Non-crossing polymers and the KPZ equation Andrea De Luca in - - PowerPoint PPT Presentation

non crossing polymers and the kpz equation
SMART_READER_LITE
LIVE PREVIEW

Non-crossing polymers and the KPZ equation Andrea De Luca in - - PowerPoint PPT Presentation

01/07/2015 1 29 / Non-crossing polymers and the KPZ equation Andrea De Luca in collaboration with P. Le Doussal arXiv:1505.04802 01/07/2015 KPZ equation PRL 56 889 (1986), Kardar, Parisi, Zhang relaxation lowest order Gaussian noise


slide-1
SLIDE 1

1 29 / 01/07/2015

Non-crossing polymers and the KPZ equation

Andrea De Luca in collaboration with P. Le Doussal arXiv:1505.04802 01/07/2015

slide-2
SLIDE 2

KPZ equation 2 29 / 01/07/2015

In 1D, the renormalization group provides exact exponents PRL 56 889 (1986), Kardar, Parisi, Zhang

relaxation (surface tension) lowest order non-linearity Gaussian noise

slide-3
SLIDE 3

Concrete examples 3 29 / 01/07/2015

Turbulent liquid crystals - PRL 104 230601

Tracy-Widom distribution

slide-4
SLIDE 4

Cole-Hopf mapping 4 29 / 01/07/2015

diffusion equation in a random potential: directed polymer partition function

slide-5
SLIDE 5

Quantum mechanics and replica 5 29 / 01/07/2015

Path integral representation (Feynman - Kac)

slide-6
SLIDE 6

High-temperature and Lieb-Liniger 6 29 / 01/07/2015

Rescaling of variables We end up with the attractive Lieb-Liniger Hamiltonian which is integrable in 1dimension!

slide-7
SLIDE 7

Bethe-ansatz approach 7 29 / 01/07/2015

Hard to treat: it contains space-time correlation of the KPZ height

decomposition in eigenstates

If we can compute the spectrum, we can find arbitrary moments...

slide-8
SLIDE 8

Bethe-ansatz equations 8 29 / 01/07/2015

The initial condition is symmetric: the dynamics lies in the bosonic sector of the Hamiltonian The coefficient implements the scattering matrix How to fix the values of rapidities?

superposition of plane waves in each sector

slide-9
SLIDE 9

Periodic boundary condition 9 29 / 01/07/2015

The values of rapidities are fixed by boundary

  • conditions. In the symplest case

Bethe-Ansatz equations for the LL model Solutions at finite L are not easy... But in the thermodynamic limit?

slide-10
SLIDE 10

String ansatz 10 29 / 01/07/2015

If for large L, we have a divergence in the LHS, which must be compensated by a pole in the RHS

bound states (strings) 5-string

1-string

slide-11
SLIDE 11

String features 11 29 / 01/07/2015

5-string energy momentum

slide-12
SLIDE 12

Needed ingredients 12 29 / 01/07/2015

Norm WF

slide-13
SLIDE 13

General expression for moments 13 29 / 01/07/2015

The sum over eigenstates becomes the sum

  • ver the possible partitioning of the n particles

into strings

sum over partitions

It is exact... but how to deal with it?

slide-14
SLIDE 14

Partition function at fixed string number 14 29 / 01/07/2015

Use the grancanonical partition function: In this way we can recover the free energy distribution

slide-15
SLIDE 15

Fredholm determinant 15 29 / 01/07/2015

Exchanging the two sums, we obtain In the large time limit, one obtains

Tracy-Widom GUE distribution EPL 90 2 (2010) Calabrese, Le Doussal, Rosso

slide-16
SLIDE 16

Non-crossing polymers 16 29 / 01/07/2015

Can we use replica approach to treat non-crossing polymers? Simplest example of interaction, together with disorder...!

slide-17
SLIDE 17

Karlin-McGregor formula 17 29 / 01/07/2015

Similar formulas for more than two polymers

slide-18
SLIDE 18

Coinciding points 18 29 / 01/07/2015

slide-19
SLIDE 19

Replica for non-crossing polymers 19 29 / 01/07/2015

The expression is analogous to the one for single polymer. But the bosonic sector gives a vanishing contribution! How to build wave functions with different symmetries?

slide-20
SLIDE 20

Wave function and Young tableau 20 29 / 01/07/2015

We look for eigen functions antisymmetric in the first two variables...

symmetric antisymmetric

More general ansatz...

slide-21
SLIDE 21

Nested Bethe Ansatz 21 29 / 01/07/2015

The auxliary variable implement the symmetry

  • f the wave function.

In general, one auxiliary variable for every doubled column

slide-22
SLIDE 22

String ansatz? 22 29 / 01/07/2015

For large L, the first equation suggests again the presence of strings What about the auxiliary variable?

slide-23
SLIDE 23

Contour integral 23 29 / 01/07/2015

The solution of the second equation are non trivial... But we are only interested on the sum

slide-24
SLIDE 24

Comparison with bosonic case 24 29 / 01/07/2015

After summing over the auxiliary variable, we get an expression very similar to the bosonic case Norm is unchanged

conserved quantities

  • f the LL
slide-25
SLIDE 25

Generalized Gibbs Ensemble 25 29 / 01/07/2015

we replace time evolution with a generalized evolution with multiples times

The average non-crossing probability is not affected by disorder!

slide-26
SLIDE 26

Comparison with numerics 26 29 / 01/07/2015

Two-lines derivation

slide-27
SLIDE 27

Recipe for higher moments 27 29 / 01/07/2015

In order to compute higher moments

  • symmetrize the polynomial in terms of

conserved charges

  • write the result as a set of derivatives

applied to the generalized moments

slide-28
SLIDE 28

Results for higher moments 28 29 / 01/07/2015

Leading order

Physical picture:

  • for most of the realization:

p is exponentially small

  • for a fraction 1/t
  • f the realization, p is O(c^2)
slide-29
SLIDE 29

Conclusions 29 29 / 01/07/2015

  • We developed a framework based on the Nested Bethe ansatz

to deal with non-crossing polymers in random media;

  • We computed exactly the large times asymptotics

for the moments of the non-crossing probability for two polymers;

  • Agreement with numerical lattice simulations:

the crossing probability is most of the time exponentially small Open questions:

  • generalization to multi-polymers
  • higher order large time asymptotics: connection with

random matrices?