Download - ON THE ENERGY TRANSFER IN FPU LATTICES H. Christodoulidi ...€¦ · NX−1 i=1 X∞ k=4 2k−2ak−2 k! (qi+1 −qi) k. (3) At the harmonic limit α = 0 is HT = HFPU = H 2 = NX−1

Transcript
Page 1: ON THE ENERGY TRANSFER IN FPU LATTICES H. Christodoulidi ...€¦ · NX−1 i=1 X∞ k=4 2k−2ak−2 k! (qi+1 −qi) k. (3) At the harmonic limit α = 0 is HT = HFPU = H 2 = NX−1

ON THE ENERGY TRANSFER IN FPU LATTICESH. Christodoulidi, A. Ponno, S. Flach and Ch. Skokos

Max-Planck-Institut fur Physik komplexer Systeme, Nothnitzer Str. 38, Dresden

AbstractWe study the evolution of dynamics in the Fermi-Pasta-Ulam-α model in order to classify and characterize the behavior ofthe system. This classification is divided into three main timeintervals, called stages, in which there is a qualitative changein the dynamics of the system. We compare every stage inthe FPU dynamics with those of Toda’s system. Thus wecan use Toda as a tool to distinguish between the chaotic andintegrable behavior in the FPU-α system.

IntroductionThe one dimensional FPU-α lattice with fixed boundary con-ditions is described by the Hamiltonian

HFPU =1

2

N−1∑

k=1

p2k +

N−1∑

k=0

[1

2(qk+1 − qk)

2 +α

3(qk+1 − qk)

3] (1)

with x(0) = x(N) = 0.The Toda lattice, described by the Hamiltonian function

HT =

N−1∑

k=1

p2k +

1

4a2

N−1∑

k=1

e2a(qk+1−qk) − N

4a2(2)

can be regarded as an approximation of the FPU-α Hamilto-nian (1) of order α2, since

HT = HFPU −N−1∑i=1

∞∑

k=4

2k−2ak−2

k!(qi+1 − qi)

k. (3)

At the harmonic limit α = 0 is

HT = HFPU = H2 =

N−1∑

k=1

Ek (4)

where Ek are the harmonic energies and ωk = 2 sin kπ2N the

harmonic frequencies for both systems.In the present work, we excite the first normal mode k = 1of the FPU-α and Toda systems, with initial conditions qi =A · sin πi

N , qi = 0 , for i = 1, 2, ..., N − 1 .

STAGE IThe harmonic energies of both systems are characterized by asharp power law growth in time, due to acoustic resonances.

10 100 1000 100001E-60

1E-55

1E-50

1E-45

1E-40

1E-35

1E-30

1E-25

1E-20

1E-15

1E-10

1E-5

1

toda fpu theoretical

Ek/E

t

(i) E=0.01

10 100 10001E-60

1E-55

1E-50

1E-45

1E-40

1E-35

1E-30

1E-25

1E-20

1E-15

1E-10

1E-5

1

toda fpu theoretical

Ek/E

t

(ii) E=0.1

10 1001E-60

1E-55

1E-50

1E-45

1E-40

1E-35

1E-30

1E-25

1E-20

1E-15

1E-10

1E-5

1

toda fpu theoretical

Ek/E

t

(iii) E=1

Figure 1: FPU-α and Toda systems with N = 32, α = 0.33.Normalized harmonic energy evolution plotted in logarithmicscale, for different values of the total energy: i) E = 0.01, ii)E = 0.1, iii) E = 1. Blue lines correspond to the power lawEk ∝ t2(k−1).

STAGE IIThe harmonic energy spectrum saturates around a constantprofile.FPU-α and Toda systems exhibit in Fourier space the natu-ral packet formation, in which the total energy is exchangedamong few low-frequency modes [1], and the exponential en-ergy localization of the tail modes [2], i.e. higher-frequencymodes.This stage holds for FPU-α up to the time that the energyspectra of tail modes slowly grow and the system tends toreach equipartition.

The exponential localization, of the averaged Toda energyspectra of the tail modes, are given by

Log(Ek

E) = −σk + % (5)

whereσ =

eC0

N√

αε1/4(6)

and% = Log(

C1

Nαε1/2) (7)

with C0 ' 3−1/2 and C1 ' 6 (See Fig.4).

10000 100000 1000000 1E7 1E81E-25

1E-23

1E-21

1E-19

1E-17

1E-15

1E-13

1E-11

1E-9

1E-7

1E-5

1E-3

0.1

10

Toda FPU

Ek/E

t

(i) E=0.01

10000 100000 1000000 1E7 1E81E-13

1E-11

1E-9

1E-7

1E-5

1E-3

0.1

10

Toda FPU

Ek/E

t

(ii) E=0.1

10000 100000 1000000 1E7 1E81E-7

1E-6

1E-5

1E-4

1E-3

0.01

0.1

1

Toda FPU

Ek/E

t

(iii) E=1

Figure 2: FPU-α and Toda systems with N = 32, α = 0.33.Normalized and time averaged harmonic energy evolution forboth systems, plotted in logarithmic scale, for different valuesof the total energy: i) E = 0.01, ii) E = 0.1, iii) E = 1.

