JOURNAL OF CHEMICAL PHYSICS

VOLUME 121, NUMBER 13

1 OCTOBER 2004

Complex autocorrelation function and energy spectrum by classical trajectory calculations
˘ˇ´ ´ Petra R. Zdanskaa)
Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, Flemingovo 2, 166 10 Prague 6, Czech Republic

Nimrod Moiseyev
Department of Chemistry and Minerva Center of Nonlinear Physics in Complex Systems, Technion—Israel Institute of Technology, Haifa 32000, Israel

Received 14 April 2004; accepted 8 July 2004 A quasiclassical method which enables evaluation of complex autocorrelation function from classical trajectory calculations is proposed. The method is applied for two highly excited nonlinearly coupled harmonic oscillators in regimes prevailed either by regular or chaotic classical motions. A good agreement of classical and quantum autocorrelation functions is found within short Ehrnfest time limit. Fourier transforms of the autocorrelation functions provide moderate resolved energy spectra, where classical and quantum results nearly coincide. The actual energy levels are obtained from approximate short-time autocorrelation functions with the help of filter diagonalization. This study is a follow up to our previous work P. Zdanska and N. Moiseyev, J. Chem. Phys. 115, 10608 2001 , where the complex autocorrelation has been obtained up to overall phase factors of recurrences. © 2004 American Institute of Physics. DOI: 10.1063/1.1787489

I. INTRODUCTION

Our present study is motivated by a quest for a method suitable for calculation of rovibronic spectra of nearly classical polyatomic systems, such as chromophores bound in van der Waals clusters,1–5 van der Waals matrices,6,7 etc. Calculation of rovibronic spectra of polyatomic systems represents a hardly tractable problem. Fully quantum treatment is possible for systems up to four to nine degrees of freedom,8,9 although for special cases systems up to 24 degrees of freedom have been calculated.10,11 Another possibility is mixed quantum-classical approach where a small important part of a system is simulated in quantum way, while the influence of the rest of the system is approximated by classical dynamics.12,13 The third type of approach uses a full semiclassical treatment.14 –17 Our present concern is a quasiclassical approach known as Wigner phase-space WPS method.18,19 All degrees of freedom are treated on the same footing, i.e., by classical mechanics, and quantum mechanics comes into the calculation only through the initial choice of positions and momenta of classical trajectories. All quantities are calculated without a specific evaluation of the semiclassical phase, which makes the WPS method promising for larger systems than those allowed by quantum and other semiclassical approaches. The other motivation of our study is a theoretical puzzle posed by the WPS approach. The WPS method is capable of very accurate approximations to cross sections19 or meanfield potentials,20 namely, quantities which are dependent on quantum time-dependent wave function—not merely on its absolute value, but the phase, too. Therefore one asks, whether complex quantum amplitudes, such as the autocora

Electronic mail: petra.zdanska@uochb.cas.cz 6175

relation function, can be obtained from WPS propagations, and if yes, then how? One possibility is to map timedependent phase-space density onto a wave function and thus to calculate the autocorrelation function in Hilbert space.21 Yet the backward mapping of classical evolved phase-space density includes a conceptual problem, as long as classical and quantum evolutions result in general case in very different phase-space densities.19 Thus the introduced error may be hard to estimate in unknown systems. Counter to one’s intuitive notions, the expectation values calculated from the classical evolved density are rigorous approximations to quantum expectation values,22 due to, loosely speaking, vanishing of the oscillatory part of the quantum density upon the integration. Therefore we suggest to utilize expectation values of the product Hamiltonian—evolution operator rather then the backward mapping of the classical evolved phase-space density, in order to calculate the phase of the autocorrelation function. Our paper is organized as follows: In Sec. II we summarize the theory for the WPS approach needed for derivations of new formulas. In Sec. III we summarize the results of our previous study:23 We show that the phase of autocorrelation function can be defined using expectation values of the product operator of Hamiltonian and evolution operator. This definition results in the phase integration which has to go through regions in time domain of zero amplitude, therefore the phase of individual recurrences is obtained only up to the relative phases between recurrences.23 In Sec. IV we propose a solution to overcome this difficulty. We define an auxiliary cross correlation function with a frozen wave packet which propagates on a classical periodic orbit. The cross-correlation function does not vanish which allows for integration of the phase, and coincides with
© 2004 American Institute of Physics

