamique un cas particulier de la Thermodynamique, une

Science qui embrasse dans des principes communs tous les changements d’état des corps, aussi bien les changements de lieu que les changements de qualités physiques “

four scientists were credited by Duhem with having carried out

“the most important researches on that subject”:

  • F. Massieu had managed to derive Thermodynamics from a “characteristic

function and its partial derivatives”

J.W. Gibbs had shown that Massieu’s functions “could play the role of potentials

in the determination of the states of equilibrium” in a given system.

  • H. von Helmholtz had put forward “similar ideas”
  • A. von Oettingen had given “an exposition of Thermodynamics of remarkable

generality” based on general duality concept in “Die thermodynamischen Beziehungen antithetisch entwickelt“, St. Petersburg 1885

W TS E G + − = Ω ) (

slide-7
SLIDE 7

7 / 7 /

Thales Air Systems Date

Influence of Massieu on Poincaré

[M. Massieu showed that, if we make choice for independent variables of v and T or of p and T, there is a function, moreover unknown, of which three functions of variables, p, U and S in the first case, v, U and S in the second, can be deducted

  • easily. M. Massieu gave to this function, the form
  • f which depends on the choice of variables, name
  • f characteristic function.]

[Because from functions of M. Massieu, we can deduct the other functions of variables, all the equations of the Thermodynamics can be written not so as to contain more than these functions and their derivatives; it will thus result from it, in certain cases, a large simplification. We shall see soon an important application of these functions.]

2nd edition of Poincaré Lecture on « Thermodynamics »

slide-8
SLIDE 8

8 / 8 /

1908

Massieu Characteristic Function by Henri Poincaré

slide-9
SLIDE 9

9 / 9 /

1912

Characteristic Function by Henri Poincaré in Probability

  • H. Poincaré has introduced « Characteristic Function » in

Probability with LAPLACE TRANSFORM not with FOURIER TRANSFORM

slide-10
SLIDE 10

10 / 10 /

Thales Air Systems Date

Koszul'Vinberg Characteristic Function

Jean-Marie Souriau has extended the Characteristic Function in

Statistical Physics:

looking for other kinds of invariances through co-adjoint action of a group on its

momentum space

defining physical observables like energy, heat and momentum as pure

geometrical objects.

In covariant Souriau model, Gibbs equilibriums states are indexed

by a geometric parameter, the Geometric Temperature, with values in the Lie algebra of the dynamical Galileo/Poincaré groups, interpreted as a space-time vector (a vector valued temperature of Planck), giving to the metric tensor a null Lie derivative.

Fisher Information metric appears as the opposite of the derivative

  • f Mean “Moment map” by geometric temperature, equivalent to a

Geometric Capacity or Specific Heat.

We will synthetize the analogies between both Koszul and Souriau

models, and will reduce their definitions to the exclusive “Inner Product” selection using symmetric bilinear “Cartan-Killing form” (introduced by Elie Cartan in 1894).

slide-11
SLIDE 11

11 / 11 /

Elie Cartan by Henri Poincaré

Thales Air Systems Date

« A l’Exception d’Henri Poincaré qui écrivit peu avant sa mort un rapport sur les travaux d’Elie Cartan à l’occasion de la candidature de celui-ci à la Sorbonne, les mathématiciens français ne voyaient pas l’importance de l’œuvre. » Paulette Libermann La géométrie différentielle d’Elie Cartan à Charles Ehresmann et André Lichnerowicz Géométrie au XXième siècle, HERMANN, 2005

slide-12
SLIDE 12

12 / 12 /

Cartanian Filiation: J.L Koszul & J.M. Souriau

Thales Air Systems Date

2 2 2 2

log ) ( log x (x) x p E

x

∂ ∂ =       ∂ ∂ −

ψ ξ

ξ

Ω ∈ ∀ = ∫

Ω − Ω

x d e x

x

) (

*

,

ξ ψ

ξ

( ) ( )

) ( . , ,

2 2 1 2 1

1 Z

ad Q Z Z f Z Z f

Z

+ =

β

[ ] ( ) ( ) ()

( ) [

]

,. . Im , , , , ,

2 1 2 1 2 1

β β

β β β

= ∈ ∀ ∈ ∀ = ad Z Z Z Z f Z Z g

∂ ∂ − = ∈ ∈

Q Q C f Ker e Temperatur

Capacity Heat

, , : ) ( β β β

β

∈ − −

=

*

  • ξ

x x x

d e e p ξ ξ

ξ ξ , ,

/ ) ( G e D f to associated cocycle with ) )( ( θ θ =

( )

[ ] ( )

Map Moment Souriau with , , . ) ( ), (

2 1 2 1 , 2 , 1

µ µ ξ ξ σ Z Z f Z Z Z Z

M M

+ =

( )

( )

Involution Cartan , where ) , ( with ) ( , , g ad ad Tr y x B y x B y x

y x

∈ = − = η η β ψ ∂ ∂ − = ∂ ∂ =

Q x (x) I Fisher

2 2 log

Koszul Forms Koszul Characteristic Function Koszul Hessian Metric Jean-Louis Koszul

[ ] [ ] ( ) [ ] ( )

2 1 2 1

, , , , , Z Z f Z Z g β β β

β β

= (x) d

= ψ α log

= = ψ α log

2

d D g

Jean-Marie Souriau Souriau Moment Map Souriau Geometric Temperature/Heat Capacity Souriau/Fisher Metric from Symplectic Cocycle Elie CARTAN

Leçons sur les invariants intégraux, Hermann, 1922 La théorie des groupes finis et continus et la géométrie différentielle (Written by J. Leray)

slide-13
SLIDE 13

13 / 13 /

Jean'Marie Souriau and his Lie Group Thermodynamics

Thales Air Systems Date

J.M. Souriau, “Sur la Stabilité des Avions,” ONERA Publ., 62, vi+94, 1953 Engines could be positionned everywhere and a stable command could be defined J.M. Souriau, Calcul linéaire, P.U.F., Paris, 1964. Multilinear Algebra Le Verrier-Souriau Algorithm Computation of Matrix Characteristic Equation J.M. Souriau, Structure des systèmes dynamiques, Dunod, Paris, 1970 Symplectic Geometry Structure of Classical & Quantum Mechanics

  • Moment Map
  • Geometric Noether

Theorem

  • General Barycentric

Theorem

  • Mass = Symplectic

cohomology of the action of the Galilean group (not for Poincaré Group in Relativity) J.M. Souriau, Définition covariante des équilibres thermodynamiques, Supp. Nuov. Cimento,1,4, p.203-216, 1966 J.M. Souriau, Thermodynamique et géométrie. In diff. Geo. methods in mathematical physics, II,

  • vol. 676 of Lecture Notes in Math., pages 369-397. Springer, 1978

Lie Group Thermodynamics (Gibbs Equilibrium is not covariant by Gallileo/Poincaré Groups)

  • Geometric Temperature (Vector in Lie Algebra of Dynamical Group) & Geometric Entropy
  • FISHER METRIC DEFINED THROUGH SYMPLECTIC COCYCLE OF DYNAMICAL GROUP
  • FISHER METRIC = GEOMETRIC HEAT CAPACITY

AIRBUS BOEING J.M. Souriau, Les groupes comme universaux, Géométrie au XXième siècle, Hermann, 2005 J.M. Souriau, Grammaire de la Nature, 2007 Souriau theorem revisited by AIRBUS/BOEING

« La masse totale d’un système dynamique isolé est la classe de cohomologie du défaut d’équivariance de l’application moment »

slide-14
SLIDE 14

14 / 14 /

Souriau Books

Thales Air Systems Date

http://www.jmsouriau.com/structure_des_systemes_dynamiques.htm Chapter 4 on « Statistical Mechanics » http://www.jmsouriau.com/Publications/JMSouriau-SSD-Ch4.pdf

slide-15
SLIDE 15

15 / 15 /

Souriau Model for Gaussian Law

Thales Air Systems Date

J.M. Souriau, Structure des systèmes dynamiques, Chapitre 4 « Mécanique Statistique »

Souriau « Geometric Temperature » idea come from his book « Calcul Linéaire » (chap. on «Multilinear Algebra »

slide-16
SLIDE 16

16 / 16 /

Souriau Mechanical Statistics

Thales Air Systems Date

J.M. Souriau, Structure des systèmes dynamiques, Chapitre 4 « Mécanique Statistique »

slide-17
SLIDE 17

17 / 17 /

Covariant Souriau Statistical Mechanics

Thales Air Systems Date

J.M. Souriau, Structure des systèmes dynamiques, Chapitre 4 « Mécanique Statistique »

Classical Gibbs Equilibrium is not covariant according to Dynamic Group of Mechanics (Gallileo Group and Poincaré Group) !!!

slide-18
SLIDE 18

18 / 18 /

!"

