Professional Documents
Culture Documents
Filoche Arcachon 2020
Filoche Arcachon 2020
Filoche Arcachon 2020
M. Filoche
Laboratoire de Physique de la Matière Condensée
CNRS, Ecole Polytechnique, IP Paris
School on Anderson localization: Landscape theory, experiments with cold atoms, 7-9 October 2020
Acknowlegdments
Mathematics
• Svitlana Mayboroda, Douglas N. Arnold (Univ. Minnesota)
• Guy David (Univ. Paris Saclay)
• David Jerison (MIT)
Theoretical physics
Cold Atoms
Douglas Arnold Alain Aspect Guy David David Jerison Marcel Filoche
(Univ. of Minnesota) (IOGS) (Univ. Paris Sud) (MIT) (Ecole Polytechnique)
Richard Friend Svitlana Mayboroda Yves Meyer Jim Speck Claude Weisbuch
(Univ. of Cambridge) (Univ. of Minnesota) (ENS Paris Saclay) (UCSB) (UCSB/ Ecole Polytechnique)
Disorder-induced single-particle localization: the Anderson model
Lagendijk, van Tiggelen, Wiersma, Phys. Today (2009), Billy et al., Nature (2008)
Fifty years of Anderson localization
Localization of acoustic waves
B. Sapoval
Statistical approaches to Anderson localization
!B
Diffusion
Mobility edge
Localization
Diffusion in weak disorder Localization in
“extreme” disorder
x µ ( Ec - E ) s µ ( E - Ec )
-n s
§ Scaling theory (renormalization) Abrahams, Anderson, Liciardello & Ramakrishnan, Phys. Rev. Lett. 1979
However
§ Disorder is described through its statistical properties.
§ Values of the critical exponents have to be determined numerically or with approximate
theories.
§ The exact value of the mobility edge depends on the disorder type (correlations).
§ No analytical prediction when correlations or interactions come into play.
“Our results deviate significantly from previous theoretical estimates using an approximate,
self-consistent approach of localization.”
Delande & Orso, Phys. Rev. Lett. 2014
The Localization Landscape Theory
Kinetic energy Potential energy reduced kinetic energy effective potential energy
u predicts the fundamental energy and quantum state inside each localization region
Fundamental state in each region Fundamental energy
! 3
y0( m)
»
u 1
u u òòò ) d r
u ( r
u E0( m ) » u =
Wm
2 !
òòò ( r ) d r
2 3
u u
Wm
æ dö æ1ö
E0( m ) » ç1 + ÷ min ç ÷ Arnold, David, Filoche, Jerison, Mayboroda
è 4ø èuø SIAM J. Sci. Comput. 41, No. 1, pp. B69-B92 (2019)
Predicting the lower end of the spectrum
Energies
⎛ d⎞
E0 ≈ ⎜ 1+ ⎟ Wmin
Wmin Ratio E/Wmin ⎝ 4⎠
Localization landscape:
a
Disordered 1D potentials
Disordered 2D potential
Arnold, David, Filoche, Jerison, Mayboroda, SIAM J. Sci. Comput., 41, No. 1, pp. B69-B92 (2019)
Exponential decay in the effective potential
!!
! ! ææ rr22 ö
inf ççç òò ((W
Agmon distance rrWV ,,EE ( r11, r22 ) = inf V ( s ) - E )++ ddss ÷÷
paths ç !
paths !
èè r1r1 ø
! !!
y ( r ) " e V ,,EE r,,¶¶BBEE ))
(
(
- rW r 2 m(V - E )
- x
Decay of eigenfunction "e !2
"
S. Agmon, Lect. Notes Math. 1159, 1 (1985)
log ( Ñy )
y
wells
wells
Predicting the density of states
Uncertainty principle: k
Counting eigenvalues below E Counting volume in phase space
“W-based” law:
Arnold, David, Jerison, Mayboroda, Filoche, Phys. Rev. Lett. 116, 056602 (2016)
Predicting the density of states: the “W-based” law
1D 2D
correlated periodic
Wigner function
Weyl’s transform
Localization landscape in cold atom systems: Spectral functions
Wigner function
Weyl’s transform
Idea 2: the distribution is energy is to be found in the Wigner function of the quantum state
Conjecture:
Localization landscape in cold atom systems: Spectral functions
Speckle potential
Speckle potential
§ One mathematical object, the localization landscape, contains most of the information
about the localization of quantum states in complex systems. It is obtained by solving
one single linear problem:
æ !2 ö
ç- D +V ÷ u = 1
è 2m ø
§ The landscape predicts the localization subregions, the shape of the fundamental eigenfunction in each
localization region, and its energy.
§ W=1/u can be understood as an effective confining potential that is experienced by the quantum states. This
new potential can be used to compute the density of states and assess the long range decay of the states.
§ The counting function can be approximated in two different ways: either through a revisited Weyl’s formula or
through the so-called Landscape Law. The latter provides for the first time bounds from above and below to the
counting function on the entire spectrum.
§ A Weyl-Wigner approach allows us to compute the spectral functions in a way that seems effective both in the
semi-classical and in the quantum regime.