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 321E-22

1E-20

1E-18

1E-16

1E-14

1E-12

1E-10

1E-8

1E-6

1E-4

0.01

1 104

106

108

-

Ek/

E

k

E=0.01 (i) Toda

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 321E-12

1E-10

1E-8

1E-6

1E-4

0.01

1 104

106

108

-

Ek/

E

k

E=0.1 (ii) Toda

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 321E-7

1E-6

1E-5

1E-4

1E-3

0.01

0.1

1

104

106

108

-

Ek/

E

k

E=1 (iii) Toda

Figure 3: Toda system with N = 32, α = 0.33. Normalizedand time averaged energy spectra for total energy: i) E =0.01, ii) E = 0.1, iii) E = 1. In each panel we plot thespectra at times 104, 106, 108. The black line corresponds toEq. (5).

100 10006

8

10

12

14

16

N

1/

Toda

32 64 128 256 512 1024

Figure 4: Toda system with α = 0.33. Numerical evidencefor the validity of Eq. (6). The plot of Log(σN) versus−Log(ε) is a line that saturates to Eq. (6) as the degreesof freedom N increase. The red line, which corresponds toN = 1024, is −0.256Log(ε) + 0.4278.

STAGE IIIEnergy is diffused in FPU-α system from the packet to thetail.We numerically estimate this diffusion by computing i) themoments of the normalized energy spectra, defined as

ms =

N∑

k=1

ks(Ek

E)s (8)

ii) the sum of the harmonic energies of the last third part ofthe mode interval i.e. [N/3, N ], which we call tail energy

η =

N∑

k=N/3

Ek

E. (9)

Numerical evaluations of both quantities show a power lawincrease in time, of the form

ms ∝ Dstγs, η ∝ Dtγ (10)

for the moments and the tail energy respectively (see Figs.5and 7 ).

By ms and η we denote the moments and the tail energy, de-fined for the averaged harmonic energies Ek.The exponents of these powers laws are numerically evalu-ated and appear in Figs.6 and 7(ii). Both of them fluctuatevery strongly indicating a dense region of invariant objects.As E increases, these objects are destroyed and the systemtends to equipartition linearly in time (η ∝ t).

10 100 1000 10000 100000 1000000 1E7 1E80.01

1

100

10000

1000000

1E8

1E10

1E12

1E14

1E16

1E18

1E20

mj

t

(i) E=0.01

10 100 1000 10000 100000 1000000 1E7 1E80.01

1

100

10000

1000000

1E8

1E10

1E12

1E14

1E16

1E18

1E20(ii) E=0.1

mj

t10 100 1000 10000 100000 1000000 1E7 1E8

0.01

1

100

10000

1000000

1E8

1E10

1E12

1E14

1E16

1E18

1E20

1E22

1E24

mj

t

(iii) E=1

Figure 5: FPU-α model with N = 32, α = 0.33. Evolutionof moments mj, j = 1, ..., 16, plotted in logarithmic scalefor total energy i) E = 0.01, ii) E = 0.1, iii) E = 1. Thevariation of moments, from m1 to m16, is shown by the linecolour, which ranges from yellow to blue. The black curvescorrespond to moments mj,j = 1, ..., 16.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

j

E

(i) instant

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

j

E

(ii) average

Figure 6: FPU-α model with N = 32, α = 0.33. i) The slopeγj of the least squares linear fit, of moments mj, j = 1, ..., 16versus the total energy of the system E. Same line colourswith Fig.5 are used. ii) The same for the slope γj of momentsmj, j = 1, ..., 16.

10 100 1000 10000 100000 1000000 1E7 1E81E-18

1E-16

1E-14

1E-12

1E-10

1E-8

1E-6

1E-4

0.01

1

100

10000

E=1

E=0.1

t

(i) N=32

E=0.01

1/3

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

E

instant average

(ii) N=32

Figure 7: FPU-α model with N = 32, α = 0.33. i) Evolutionof the tail energies, instantaneous η and averaged η (orangecurves). The black straight lines correspond to the linear fitof η in the time windows [104, 3 · 106] for the total energiesE = 0.01, E = 0.1 and [103, 3 · 105] for E = 1. Equipartitionof the system is reached when η = 1/3. ii) The slope γ of thelinear fit of η and η versus E.

10 100 1000 10000 100000 1000000 1E7 1E8506.21095

506.21096

506.21097

506.21098

506.21099

506.21100

506.21101(i) E=0.01

J

t

FPU Toda

10 100 1000 10000 100000 1000000 1E7 1E8508.688

508.689

508.690

508.691

508.692

508.693

508.694(ii) E=0.1

FPU Toda

J

t

10 100 1000 10000 100000 1000000 1E7 1E8

522

524

526

528

530

532

534

J

t

FPU Toda

(iii) E=1

Figure 8: Numerical computation of the 2nd Toda integral Jfor both systems, for the total energies i) E = 0.01, ii) E =0.1, iii) E = 1.

ConclusionsThe energy transfer in the FPU-α model from the lower fre-quency modes to the tail modes is initially very sharp, afterthat stops for a certain time window [3] and then starts againwith a linear in time process, that leads the system to equipar-tition. Comparison with the Toda model shows that only thelast part is due to non-integrability of FPU-α.

AcknowledgementsWe thank G. Benettin and S. Ruffo for fruitful discussions.H.Ch. appreciated the warm hospitality of the MPIPKS,Dresden.

References[1] L. Berchialla, L. Galgani, A. Giorgilli DCDS 11, 855-866, (2004).[2] S. Flach, M. V. Ivanchenko, O. I. Kanakov, Phys. Rev. Lett. 95 064102 (2005).[3] S. Flach, A. Ponno Physica D 237, 908-917, (2008).