Hirohiko Shima Book, « Geometry of Hessian

Structures », world Scientific Publishing 2007, dedicated to Jean-Louis Koszul

Hirohiko Shima Keynote Talk at GSI’13 http://www.see.asso.fr/file/5104/download/9914

  • Prof. M. Boyom tutorial :

http://repmus.ircam.fr/_media/brillouin/ressources/une

  • source-de-nouveaux-invariants-de-la-geometrie-de-l-

information.pdf

Jean-Louis Koszul J.L. Koszul, « Sur la forme hermitienne canonique des espaces homogènes complexes », Canad. J. Math. 7, pp. 562-576., 1955 J.L. Koszul, « Domaines bornées homogènes et orbites de groupes de transformations affines », Bull. Soc. Math. France 89, pp. 515-533., 1961 J.L. Koszul, « Ouverts convexes homogènes des espaces affines », Math. Z. 79, pp. 254-259., 1962 J.L. Koszul, « Variétés localement plates et convexité », Osaka J. Maht. 2,

  • pp. 285-290., 1965

J.L. Koszul, « Déformations des variétés localement plates », .Ann Inst Fourier, 18 , 103-114., 1968

Jean'Louis Koszul and his Hessian Geometry

slide-19
SLIDE 19
  • Thales Air Systems Date

Koszul Information Geometry

slide-20
SLIDE 20

20 / 20 / Projective Legendre Duality and Koszul Characteristic Function

LEGENDRE TRANSFORM FOURIER/LAPLACE TRANSFORM

ENTROPY= LEGENDRE(- LOG[LAPLACE]) Ψ = Φ − =

2 2 log

d d g

INFORMATION GEOMETRY METRIC

S d d g

2 * 2 *

= Ψ =

Ω −

− = Φ − = Ψ

*

,

log ) ( log ) ( dy e x x

y x

) ( , ) (

* * *

x x x x Ψ − = Ψ ξ ξ ξ d p p

x x

− = Ψ

*

) ( log ) (

* Φ(x) x,ξ

  • ξ,x

ξ,x x

e dξ e e p

*

+ − − −

= =

/ ) (ξ

=

*

) ( .

*

ξ ξ ξ d p x

x Legendre Transform of minus logarithm of characteristic function (Laplace transform) = ENTROPY !!!

ds2=d2ENTROPY ds2=-d2LOG[LAPLACE]

slide-21
SLIDE 21

21 / 21 / Koszul'Vinberg Characteristic Function/Metric of convex cone

J.L. Koszul and E. Vinberg have introduced an affinely invariant

Hessian metric on a sharp convex cone through its characteristic function.

  • is a sharp open convex cone in a vector space of finite

dimension on (a convex cone is sharp if it does not contain any full straight line).

  • is the dual cone of and is a sharp open convex cone.

Let the Lebesgue measure on dual space of , the following

integral: is called the Koszul-Vinberg characteristic function

Ω E R

*

Ω Ω ξ d

*

E E Ω ∈ ∀ = ∫

Ω − Ω

x d e x

x

) (

*

,

ξ ψ

ξ

slide-22
SLIDE 22

22 / 22 / Koszul'Vinberg Characteristic Function/Metric of convex cone

  • Koszul-Vinberg Metric :

We can define a diffeomorphism by:

with

When the cone

is symmetric, the map is a bijection and an isometry with a unique fixed point (the manifold is a Riemannian Symmetric Space given by this isometry): , and

  • is characterized by
  • is the center of gravity of the cross section
  • f :

= ψ log

2

d g

[ ]

( )

log log 2 1 log log ) ( log

2 u 2 2 2

∫∫ ∫∫ ∫ ∫ ∫

− + = = dudv dudv d d du du d du d x d

v u v u v u u u u

ψ ψ ψ ψ ψ ψ ψ ψ ψ ψ ψ

) ( log

*

x d x

x Ω

− = − = ψ α ) ( ) ( ), ( tu x f dt d x f D u x df

t u

+ = =

=

x

x α − =

*

x x =

* *)

( n x x =

*

, cste x x =

Ω Ω

) ( ) (

*

*

ψ ψ

*

x

{ }

n y x y y x = Ω ∈ = , , / ) ( min arg

* *

ψ

*

x

{ }

n y x y = Ω ∈ , ,

* *

∫ ∫

Ω − Ω −

=

* *

, , *

/ . ξ ξ ξ

ξ ξ

d e d e x

x x

slide-23
SLIDE 23

23 / 23 /

Koszul Entropy via Legendre Transform

we can deduce “Koszul Entropy” defined as Legendre Transform of minus

logarithm of Koszul-Vinberg characteristic function : with and where

Demonstration:

we set Using and we can write: and

) ( , ) (

* * *

x x x x Φ − = Φ Φ =

x

D x*

*

=

x

D x ) ( log ) ( x x

− = Φ ψ Ω ∈ ∀ = ∫

Ω − Ω

x d e x

x

) (

*

,

ξ ψ

ξ

∫ ∫

Ω − Ω −

=

* *

, , *

/ . ξ ξ ξ

ξ ξ

d e d e x

x x

∫ ∫

Ω − Ω − Ω

− = = −

* *

, , *

/ , ) ( log , ξ ξ ξ ψ

ξ ξ

d e d e h x d h x

x x h

∫ ∫

Ω − Ω − −

= −

* *

, , , *

/ . log , ξ ξ

ξ ξ ξ

d e d e e x x

x x x

∫ ∫ ∫ ∫ ∫ ∫ ∫

Ω − Ω − − Ω − Ω − Ω − Ω − Ω − −

        −         = Φ + − = Φ

* * * * * * *

, , , , , * * , , , , * *

/ . log log . ) ( log / . log ) ( ξ ξ ξ ξ ξ ξ ξ

ξ ξ ξ ξ ξ ξ ξ ξ ξ

d e d e e d e d e x d e d e d e e x

x x x x x x x x x

slide-24
SLIDE 24

24 / 24 /

Koszul'Vinberg Characteristic Function Legendre Transform

Thales Air Systems Date

                        − = Φ =             −             = Φ           − = Φ         −         = Φ + − = Φ − = Φ

∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫ ∫

Ω Ω − − Ω − − Ω Ω − − Ω Ω − − − Ω Ω − − Ω − Ω Ω − − − Ω − Ω − Ω − − Ω − Ω − Ω − Ω − Ω − −

* * * * * * * * * * * * * * * * * * * *

, , , , * * , , , , , , , , * * , , , , * * , , , , , * * , , , , * * *

log . ) ( 1 with . log . log ) ( . log log ) ( / . log log . ) ( log / . log ) ( , ) ( ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ

ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ

d d e e d e e x d d e e d d e e e d d e e d e x d d e e e d e x d e d e e d e d e x d e d e d e e x x x x

x x x x x x x x x x x x x x x x x x x x x x x x x

slide-25
SLIDE 25

25 / 25 /

[ ] ( )

[ ] [ ]

Φ = Φ ≥ Φ ⇒ Φ − ≥ Φ Φ ≤ Φ ⇒ Φ

* * *

) ( ) ( ) ( ) ( ) ( , ) ( : Transform Legendre ) ( conv. : Ineq. Jensen

* * * * * * * *

ξ ξ ξ ξ ξ ξ E d p x x x x x E E

x

Koszul Entropy via Legendre Transform

We can then consider this Legendre transform as an entropy, that we could

named “Koszul Entropy”: With and

ξ ξ ξ ξ ξ ξ

ξ ξ ξ ξ

d p p d d e e d e e

x x x x x x

∫ ∫ ∫ ∫

Ω Ω − − Ω Ω − −

− = − = Φ

* * * *

) ( log ) ( log

, , , , * Φ(x) x,ξ dξ e x,ξ

  • ξ,x

ξ,x x

e e dξ e e p

*

  • ξ,x

*

+ − ∫ − − − −

= = =

log

/ ) (ξ

( ) ξ

ξ ξ ξ ξ ξ d e d ξ.e d p Φ D x

ξ Φ

  • Φ(x)

x,ξ x x

* *

− Ω + − Ω

∫ ∫ ∫

= = = =

* *

. ) ( .

* [ ]

1 log ) ( ) ( log log ) (

* * * * * * *

) ( ) ( ) ( ) ( ,

= ⇒ − Φ = Φ − = − = Φ

∫ ∫ ∫ ∫

Ω Φ − Ω Φ − Ω Φ + Φ − Ω −

ξ ξ ξ ξ

ξ ξ ξ ξ

d e d e x x d e d e x

x x

[ ]

[ ] ( )

ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ

ξ ξ

E E d p d p x d p d p p e e p

x x x x x x x x * * * * * * * * ) ( ) ( ,

) (

  • r

) ( . ) ( ) ( if

  • nly

and if ) ( ) ( ) ( ) ( ) ( log ) ( log log ) ( log

* * * * *

Φ = Φ         Φ = Φ Φ = Φ = − ⇒ Φ − = = =

∫ ∫ ∫ ∫

Ω Ω Ω Ω Φ − Φ + −

slide-26
SLIDE 26

26 / 26 /

Barycentre & Koszul Entropy

Thales Air Systems Date

[ ]

[ ]

( )

ξ ξ E E

*

) (

*

Φ = Φ

[ ]

) ( , ) (

* * *

x x x Sup x

x

Φ − = Φ

( ) ξ

ξ ξ ξ ξ ξ d e d ξ.e d p Φ D x

ξ Φ

  • Φ(x)

x,ξ x x

* *

− Ω + − Ω

∫ ∫ ∫

= = = =

* *

. ) ( .

