Beruflich Dokumente
Kultur Dokumente
Plan of talk
A short history of the subject
Random systems, from nuclear Hamiltonians to mesoscopic systems to
financial markets. . .
Large N limit of U(N) gauge theory Topological expansion and the
counting of maps and other planar objects Stat mech models on
random lattices
Double scaling limit and 2D quantum gravity
QCD, Dijkgraaf-Vafa, etc
Feynman diagrams and large N limit
Counting of maps or triangulations (cf Edouard Maurel-Segalas talk)
Computational methods: saddle point; [loop equations (cf EMS)];
[orthogonal polynomials (cf Paul Zinn-Justin)]
Eugene P. Wigner
1902 - 1995
Freeman Dyson
1923 -
Data
Ensemble,
Other random systems and random matrices : from transport properties (for
ex. universal fluctuations of electric conductivity) in disordered mesoscopic
systems to financial markets. . .
12 xi Ai j x j
d xe
(2)n/2
1
det 2 A
n/2 1
1
1
(2)
b
A
x
A
x
+
b
x
i
i
j
j
i
i
ij bj
2
e
=
d xe 2
1
2
det A
Differentiate w.r.t. bi R n
1
d x xk1 xk2 xk e 2 x.A.x
1 b.A1 .b
e2
hxk1 xk2 xk i :=
=
R
1
2 x.A.x
n
b=0
b
b
k1
k
d xe
1
A1
A
kP1 kP2
kP1 kP
all distinct
y
X
XXX
pairings P of the k
X
...
Wick theorem
xk x k
2
1
xk
propagators
HH
j
H
...
P kP kP
1
kP
h(x4 ) p i =
graphs
=
=
log Z
1
2 + g x4
Ax
4!
dx e 2
2 12
A
2 21
=
A
gp
p!
p=0
gp
A2p
|Aut G |
x4 p 1 Ax2
e 2
dx
4!
=: eF
1
2 + g tr M 4 ]
N[
tr
M
2
4
= dM e
nm
4-valent vertex :
i
l
1
k = N i jk
= gN jk m np qi
[t Hooft]
10
gN
A topological expansion :
gN
= log Z =
conn. surf.
22h (h)
N
F (g).
h=0
22genus()
g#vert.()
symm. factor
11
[Br
ezin, Itzykson, Parisi, Z. 1978]
gn
planar diagrams symm.factor
with n 4vertices
12
13
4 +gB4 )
14
4 +gB4 )
+
+
+ +
+ +
+
+
+
+
+
+
+ ++
++
etc etc, many variations on that theme
Statistical mechanics models on a random lattice
15
M1
2 +M 2 )
tr
(M
1
2 )
e
and
M2
2
gN
h N1 tr M1 M2 M1 M2 )i
gN
Similar behavior in non Gaussian theory etr (V1 (M1 )+V2 (M2 )) provided
hM1 i = 0, hM2 i = 0.
16
If hM1 i =
6 0, hM2 i =
6 0,
1
0
0
1
tr M1 M2 M1 M2
1
0
0
1
11
00
00
11
11
00
00
11
1
0
0
1
1
0
0
1
0
1
0
1
1
0
0
1
0
1
0
1
1
0
0
1
1
0
0
1
0
1
0
1
1
0
0
1
0
1
0
1
1
0
0
1
11
00
00
11
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
Compare (a1 a2 a1 a2 ) = (a21 )2 (a2 ) + 2 (a1 )(a22 ) 2 (a1 )2 (a2 ) for two
free variables
17
log F
g
continuum 2D gravity. Keep all genera h in h=0 N 22h F (h) (g) by letting
gc g 0 as N in such a way that (gc g)(2str )/2 N = fixed.
Very interesting limit : appearance of integrable equations (KdV . . .),
solutions to Painleve equations . . .
Thus, double scaling limit models of 2D quantum gravity
18
19
Computational techniques
Consider integral over N N Hermitian matrices
Z=
dMeNtr V (M) ,
Z N
2 N i=1 V (i )
,
(
)
e
d
i
j
i
i=1
i< j
20
Z N
di exp
i=1
2 log |i j | N V (i )
i< j
i=1
1
i j = V (i ) .
j6=i
()
21
G (x) = 2
=
G
(x)
+V
(x)G(x) P(x)
(x i )(x j )
N i, j=1,,N
N
E
D
(x)V ( )
V
i
a polynomial in x of degree d 1, i.e.
with P(x) := N1 Ni=1
xi
G2 (x) V (x)G(x) +
1
G (x) + P(x) = 0 .
N
(Beware ! Not exact for N finite!) For N large, neglect the 1/N term
quadratic equation for G(x), with yet unknown polynomial P, hence
q
1
G(x) =
V (x) V (x)2 4P(x)
2
22
() on a support S ,
S d() = 1,
and G(x) =
R 2a d ()
2a
1
() = 2 4 2 on the segment [2, 2].
For more general potentials, assume first S to be still a finite segment
[2a , 2a ], in such a way that (*) becomes
2 P.P.
Z 2a
d ()
2a
= V ()
if [2a , 2a ] .
23
3ga4 + a2 1 = 0
(EQa2 )
and we may compute all invariant quantities like the free energy or the
moments
E Z
D1
tr M 2p = d 2p () .
G2p :=
N
For example G2 = (4 a2 )a2 /3, G4 = (3 a2 )a4 , etc. All these functions
24
1
of a2 are singular as functions of g at the point gc = 12
where the two
roots of (EQa2 ) coalesce. For example the genus 0 free energy
1 2
Z(g)
1
2
(0)
2
= log a (a 1)(9 a2 )
F (g) : = lim (1/N ) log
N
Z(0)
2
24
3g (2p 1)!
= ( 2 )p
[Tutte 62, BIPZ 78]
p!(p + 2)!
p=1 t
...
has a power-law singularity
F (0) (g) |g gc |5/2
ggc
(0)
(g) =
fn gn
n=0
fn const|gc |n n7/2 .
n
Comments
i) Nature of the 1/N 2 and of the g expansions, algebraic singularity at
finite gc
ii) Universal singular behavior at gc
iii) Extension to several cuts, the role of the algebraic curve (cf Eynard).
iv) Connected correlation functions and free (or non crossing) cumulants
v) Factorization property, localization of the matrix integral and the
master field [. . .]
25
26
2. Orthogonal polynomials
Z
d Pm ()Pn ()eNV () = hn mn
[Mehta, Bessis, . . .]
cf Paul Z-J . . .
3. Loop (or Schwinger-Dyson) equations
Z
dM
{ eNtrV (M) } = 0
Mi j
27
1
1
i = jn mn
Z( j) = h tr
N 1 jM
n=0
e (z)
W (z) = 1 + zn fn = 1 + W
n=1
with
or equivalently
Z( j) = W (z( j))
with z( j) = jZ( j)
28
Graphically [Cvitanovic]
Z( j) =
Z(j)
1 +
111111111
000000000
~
000000000
111111111
W(j)
000000000
111111111
000000000
111111111
000000000
111111111
Z
= W ( jZ( j))
...
...
f1 1 f2 2
k!
mk = k (k+1 q )! 1 ! 2 !
or conversely
29
End of Act I
30