*

        Φ = Φ

∫ ∫

Ω Ω

* *

) ( . ) ( ) (

* *

ξ ξ ξ ξ ξ ξ d p d p

x x

Φ(x) x,ξ dξ e x,ξ

  • ξ,x

ξ,x x

e e dξ e e p

*

  • ξ,x

*

+ − ∫ − − − −

= = =

log

/ ) (ξ Barycenter of Koszul Entropy = Koszul Entropy of Barycenter

slide-27
SLIDE 27

27 / 27 /

Koszul metric & Fisher Metric

To make the link with Fisher metric given by matrix

, we can

  • bserve that the second derivative of

is given by:

We could then deduce the close interrelation between Fisher metric

and hessian of Koszul-Vinberg characteristic logarithm.

[ ]

2 2 2 2 2 2 2 2 2 2 2 2 *

log ) ( log ) ( , ) ( ) ( log , ) ( ) ( ) ( log x (x) x (x) x p E x I x (x) x x x x p x x p

x x x

∂ ∂ = ∂ Φ ∂ − =       ∂ ∂ − = ⇒ ∂ Φ ∂ = ∂ − Φ ∂ = ∂ ∂ − Φ = Φ − =

ψ ξ ξ ξ ξ ξ ξ

ξ 2 2 2 2

log ) ( log ) ( x (x) x p E x I

x

∂ ∂ =       ∂ ∂ − =

ψ ξ

ξ

FISHER METRIC (Information Geometry) = KOSZUL HESSIAN METRIC (Hessian Geometry) ) (x I ) ( log ξ

x

p

slide-28
SLIDE 28

28 / 28 /

Koszul Metric and Fisher Metric as Variance

We can also observed that the Fisher metric or hessian of KVCF

logarithm is related to the variance of :

Thales Air Systems Date

ξ

∫ ∫ ∫

− − −

− = ∂ ∂ ⇒ =

* * *

  • ξ,x
  • ξ,x
  • ξ,x

e dξ e x (x) Ψ dξ e (x) Ψ . 1 log log log ξ                 + −         − = ∂ ∂

∫ ∫ ∫ ∫

− − − − 2 2 2 2 2

. . . 1 log

* * * *

  • ξ,x
  • ξ,x
  • ξ,x
  • ξ,x

e dξ e dξ e dξ e x (x) Ψ ξ ξ

2 2 2 2 2 2

) ( . ) ( . . . log         − =             − = ∂ ∂

∫ ∫ ∫ ∫ ∫ ∫

− − − −

* * * * * *

  • x
  • x
  • ξ,x

ξ,x

  • ξ,x

ξ,x

p dξ p dξ dξ e e dξ dξ e e x (x) Ψ ξ ξ ξ ξ ξ ξ

[ ]

[ ]

) ( log ) ( log ) (

2 2 2 2 2 2

ξ ξ ξ ψ ξ

ξ ξ ξ

Var E E x (x) x p E x I

x

= − = ∂ ∂ =       ∂ ∂ − =

slide-29
SLIDE 29

29 / 29 /

New Definition of Maximum Entropy Density

How to replace by mean value of , in :

with

Legendre Transform will do this inversion by inversing We then observe that Koszul Entropy provides density of

Maximum Entropy with this general definition of density: with and where and

Thales Air Systems Date

( ) ( )

− −

Θ − Θ −

=

*

  • ξ,

ξ,

dξ e e p

ξ ξ ξ ξ

1 1

) ( ) (

1 ξ −

Θ = x dx x d x ) ( ) ( Φ = Θ = ξ ξ ξ ξ ξ

ξ

d p

=

*

) ( .

Ω −

− = Φ

*

,

log ) ( ξ

ξ d

e x

x

x ξ ) (

*

x = ξ

Ω − −

=

*

, ,

) ( ξ ξ

ξ ξ

d e e p

x x x

=

*

) ( . ξ ξ ξ ξ d px dx x d ) ( Φ = ξ

slide-30
SLIDE 30

30 / 30 /

Thales Air Systems Date

Cartan'Killing Form and Invariant Inner Product

It is not possible to define an ad(g)-invariant inner product for any

two elements of a Lie Algebra, but a symmetric bilinear form, called “Cartan-Killing form”, could be introduced (Elie Cartan PhD 1894)

This form is defined according to the adjoint endomorphism

  • f that is defined for every element of with the help of the

Lie bracket:

The trace of the composition of two such endomorphisms defines

a bilinear form, the Cartan-Killing form:

The Cartan-Killing form is symmetric: and has the associativity property: given by:

x

ad g x g

[ ]

y x y adx , ) ( =

( )

y xad

ad Tr y x B = ) , ( ) , ( ) , ( x y B y x B =

[ ] ( ) [ ] ( )

z y x B z y x B , , , , =

[ ] ( )

[ ]

( )

[ ]

( )

[ ] ( )

[ ]

( )

[ ] ( )

z y x B ad ad ad Tr z y x B ad ad ad Tr ad ad Tr z y x B

z y x z y x z y x

, , , , , , , ,

,

= = = =

slide-31
SLIDE 31

31 / 31 /

Thales Air Systems Date

Cartan'Killing Form and Invariant Inner Product

Elie Cartan has proved that if is a simple Lie algebra (the Killing

form is non-degenerate) then any invariant symmetric bilinear form

  • n is a scalar multiple of the Cartan-Killing form.

The Cartan-Killing form is invariant under automorphisms

  • f the algebra :

To prove this invariance, we have to consider:

Then

g g ) (g Aut ∈ σ g

( ) ( )

y x B y x B , ) ( ), ( = σ σ

[ ] [ ]

[ ] [

]

1 ) ( 1

rewritten ), ( ) ( , ) ( ) ( ), ( ,

− −

= = ⇒    = = σ σ σ σ σ σ σ σ σ

σ

  • x

x

ad ad z x z x y z y x y x

( )

( ) ( )

( )

( )

) , ( ) ( ), ( ) ( ), (

1 ) ( ) (

y x B ad ad Tr y x B ad ad Tr ad ad Tr y x B

y x y x y x

= = = =

σ σ σ σ σ σ

σ σ

slide-32
SLIDE 32

32 / 32 /

Thales Air Systems Date

Cartan'Killing Form and Invariant Inner Product

A natural G-invariant inner product could be introduced by Cartan- Killing form:

Cartan Generating Inner Product: The following Inner product

defined by Cartan-Killing form is invariant by automorphisms of the algebra where is a Cartan involution (An involution on is a Lie algebra automorphism

  • f whose square is equal to the

identity).

( )

) ( , , y x B y x θ − = g ∈ θ g θ g

slide-33
SLIDE 33

33 / 33 /

Thales Air Systems Date

From Cartan'Killing Form to Koszul Information Metric

( )

( )

Involution Cartan , with ) ( , , Form Killing Cartan ) , ( g y x B y x ad ad Tr y x B

y x

∈ − = − = θ θ Ω ∈ ∀ − = Φ

Ω −

x d e x

x

log ) ( Function stic Characteri Koszul

*

,

ξ

ξ

∫ ∫ ∫

− − Ω Ω

= = − = Φ Φ − = Φ

*

  • ξ,x

ξ,x x x x x

dξ e e p d p x d p p x x x x x ) ( Density Koszul ) ( . with ) ( log ) ( ) ( ) ( , ) ( Entropy Koszul

* *

* * * * * *

ξ ξ ξ ξ ξ ξ ξ

2 , 2 2 2 2 2

*

log ) ( ) ( log ) ( Metric Koszul x d e x (x) x I x p E x I

x x

∂ ∂ = ∂ Φ ∂ − ==       ∂ ∂ − =

Ω −

ξ ξ

ξ ξ

slide-34
SLIDE 34

34 / 34 /

Koszul Density: Application for SPD matrices

We can then named this new density as “Koszul Density”:

With

Φ(x) x,ξ dξ e x,ξ

  • ξ,x

ξ,x x

e e dξ e e p

*

  • ξ,x

*

+ − ∫ − − − −

= = =

log

/ ) (ξ

( ) ξ

ξ ξ ξ ξ ξ d e d ξ.e d p Φ D x

ξ Φ

  • Φ(x)

x,ξ x x

* *

− Ω + − Ω

∫ ∫ ∫

= = = =

* *

. ) ( .

*

( )

( )

[ ]

( )

Ω − − + + −

= = =

− * 1

). ( . with det ) (

1 det log 2 1

ξ ξ ξ ξ ξ α ξ

ξ ξ α α ξ

d p e e p

x Tr x n x Tr x

( )

           + = + = − = = = = ∈ ∀ =

− Ω + − Ω = Ω = Ω − Ω

1 * 2 1 dual

  • self

) ( , ,

2 1 det log 2 1 log ) ( det ) ( ) ( , , ,

* *

x n x d n d x I x d e x R Sym y x xy Tr y x

n n xy Tr y x x n

ψ ξ ψ ξ ψ

ξ

slide-35
SLIDE 35

35 / 35 /

Relation of Koszul density with Maximum Entropy Principle

The density from Maximum Entropy Principle is given by: If we take such that: Then by using the fact that with equality if and

  • nly if , we find the following:

Thales Air Systems Date

       = =        −

∫ ∫ ∫

Ω Ω Ω * (.)

* * *

) ( . 1 ) ( such ) ( log ) ( x d p d p d p p Max

x x x x px

ξ ξ ξ ξ ξ ξ ξ ξ

∫ − − − −

= =

*

  • ξ,x

*

dξ e x,ξ

  • ξ,x

ξ,x x

e dξ e e q

log

/ ) (ξ        − − = = = =

∫ ∫ ∫ ∫

Ω − ∫ − − − −

− *

, log

log , log ) ( log 1 / ). ( ξ ξ ξ ξ

ξ d

e x e q dξ e dξ e dξ q

x dξ e x,ξ x

  • ξ,x
  • ξ,x
  • x

*

  • ξ,x

* * *

( )

1

1 log

− ≥ x x 1 = x ξ ξ ξ ξ ξ ξ ξ ξ d p q p d q p p

x x x x x x

∫ ∫

Ω Ω

        − − ≤ −

* *

) ( ) ( 1 ) ( ) ( ) ( log ) (

slide-36
SLIDE 36

36 / 36 /

Relation of Koszul density with Maximum Entropy Principle

We can then observe that:

because

We can then deduce that: If we develop the last inequality, using expression of :

Thales Air Systems Date

) ( ) ( ) ( ) ( 1 ) (

* * *

= − =         −

∫ ∫ ∫

Ω Ω Ω

ξ ξ ξ ξ ξ ξ ξ ξ d q d p d p q p

x x x x x

1 ) ( ) (

* *

= = ∫

Ω Ω

ξ ξ ξ ξ d q d p

x x

ξ ξ ξ ξ ξ ξ ξ ξ ξ ξ d q p d p p d q p p

x x x x x x x

∫ ∫ ∫

Ω Ω Ω

− ≤ − ⇒ ≤ −

* * *

) ( log ) ( ) ( log ) ( ) ( ) ( log ) ( ) (ξ

x

q ξ ξ ξ ξ ξ ξ ξ

ξ

d d e x p d p p

x x x x

∫ ∫ ∫

Ω Ω − Ω

        − − − ≤ −

* * *

,

log , ) ( ) ( log ) (

∫ ∫ ∫

Ω − Ω Ω

+ ≤ −

* * *

,

log ) ( . , ) ( log ) ( ξ ξ ξ ξ ξ ξ ξ

ξ d

e d p x d p p

x x x x

) ( , ) ( log ) (

*

*

x x x d p p

x x

Φ − ≤ − ∫

ξ ξ ξ ) ( ) ( log ) (

* *

*

x d p p

x x

Φ ≤ − ∫

ξ ξ ξ

slide-37
SLIDE 37
  • Thales Air Systems Date

Souriau Lie Group Thermodynamics

slide-38
SLIDE 38
  • Thales Air Systems Date

Covariant Definition of Gibbs Equilibrium by Souriau

slide-39
SLIDE 39

39 / 39 /

Thales Air Systems Date

Covariant Definition of Thermodynamic Equilibriums

Jean-Marie Souriau , student of Elie Cartan at ENS Ulm in 1946,

has

given a covariant definition of thermodynamic equilibriums formulated statistical mechanics and thermodynamics in the framework

  • f Symplectic Geometry

by use of symplectic moments and distribution-tensor concepts, giving a geometric status for:

Temperature Heat Entropy This work has been extended by C. Vallée & G. de Saxcé, P.

Iglésias and F. Dubois.

slide-40
SLIDE 40

40 / 40 /

Thales Air Systems Date

Covariant Definition of Thermodynamic Equilibriums

The first general definition of the “moment map” (constant of the

motion for dynamical systems) was introduced by Jean-Marie Souriau during 1970’s

with geometric generalization such earlier notions as the Hamiltonian and the

invariant theorem of Emmy Noether describing the connection between symmetries and invariants (it is the moment map for a one-dimensional Lie group

  • f symmetries).

In symplectic geometry the analog of Noether’s theorem is the

statement that the moment map of a Hamiltonian action which preserves a given time evolution is itself conserved by this time evolution.

The conservation of the moment of a Hamilotnian action was called

by Souriau the “Symplectic or Geometric Noether theorem”

considering phases space as symplectic manifold, cotangent fiber of

configuration space with canonical symplectic form, if Hamiltonian has Lie algebra, moment map is constant along system integral curves.

Noether theorem is obtained by considering independently each

component of moment map

slide-41
SLIDE 41

41 / 41 /

Thales Air Systems Date

Souriau Covariant Model

Let be a differentiable manifold with a continuous positive

density and let E a finite vector space and a continuous function defined on with values in E. A continuous positive function solution of this problem with respect to calculus of variations:

is given by:

and and

Entropy can be stationary only if

there exist a scalar and an element belonging to the dual of E.

Entropy appears naturally as Legendre transform of :

M ω d ) (ξ U M ) (ξ p        = =       − =

∫ ∫ ∫

Q d p U d p d p p s ArgMin

M M M p

ω ξ ξ ω ξ ω ξ ξ

ξ

) ( ) ( 1 ) ( such that ) ( log ) (

) ( ) ( . ) (

) (

ξ β β

ξ

U

e p

− Φ

=

− = Φ

M U

d e ω β

ξ β ) ( .

log ) (

∫ ∫

− −

=

M U M U

d e d e U Q ω ω ξ

ξ β ξ β ) ( . ) ( .

) (

− =

M

d p p s ω ξ ξ ) ( log ) ( Φ β Φ ) ( . ) ( β β Φ − = Q Q s

slide-42
SLIDE 42

42 / 42 /

Thales Air Systems Date

Souriau Covariant Model

This value is a strict minimum of s, and the

equation: has a maximum of one solution for each value of Q.

The function is differentiable and we can write

and identifying E with its dual:

Uniform convergence of

proves that and that is convex.

Then, and are mutually inverse and differentiable,

where .

Identifying E with its bidual:

) ( . ) ( β β Φ − = Q Q s

∫ ∫

− −

=

M U M U

d e d e U Q ω ω ξ

ξ β ξ β ) ( . ) ( .

) ( ) (β Φ Q d d . β = Φ β ∂ Φ ∂ = Q

M U

d e U U ω ξ ξ

ξ β ) ( .

) ( ) (

2 2

> ∂ Φ ∂ − β ) (β Φ − ) (β Q ) (Q β dQ ds . β = Q s ∂ ∂ = β

slide-43
SLIDE 43

43 / 43 /

Thales Air Systems Date

Souriau'Gibbs Canonical Ensemble

In statistical mechanics, a canonical ensemble is the statistical

ensemble that is used to represent the possible states of a mechanical system that is being maintained in thermodynamic equilibrium.

Souriau has defined this Gibbs canonical ensemble on

Symplectic manifold M for a Lie group action on M

The seminal idea of Lagrange was to consider that a statistical

state is simply a probability measure on the manifold of motions

In Jean-Marie Souriau approach, one movement of a dynamical

system (classical state) is a point on manifold of movements.

For statistical mechanics, the movement variable is replaced by

a random variable where a statistical state is probability law on this manifold.

slide-44
SLIDE 44

44 / 44 /

Thales Air Systems Date

Souriau'Gibbs Canonical Ensemble

Symplectic manifolds have a completely continuous measure,

invariant by diffeomorphisms: the Liouville measure

All statistical states will be the product of Liouville measure by the

scalar function given by the generalized partition function defined by the generalized energy (the moment that is defined in dual of Lie Algebra of this dynamical group) and the geometric temperature , where is a normalizing constant such the mass of probability is equal to 1,

Jean-Marie Souriau generalizes the Gibbs equilibrium state to all

Symplectic manifolds that have a dynamical group.

To ensure that all integrals could converge, the canonical Gibbs

ensemble is the largest open proper subset (in Lie algebra) where these integrals are convergent. This canonical Gibbs ensemble is convex.

the mean value of the energy a generalization of heat capacity Entropy by Legendre transform

λ

U

e

. β − Φ

U β Φ

− = Φ

M Ud

e ω

β.

log

β ∂ Φ ∂ = Q β ∂ ∂ − = Q K Φ − = Q s . β

slide-45
SLIDE 45

45 / 45 /

Thales Air Systems Date

Souriau Lie Group Thermodynamic

For the group of time translation, this is the classical thermodynamic Souriau has observed that if we apply this theory for non-

commutative group (Galileo or Poincaré groups):

the symmetry has been broken Classical Gibbs equilibrium states are no longer invariant by this group This symmetry breaking provides new equations, discovered by

Jean-Marie Souriau.

For each temperature , Jean-Marie Souriau has introduced a

tensor , equal to the sum of cocycle and Heat coboundary (with [.,.] Lie bracket):

This tensor has the following properties:

  • is a symplectic cocycle
  • The following symmetric tensor , defined on all values of is positive

definite:

β

β

f f

( ) ( ) [ ]

2 1 2 2 2 1 2 1

, ) ( with ) ( . , ,

1 1

Z Z Z ad Z ad Q Z Z f Z Z f

Z Z

= + =

β β

f

β

f

β

β f Ker ∈

β

g (.)

β

ad

[ ] [ ] ( ) [ ] ( )

2 1 2 1

, , , , , Z Z f Z Z g β β β

β β

=

slide-46
SLIDE 46

46 / 46 /

Thales Air Systems Date

Souriau Lie Group Thermodynamic

Souriau equations are universal, because they are not dependent

  • f the symplectic manifold but only of:

the dynamical group G its symplectic cocycle the temperature the heat Souriau called this model “Lie Groups Thermodynamics”: “Peut-être cette thermodynamique des groups de Lie a-t-elle un intérêt

mathématique”.

For dynamic Galileo group (rotation and translation) with only one

axe of rotation:

this thermodynamic theory is the theory of centrifuge where the temperature

vector dimension is equal to 2 (sub-group of invariance of size 2)

these 2 dimensions for vector-valued temperature are “thermic conduction” and

“viscosity”, unifying “heat conduction” and “viscosity”.

( ) ( ) [ ]

2 1 2 2 2 1 2 1

, ) ( with ) ( . , ,

1 1

Z Z Z ad Z ad Q Z Z f Z Z f

Z Z

= + =

β β

β f Ker ∈

[ ] [ ] ( ) [ ] ( )

2 1 2 1

, , , , , Z Z f Z Z g β β β

β β

= β f Q

slide-47
SLIDE 47

47 / 47 /

The Galileo group of an observer is the group of affine maps Matrix Form of Gallileo Group Symplectic cocycles of the Galilean group: V. Bargmann (Ann.

  • Math. 59, 1954, pp 1–46) has proven that the symplectic

cohomology space of the Galilean group is one-dimensional.

Lie Algebra of Gallileo Group

Dynamic Gallileo Group

Thales Air Systems Date

                    =           1 1 1 1 ' ' t x e w u R t x

  • )

3 ( , and , ' . . '

3

SO R R e R w u x e t t w t u x R x ∈ ∈ ∈    + = + + =

+

  × ∈ ∈ ∈          

+

x x so R R

  • ω

ω ε γ η ε γ η ω : ) 3 ( , and ,

3

slide-48
SLIDE 48

48 / 48 /

Thales Air Systems Date

Fundamental Souriau Theorem

Let be the largest open proper subset of , Lie algebra of G,

such that and are convergent integrals

this set is convex and is invariant under every transformation ,

where is the adjoint representation of G, with:

  • where is the cocycle associated with the group G and the moment,

and is the image under of the probability measure .

Rmq: is changed but with linear dependence to , then Fisher

metric is unchanged by dynamical group:

− M U

d e ω

ξ β ) ( .

− M U

d e ω ξ

ξ β ) ( .

. Ω

  • a
  • a

a ) (β β

  • a

( )

( )

) ( .

1

β θ β θ

  • a

a a + Φ = − Φ → Φ

s s →

( )

) ( ) ( Q a a Q a Q

  • θ

θ = + → ) (ς ς

+

M

a θ ) (ς

+ M

a

M

a ς Φ β

( )

( ) [ ]

( )

β β β β θ β I a a I = ∂ Φ ∂ − = ∂ − Φ ∂ − =

− 2 2 2 1 2

) (

slide-49
SLIDE 49

49 / 49 /

Fundamental Souriau Theorem

Thales Air Systems Date

*

  • R

R

( )

β Φ

( ) ( )

) ( . β θ β

  • a

a + Φ

( ) ( )

β β Φ − = Q Q s . Q ) (Q a

θ

β ) (β

  • a

ς ) (ς

+ M

a e a G

slide-50
SLIDE 50

50 / 50 /

Thales Air Systems Date

Souriau Lie Group definition of Fisher Metric

Let be the derivative of (symplectic cocycle of G) at the identity

element and let us define: Then

  • is a symplectic cocycle of ,that is independent of the moment of G
  • There exists a symmetric tensor defined on the image of

such that: and that gives the structure of a positive Euclidean space

f θ

( ) ( ) [ ]

2 1 2 2 2 1 2 1

, ) ( with ) ( . , , ,

1 1

Z Z Z ad Z ad Q Z Z f Z Z f

Z Z

= + = Ω ∈ ∀

β

β

β

f

  • (

)

Ω ∈ ∀ = β β β

β

, , f

β

g

[ ]

β

β

., (.) = ad

[ ] ( ) ( ) ()

( )

. Im , , , , ,

2 1 2 1 2 1 β β β β

ad Z Z Z Z f Z Z g ∈ ∀ ∈ ∀ =

  • (

) ()

( )

. Im , , ,

2 1 2 1 β β

ad Z Z Z Z g ∈ ∀ ≥

slide-51
SLIDE 51

51 / 51 /

Thales Air Systems Date

Koszul Information Geometry, Souriau Lie Group Thermodynamics

Koszul Information Geometry Model Souriau Lie Groups Thermodynamics Model Characteristic function

Ω ∈ ∀ − = Φ

Ω −

x d e x

x

log ) (

*

,

ξ

ξ

∀ − = Φ

β ω β

ξ β

log ) (

) ( . M U

d e

Entropy

ξ ξ ξ d p p x

x x

− = Φ

*

) ( log ) ( ) (

* *

− =

M

d p p s ω ξ ξ ) ( log ) (

Legendre Transform

) ( , ) (

* * *

x x x x Φ − = Φ ) ( . ) ( β β Φ − = Q Q s

Density

  • f probability

− − + −

= =

*
  • ξ,x

ξ,x x Φ(x) x,ξ x

dξ e e p e p ) ( ) ( ξ ξ

− − Φ + −

= =

M U U U

d e e p e p ω ξ ξ

ξ β ξ β β β ξ β β ) ( . ) ( . ) ( ) ( .

) ( ) (

Dual Coordinate Systems

* *

and Ω ∈ Ω ∈ x x

∫ ∫ ∫

− Ω − Ω

= =

*
  • ξ,x

ξ,x x

dξ e d e d p x

* *

. ) ( .

*

ξ ξ ξ ξ ξ

∈ Q and β

∫ ∫ ∫

− −

= =

M U M U M

d e d e U d p U Q ω ω ξ ω ξ ξ

ξ β ξ β β ) ( . ) ( .

) ( ) ( ). ( heat Geometric

  • r

Map Moment Souriau

  • f

Mean : map Moment Souriau : e Temperatur Geometric Souriau : Q U β

Dual Coordinate Systems

x x x ∂ Φ ∂ = ) (

*

and

* * *

) ( x x x ∂ Φ ∂ = β ∂ Φ ∂ = Q

and

Q s ∂ ∂ = β

Hessian Metric

) (

2 2

x d ds Φ − =

( )

β Φ − =

2 2

d ds

Fisher metric

2 , 2 2 2 2 2

*

log ) ( ) ( log ) ( x d e x (x) x I x p E x I

x x

∂ ∂ = ∂ Φ ∂ − =       ∂ ∂ − =

Ω −

ξ ξ

ξ ξ 2 ) ( . 2 2 2 2 2

log ) ( ) ( log ) ( β ω β β β β ξ β

ξ β β ξ

∂ ∂ = ∂ Φ ∂ − =         ∂ ∂ − =

− M U

d e ) ( I p E I Capacity Geometric Souriau : ) (

2 2

β β β β β ∂ ∂ − = ∂ ∂ − = ∂ Φ ∂ − = Q K Q ) ( I

slide-52
SLIDE 52

52 / 52 /

Thales Air Systems Date

Geometric heat Capacity / Specific heat

We observe that the Information Geometry metric could be

considered as a generalization of “Heat Capacity”. Souriau called it the “Geometric Capacity”. This geometric capacity is related to calorific capacity.

  • is related to the mean, and is related to the variance of

β ∂ Φ ∂ = Q β β β β ∂ ∂ − = ∂ Φ ∂ − = Q ) ( I

2 2

) ( T Q kT T kT T Q Q K ∂ ∂ =             ∂ ∂ ∂ ∂ − = ∂ ∂ − =

2

1 1 β kT 1 = β Q K U

[ ]

[ ]

2 2 2 2

) ( ). ( ) ( . ) ( ) (         − = − = ∂ ∂ − =

∫ ∫

M M

d p U d p U U E U E Q I ω ξ ξ ω ξ ξ β β

β β ξ ξ

[ ]

U E d p U β Φ Q

M ξ β

ω ξ ξ = = ∂ ∂ =

) ( ). (

slide-53
SLIDE 53

53 / 53 /

FOURIER HEAT EQUATION & HEAT CAPACITY

D C T Q . = ∂ ∂ T Q T Q IFisher ∂ ∂ = ∂ ∂ − =

2

1 ) ( β β

[ ]

1 2 1 1

) ( .

− − −

∆ = ∂ ∂ ⇒ ∆ = ∂ ∂ β β β κ β κ

Fisher

I t T D C t T

slide-54
SLIDE 54

54 / 54 /

ΘερEός : Souriau « Lie Group » Thermometer

Thales Air Systems Date

Souriau has built a thermometer (θερός) device principle that could measure the Geometric Temperature using “Relative Ideal Gas Thermometer” based on a theory of Dynamical Group Thermometry and has also recovered the (Geometric) Laplace barometric law [ ]

     = ∂ ∂ − = Entropy Geometric : s e Temperatur (Planck) Geometric : Heat Geometric : with U Var ) ( β β β Q Q I Fisher

∂ Φ ∂ = β Q

( ) ( )

β β Φ − = Q Q s ,

∂ ∂ = Q s β

r g m

e r p

  • ,

) (

β −

slide-55
SLIDE 55

55 / 55 /

Thales Air Systems Date

Koszul Information Geometry, Souriau Lie Group Thermodynamics

Koszul Information Geometry Model Souriau Lie Groups Thermodynamics Model Convex Cone

Ω ∈ x Ω convex cone Ω ∈ β Ω convex cone: largest open subset of ,

Lie algebra of G, such that ∫

− M U

d e ω

ξ β ) ( .

and ∫

− M U

d e ω ξ

ξ β ) ( .

.

are convergent integrals

Transformation

( )

Ω ∈ → Aut g gx x with ) (β β

  • a

Transformation

  • f Potential

(non invariant) ( )

g x gx x det log ) ( ) ( ) ( + Φ = Φ → Φ

Ω Ω Ω

( )

( )

( )

( )β

θ β β β

1

) (

− Φ = Φ → Φ a a

Transformation

  • f Entropy

(invariant) ( ) ( )

* * * * *

* * *

) ( x x gx x

Ω Ω Ω Ω

Φ =       ∂ Φ ∂ Φ → Φ x x x ∂ Φ ∂ =

) ( with

*

( ) ( ) ( ) ( )

( )

( ) ( ) ( ) ( )

( )

( )

( )β

θ β β β θ β β θ β β β β β

1

' ' ) ( ) ( ' ' ' ' .with ) ( . ' ' '. ' '

− Φ = Φ = Φ = Φ + = ∂ + Φ ∂ = ∂ Φ ∂ = = = Φ − = Φ − = → a a a Q a a a a Q a Q s Q Q Q s Q s

  • Information

Geometry Metric (invariant)

( )

( ) [ ]

( )

x I x x x g x gx I = ∂ Φ ∂ − = ∂ + Φ ∂ − =

Ω Ω 2 2 2 2

) ( det log ) (

( )

( )

( ) [ ]

( ) ( )

β β β β β θ β β I a a I = ∂ Φ ∂ − = ∂ − Φ ∂ − =

− 2 2 2 1 2

) (

slide-56
SLIDE 56

56 / 56 /

Thales Air Systems Date

Invariance of Fisher Metric

In both Koszul and Souriau models, the Information Geometry

Metric and the Entropy are invariant respectively to:

the automosphisms

  • f the convex cone

to adjoint representation of Dynamical group G acting on , the convex

cone considered as largest open subset of , Lie algebra of G, such that and are convergent integrals.

g Ω

  • a

− M U

d e ω

ξ β ) ( .

− M U

d e ω ξ

ξ β ) ( .

.

( )

( )

[ ]

( )

x I x x x g x gx I = ∂ Φ ∂ − = ∂ + Φ ∂ − =

Ω Ω 2 2 2 2

) ( det log ) (

( )

( )

( )

[ ]

( ) ( )

β β β β β θ β β I a a I = ∂ Φ ∂ − = ∂ − Φ ∂ − =

− 2 2 2 1 2

) (

  • ( )

Ω ∈ → Aut g gx x with ) (β β

  • a

slide-57
SLIDE 57
  • Thales Air Systems Date

Lie Group Action on Symplectic Manifold with Modern notation (cf. Charles'Michel Marle)

slide-58
SLIDE 58

58 / 58 /

Notation

#

Let a Lie Group and tangent space of at its neutral

element

  • Adjoint representation of

with

  • Tangent application of at neutral element
  • f

For with Curve from

tangent to : and transform by :

Thales Air Systems Date

G G Te G e

( )

g e g e

i T Ad G g G T GL G Ad = ∈ →

  • :

1

:

ghg h ig

  • G

Ad Ad ad

[ ]

Y X Y ad G T Y X G T End G T Ad T ad

X e e e e

, ) ( , ) ( : = ∈ → =

  • e

G ) (K GL G

n

= C R K

  • r

= ) (K M G T

n e

=

1

) ( ), (

= ∈ ∈ gXg X Ad G g K M X

g n

[ ]

Y X YX XY Y Ad T Y ad K M Y X

X e X n

, ) ( ) ( ) ( ) ( , = − = = ∈ ) ( c I e

d =

= ) 1 ( c X = ) exp( ) ( tX Ad t = γ ) exp( ) ( tX t c = Ad YX XY tX Y tX dt d Y t dt d Y Ad T Y ad

t t X e X

− = = = =

= − = 1

) exp( ) exp( ) ( ) ( ) ( ) ( γ

slide-59
SLIDE 59

59 / 59 /

Action of a Lie Group on a Symplectic Manifold

Let be an action of Lie Group G on differentiable

manifold M, the fundamental field associated to an element of Lie algebra of group G is the vectors field on M:

  • is hamiltonian on a Symplectic Manifold , if is symplectic

and if for all , the fundamental field is globally hamiltonian

There exist linear application from to differential function on We can then associate a differentiable application ,called

moment of action :

Thales Air Systems Date

  • M

M G → × Φ : X

M

X

( ) ( )

, exp ) (

=

− Φ =

t M

x tX dt d x X

( )

) , ( ) , ( ,

2 1 2 1

x g g x g g Φ = Φ Φ x x e = Φ ) , (

with and

Φ M Φ

X

M

X

X

J X R M C → →

) , (

  • X

J

  • M

Φ J

= → X X x J x J x J x M J

X

, ), ( ) ( such that ) ( :

slide-60
SLIDE 60

60 / 60 /

Action of a Lie Group on a Symplectic Manifold

We associate a bilinear and anti-symmetric form , Symplectic

Cocycle of Lie algebra :

If with constant , then:

[ ] {

}

Y X Y X

J J J Y X , ) , (

,

− = Θ Θ

{ }

Bracket Poisson : .,.

with

[ ] [ ] [ ]

) , , ( ) , , ( ) , , ( = Θ + Θ + Θ Y X Z X Z Y Z Y X

with

  • µ

+ = J J'

µ

[ ]

Y X Y X Y X , , ) , ( ) , ( ' µ + Θ = Θ

With cobord of

[ ]

Y X Y X , , ) , ( µ µ = ∂

slide-61
SLIDE 61

61 / 61 /

Action of a Lie Group on a Symplectic Manifold

$

There exist a unique affine action such that linear part is

coadjoint representation and that induce equivariance of moment is called Cocycle associated to

The differential

  • f 1-cocyle associated to at neutral

element :

If then :

a ) ( ) , ( :

*

g Ad g a G a

g

θ ξ ξ + = → ×

  • with

X Ad X Ad

g g 1 *

, ,

= ξ ξ J

( ) ( )

) ( ) ( )) ( , ( ) , (

*

g x J Ad x J g a x g J

g

θ + = = Φ J

G : θ e θ

e

T θ J

[ ] {

}

Y X Y X e

J J J Y X Y X T , ) , ( ), (

,

− = Θ = θ µ + = J J'

[ ]

Y X Y X Y X , , ) , ( ) , ( ' µ + Θ = Θ µ µ θ θ

*

) ( ) ( '

g

Ad g g − + =

Where is cobord of G

µ µ

* g

Ad −

slide-62
SLIDE 62
  • Thales Air Systems Date

From Euler'Poincaré('Tchesayev) Equation of Geometric Mechanics to Poincaré'Marle'Souriau Equation of Geometric Thermodynamics

slide-63
SLIDE 63

63 / 63 /

Seminal Paper of Poincaré 1901, revisited by Marle

%&'()($ *($)+ ) +,,,--) .) /012/.& )&13&

Heni Poincaré proved that when a Lie algebra acts locally transitively

  • n the configuration space of a Lagrangian mechanical system, the

Euler-Lagrange equations are equivalent to a new system of differential equations defined on the product of the configuration space with the Lie algebra

%4'+ 5* *)6(78 ($*($ 8) )41) &5/9) 43&/

Marle has written the Euler-Poincaré equations, under an intrinsic

form, without any reference to a particular system of local coordinates

Marle has proven that they can be conveniently expressed in terms of

the Legendre and momentum maps of the lift to the cotangent bundle

  • f the Lie algebra action on the configuration space

Thales Air Systems Date

slide-64
SLIDE 64

64 / 64 /

Seminal Paper of Poincaré

Thales Air Systems Date

« Ayant eu l’occasion de m’occuper du mouvement de rotation d’un corps solide creux, dont la cavité est remplie de liquide, j’ai été conduit à mettre les équations générales de la mécanique sous une forme que je crois nouvelle et qu’il peut être intéressant de faire connaître » Henri Poincaré, CRAS, 18 Février 1901 « Elles sont surtout intéressantes dans le cas où U étant nul, T ne dépend que des η η η η » Henri Poincaré

slide-65
SLIDE 65

65 / 65 /

Lagrangian Mechanical system with M is smooth configuration space Poincaré assumes that a finite dimensional Lie algebra

acts on the configuration manifold M

New expression of the functional If

is a parametrized continuous, piecewise differentiable curve in M, and any lift of to M g, we have:

EULER'POINCARE EQUATION by Lift of admissible Curve

Thales Air Systems Date

[ ]

R TM L M t t dt dt t d L I

t t

→ →       = ∫ : and , : with ) ( ) (

1

1

γ γ γ

  • )

( ) , ( ) , ( : x X X x X x TM M

M

= → × ϕ ϕ

  • map)

moment Souriau to d (associate :

*

  • ×

→ M M T

t

ϕ

[ ] ( )

: with and : with , ) ( ) , ( : and ) , ( , : with ) ( ) (

1

1

× = → × = = → × = = × → = ∫ M p V p M M p p V(t) dt dγ x X L X x R M L L V t M t t dt t L I

M M M t t

γ γ γ ϕ γ γ γ γ γ

  • γ

γ γ ) ( ) ( γ γ I I =

slide-66
SLIDE 66

66 / 66 /

MARLE Intrinsic Expression of EULER'POINCARE EQUATION

J.M. Marle intrinsic expression (independant of any choice of local

coordinates) of the Euler-Poincaré equation:

Poincaré “This equation is useful mainly when

  • nly depends on its second variable (Lagrangian reduction)

Thales Air Systems Date

( )

( )

( )

× = Ω Ω =       − M L d p t V t t V t L d ad dt d

t t V

: ) ( ), ( ) ( ), (

1 * 2 * ) (

  • ϕ

γ γ

[ ] [ ]

∈ − = → = = X V X ad X ad ad X,V

  • V,X

X ad

V V V V

and , , , , such that : and

* *

ξ ξ ξ variable 2nd its and 1st its respect to with : function the

  • f

als differenti partial the be : :

* 2 * 1

R M L M L d M T M L d → ×      → × → ×

  • *

:

*

× M p R M L → × :

X

slide-67
SLIDE 67

67 / 67 /

EULER'POINCARE'MARLE EQUATION

* *#5($ #**

Souriau Moment Map: The Legendre Map

Thales Air Systems Date

M T J

*

:

∈ = X M T X X J

M M

, , ) ( , ), (

*

ξ ξ π ξ ξ

  • (

)

X X

M M M

), ( ) ( ξ π ϕ ξ π =

  • (

)

) ( ), ( ) ( :

*

ξ ξ π ξ ϕ ξ ϕ J M M T

M t t

= × →

  • (

)

J p J

M t t

, π ϕ ϕ = =

  • )

al differenti (vertical

*

L d M T TM

vert

  • ϕ

ϕ ϕ ϕ

  • J

L d p J p L d

t t

= ⇒ = =

2 2

with

slide-68
SLIDE 68

68 / 68 /

EULER'POINCARE'MARLE REDUCTION

Euler-Poincaré Equation with Legendre and Moment Maps: The Euler-Poincaré Equation and Reduction: Following the remark

made by Poincaré at the end of his note, let us now assume that the map only depends on its 2nd variable

The Euler-Poincaré Equation in Hamiltonian Formalism

Thales Air Systems Date

( ) ( ) ( ) ( )

) ( ), ( ) ( ) ( ), ( ) ( ), (

1 * ) (

t V t dt t d t V t L d J t V t J ad dt d

t V

γ ϕ γ γ γ ϕ = =       −

  • R

M L → × :

X

( )

( )

) (

* ) (

=       − t V L d ad dt d

t V

( )

R M T H M T TM M T L H → → ∈ − =

− − * * * 1 1

: : , ) ( ) ( , ) (

  • ξ

ξ ξ ξ ξ

slide-69
SLIDE 69

69 / 69 /

POINCARE'MARLE'SOURIAU EQUATION ON DUAL LIE ALGEBRA

Euler-Poincaré Equation in the Framework of Souriau Lie Group

Thermodynamics

Souriau Lie Group Thermodynamics: Poincaré-Marle-Souriau Equation:

Thales Air Systems Date

( )

( )

) (

* ) (

=       − t V L d ad dt d

t V *

=         ∂ Φ ∂       − β

β

ad dt d Q ad dt dQ

* β

=

( ) ( )

( )

) ( ), ( ,

1 1

Q Q Q Q Q s

− −

Θ Φ − Θ = Φ − = β β

( )

     Θ = ∈ ∂ Φ ∂ = Θ =

) ( ) (

1 Q

Q β β β β

∂ ∂ = Q Q s ) ( β

slide-70
SLIDE 70

70 / 70 /

POINCARE'MARLE'SOURIAU EQUATION ON DUAL LIE ALGEBRA

Recall that

has a natural Poisson structure called Kirillov- Kostant-Souriau structure, which allow to associate to any smooth function its Hamiltonian vector field :

If there exist a smooth function

such that , the parametrized curve satisfies the Hamilton differential equation on :

In Souriau Lie Group Thermodynamics:

Thales Air Systems Date

  • R

h →

  • :

− = ξ ξ ξ χ

ξ

, ) (

* ) ( dh h

ad R h →

  • :

J h H

  • =

[ ]

=

1 0,

: t t J ς ξ

  • (

)

) ( ) (

* ) (

t ad dt t d

dh

ξ ξ

ξ

− =

( ) ( )

β β Φ − = Q Q s ,

( )

) ( ) (

* )) ( (

t Q ad dt t dQ

t Q ds

− = dQ ds β =

slide-71
SLIDE 71

71 / 71 /

INFORMATION GEOMETRY BASED ON GEOMETRIC MECHANICS

Thales Air Systems Date

Élie CARTAN est le fils de Joseph CARTAN maréchal-ferrant. Elie CARTAN le FORGERON (du Latin FABER) l’Homme qui bat la matière sur l’ENCLUME, pour lui imprimer la COURBURE pour la mettre en FORME)

GEOMETRIC MECHANICS

  • H. POINCARE
  • SYMPLECTIC

MECHANICS J.M. SOURIAU LIE GROUP THERMODYNAMICS J.M. SOURIAU

  • HESSIAN GEOMETRY OF

INFORMATION J.L. KOSZUL

Texte de Bergson - Homo faber "En ce qui concerne l'intelligence humaine, on n'a pas assez remarqué que l'invention mécanique a d'abord été sa démarche essentielle… . Si nous pouvions nous dépouiller de tout orgueil, si, pour définir notre espèce, nous nous en tenions strictement à ce que l'histoire et la préhistoire nous présentent comme la caractéristique constante de l'homme et de l'intelligence, nous ne dirions peut-être pas Homo sapiens, mais Homo faber. En définitive, l'intelligence, envisagée dans ce qui en paraît être la démarche originelle, est la faculté de fabriquer des objets artificiels, en particulier des

  • utils à faire des outils et d'en varier indéfiniment la fabrication."

Henri Bergson, L’ Évolution créatrice (1907), Éd. PUF, coll. "Quadrige", 1996, chap. II, pp.138-140 NEW FOUNDATION OF INFORMATION THEORY (Sapiens) by GEOMETRIC MECHANICS (Faber)

slide-72
SLIDE 72
  • Legendre Transform &

Minimal Surface: Seminal Paper of Legendre

slide-73
SLIDE 73

73 / 73 /

Thales Air Systems Date

Legendre Transform and Minimal Surface

Legendre has introduced « Legendre Transform » to solve Minimal

Surface Problem

Classical “Legendre transform” with our previous notations:

( ) ( ) ( ) ( ) ( )

             =             = Φ =             Φ Φ =              =       =       =       =    = = Φ Φ − = Φ − = dQ ds dQ ds dQ ds d d d d d d Q y x q p Q Q Q q p Q s y x z Q Q Q s

2 1 2 1 2 1 2 1

and , ) , ( ) , ( with , . β β β β β β β ω β β β β β

dq d y dp d x q p y q x p y x z ω ω ω = = − + = and with ) , ( . . ) , (

slide-74
SLIDE 74

74 / 74 /

Thales Air Systems Date

Legendre Transform and Minimal Surface

In the following relation, we recover the definition of Entropy : But also relation with Mean Curvature

     = = ⇒      = = = + dQ ds ds dQ dq d y dp d x d dq y dp x β β ω ω ω . and . .

       Φ =         − − = ⇒ − =         ⇒ >> − = ⇒ << =                 Φ + Φ =           +

Φ Φ Φ

β β β β β β β β β d d Q Q Q Q H I Q Q Q Q Q d dQ Q Q d d Q H I Q H d d d d d d Q Q d d with 1 . 2 ) ( . . 1 2 ) ( 1 . 2 1 1

2 2 2 2 2 2

2nd Principle of Thermodynamics

slide-75
SLIDE 75
  • References
slide-76
SLIDE 76

76 / 76 /

References

!" [1] Koszul J.L., Variétés localement plates et convexité, Osaka J. Math. , n2, p.285-290, 1965 [2] Koszul J.L., Domaines bornées homogènes et orbites de groupes de transformations affines, Bull. Soc. Math. France 89, pp. 515- 533., 1961 [3] Koszul J.L., Ouverts convexes homogènes des espaces affines, Math. Z. 79, pp. 254-259., 1962 [4] Koszul J.L., Déformations des variétés localement plates, .Ann Inst Fourier, 18 , 103-114., 1968

  • [5] Souriau J.M., Définition covariante des équilibres thermodynamiques, Suppl. Nuov. Cimento, n1, pp. 203–216, 1966

[6] Souriau J.M., Structure of dynamical systems, volume 149 of Progress in Mathematics. Birkhäuser Boston Inc., Boston, MA,

  • 1997. A symplectic view of physics

[7] Souriau J.M., Thermodynamique et géométrie. In Differential geometrical methods in mathematical physics, II, vol. 676 of Lecture Notes in Math., pages 369-397. Springer, Berlin, 1978 [8] Souriau J.M., Géométrie Symplectique et physique mathématique, CNRS 75/P.785, Dec. 1975 [9] Souriau J.M., Mécanique classique et géométrie symplectique, CNRS CPT-84/PE-1695, Nov. 1984 [10] Souriau J.M., On Geometric Mechanics, Discrete and continuous Dynamical systems, V.ol 19, N. 3, pp. 595–607, Nov. 2007

  • [11] Barbaresco F., Information Geometry of Covariance Matrix: Cartan-Siegel Homogeneous Bounded Domains, Mostow/Berger

Fibration and Fréchet Median, Matrix Information Geometry, Springer, 2013 [12] Barbaresco F., Eidetic Reduction of Information Geometry through Legendre Duality of Koszul Characteristic Function and Entropy, Geometric Theory of Information, Springer, 2014 [13] Barbaresco F., Koszul Information Geometry and Souriau Geometric Temperature/Capacity of Lie Group Thermodynamics,, MDPI Entropy, n16, pp. 4521-4565, August 2014. http://www.mdpi.com/1099-4300/16/8/4521/pdf

6

[14] Shima H. , Geometry of Hessian Structures, world Scientific Publishing 2007 [15] Gromov M., IHES Six Lectures on Probabiliy, Symmetry, Linearity. October 2014, Jussieu. http://www.ihes.fr/~gromov/PDF/probability-huge-Lecture-Nov-2014.pdf https://www.youtube.com/watch?v=hb4D8yMdov4&index=5&list=PLx5f8IelFRgGo3HGaMOGNAnAHIAr1yu5W

slide-77
SLIDE 77
  • Past Conferences

(GSI’13, MaxEnt’14) and Announcement (GSI’15 & Brillouin Seminar)

slide-78
SLIDE 78

78 / 78 /

Past Conferences

French/Indian Workshop on « Matrix Information Geometry », Ecole Polytechnique &

Thales Research & Technology, 23-25th February 2011 (with Prof. Rajendra Bhatia)

http://www.lix.polytechnique.fr/~schwander/resources/mig/slides/

SMAI’11 Congress, Mini-Symposium on « Information Geometry », 23-27th Mai 2011

http://smai.emath.fr/smai2011/programme_detaille.php

Symposium on « Information Geometry & Optimal Transport », hosted at Institut Henri

Poincaré, 12th February 2012 (with GDR CNRS MIA)

https://www.ceremade.dauphine.fr/~peyre/mspc/mspc-thales-12/

SMF/SEE Conference on « Geometric Science of Information », Ecole des Mines de

Paris, August 2013

http://www.see.asso.fr/gsi2013

SMF/SEE Conference on MaxEnt with special Issue "Information, Entropy and their

Geometric Structures", Clos Lucé in Amboise, sponsored by Jaynes Foundation, 21- 26th Sept. 2014

https://www.see.asso.fr/node/10784

Leon Brillouin Seminar on « Geometric Science of Information », Institut Henri

Poincaré & IRCAM, launched by THALES since December 2009, with Ecole Polytechnque & INRIA/IRCAM

http://repmus.ircam.fr/brillouin/past-events

slide-79
SLIDE 79

79 / 79 /

GSI’13 Proceedings

Thales Air Systems Date

http://www.springer.com/computer/image+processing/ book/978-3-642-40019-3

slide-80
SLIDE 80

80 / 80 /

MaxEnt’14, Amboise, Clos Lucé, September 2014

34th International Workshop on Bayesian Inference and Maximum Entropy Methods in Science and Engineering Château Clos Lucé, Parc Leonardo Da Vinci, Amboise, France https://www.see.asso.fr/maxent14 http://djafari.free.fr/MaxEnt2014/Program_MaxEnt2014.html Sponsored by Jaynes Foundation Keynote speakers: Misha Gromov (IHES, Abel Prize 2009) Daniel Bennequin (Institut Mathématique de Jussieu) Roger Balian (CEA, French Academy of science) Stefano Bordoni (Bologna University) https://www.see.asso.fr/node/10784

slide-81
SLIDE 81

81 / 81 /

GSI’15 « Geometric Science of Information » 28th '30th October 2015 Ecole Polytechnique

Thales Air Systems Date

http://www.gsi2015.org

Authors will be solicited to submit a paper in a special Issue "Differential Geometrical Theory

  • f Statistics” in ENTROPY Journal, an

international and interdisciplinary open access journal of entropy and information studies published monthly online by MDPI. http://www.mdpi.com/journal/entropy/special_is sues/entropy-statistics

slide-82
SLIDE 82

82 / 82 /

Léon Nicolas Brillouin Seminar

+:;

  • <

Hosting lab:

IRCAM, Salle Igor Stravinsky Projet INRIA/CNRS/Ircam MuSync (Arshia Cont)

Animation : Arshia Cont (IRCAM & INRIA), Frank Nielsen (Ecole

Polytechnique), F. Barbaresco (Thales)

Web Site: http://repmus.ircam.fr/brillouin/home Abstracts, Videos & Slides: http://repmus.ircam.fr/brillouin/past-events

http://repmus.ircam.fr/brillouin/home

slide-83
SLIDE 83

83 / 83 /

Thank you for your attention

Thales Air Systems Date

Nous avouerons qu’une des prérogatives de la géométrie est de contribuer à rendre l’esprit capable d’attention: mais on nous accordera qu’il appartient aux lettres de l’étendre en lui multipliant ses idées, de l’orner, de le polir, de lui communiquer la douceur qu’elles respirent, et de faire servir les trésors dont elles l’enrichissent, à l’agrément de la société. Joseph de Maistre Si on ajoute que la critique qui accoutume l’esprit, surtout en matière de faits, à recevoir de simples probabilités pour des preuves, est, par cet endroit, moins propre à le former, que ne le doit être la géométrie qui lui fait contracter l’habitude de n’acquiescer qu’à l’évidence; nous répliquerons qu’à la rigueur

  • n pourrait conclure de cette différence même,

que la critique donne, au contraire, plus d’exercice à l’esprit que la géométrie: parce que l’évidence, qui est une et absolue, le fixe au premier aspect sans lui laisser ni la liberté de douter, ni le mérite de choisir; au lieu que les probabilités étant susceptibles du plus et du moins, il faut, pour se mettre en état de prendre un parti, les comparer ensemble, les discuter et les peser. Un genre d’étude qui rompt, pour ainsi dire, l’esprit à cette opération, est certainement d’un usage plus étendu que celui où tout est soumis à l’évidence; parce que les occasions de se déterminer sur des vraisemblances

  • u

probabilités, sont plus fréquentes que celles qui exigent qu’on procède par démonstrations: pourquoi ne dirions –nous pas que souvent elles tiennent aussi à des objets beaucoup plus importants ? Joseph de Maistre