0021-9606/2004/121(13)/6175/11/$22.00

Downloaded 14 Jan 2005 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp

p j . P q. R. This designed coincidence enables us to obtain the relative phase between recurrences.t .p 1 2 F .pi (t) .pi t . ˆ A t dq dpA q. The nominator is defined as the ˆ ˆ ˆ expectation value P H (t) of the product operator A H ˆ H.aip. In Sec. qi . A (t) is evaluated by Monte Carlo integration over classical trajectories. WIGNER PHASE-SPACE APPROACH ˆ A t i A qi t . is obtained using the known WPS approximation given by Eq. 9 By differentiating the identity C(t)/C * (t) exp 2i C(t) we obtain the expression for the time derivative of the phase given by C 1 d qe i/ q/2 p• q t q/2 . PHASE OF AUTOCORRELATION FUNCTION Calculation of the complex autocorrelation function is considered fully within the realm of the WPS method.pi . 13. Phase-space points are preferentially distributed in the region of P such that the weight. C (t). 3 and 6 . and the point is selected if w i N. w j P(q j . 8 III. iii The quantum expectation values. 3 and 6 . 4 and 5 . 1 where ˆ H ˆ R H . 121.p d qe i/ d pe i/ p• q q p ˆ q/2 A q ˆ p/2 A p q/2 p/2 . 4 5 Equation 11 allows us to calculate the phase using the timedependent expectation values. AH AV AT . V and VI. is ascribed to every randomly generated point q j . 3 ˆ where A corresponds with A through the Wigner-Weyl transˆ given by form of the operator A A q. . P is represented as the ensemble of phase-space points. N max P . the phase is calculated using expectation values. The potential operator is expanded in Taylor series such that.p P q. therefore in Sec. C 2 .1. While in the regular case Sec. Vol.6176 J. Redistribution subject to AIP license or copyright.jsp .org/jcp/copyright. ˆ P Following the WPS approach. of the autocorrelation function be defined as C t C t ei C t The WPS method is a quasiclassical approach to quantum dynamical problems. Phys. Our calculations give an evidence that the presented method works in regular and chaotic regimes alike. in the chaotic regime one has to define multiple periodic orbits to follow individual recurrences Sec. 11 ˆ and P is the density operator defined as ˆ P . V a single periodic orbit is sufficient to follow the periodic motion. and thus the complete autocorrelation function can be calculated: phase and amplitude. C t e ˆ iH t/ . the weight is compared with the randomly generated number. 8 . AV d qe i/ V q p• q q• p q q/2 * q q/2 13 ˆ In practice. II. consisting of three steps: i The initial phase-space density P is obtained as a Wigner transform of the initial wave function . A is calculated using Eq. Let the phase. 12 ˆ Ae ˆ i/ H t 2 are approximated by the phase space integrals. Moiseyev the required autocorrelation function at periodic time intervals. the survival probability C(t) 2 defined as the absolute value squared of the autocorrelation function C(t). ii Phase-space points. namely. VII we demonstrate the application of Fourier transform and filter-diagonalization26 –29 methods to calculate the energy spectra from classical approximations to autocorrelation functions. 4 as ˆ ˆ Weyl transform of A T P V ˆ long as A V includes the potential operator commuting with coordinate representation.p. yet the calculations differ in the effort required. see http://jcp. Downloaded 14 Jan 2005 to 132. 7 ˆ is obtained for A being the initial density operator. The obtained trajectories. which in turn are accessible by the WPS approach using Eqs. The main importance of obtaining the autocorrelation function is that it can be used to obtain the energy spectrum. VI . where AV is the ˆ ˆˆ Wigner-Weyl transform of A V P V and AT is the Wignerˆ T . 10 q * q where F is the system dimension.pi . 1 October 2004 ˘ˇ´ ´ P. represent the time-dependent classical phasespace density P(t). are evolved in time by classical propagation. It is convenient to calculate separately the potential and kinetic contributions. qi .29.25 in Secs. which are approximated by Eqs. Chem. we calculate the Wignerˆ Weyl transform AH of the operator A H using Eqs. No. VIII we present our conclusions.pi t . defined as ˆ A t e i/ ˆ Ht ˆ ˆ ˆ e iH t/ P H e ˆ iH t/ C 2 .68. ˆ A (t). qi (t). Zdanska and N. The survival probability.. the survival probability is approximated within the WPS method by C t 2 i P qi t .p j ). 6 Following this. We demonstrate our techniques for the model PullenEdmonds Hamiltonian of two nonlinearly coupled harmonic oscillators24. q/2 .

pi t i P qi t . 1 we obtain the corresponding Wigner distribution given by.23 The solution of this difficulty is addressed in Sec. 8 . Phys.pi t i P qi t . therefore the WPS approximation to autocorrelation amplitude is exact in the case of harmonic oscillator. AV i n 0 n In this section we discuss inconsistent expressions for autocorrelation function.pi t P qi t . 15 and we obtain the series for AV given by. 10 . Let us demonstrate this point for the case of autocorrelation amplitude. 10 such that. Inconsistency within WPS approach: The harmonic approximation This result is substituted to Eq. doing this and using Hamilton’s equations of motion we obtain the expression. is by differentiating Eq. 21 The phase of the autocorrelation function is calculated by numerical integration using the phase derivative over time. d qe i/ p• q q P q/2 * q q/2 qn n i n pn . 13 . is applicable also in the case of autocorrelation amplitude. 26 where the harmonic Hamiltonian is given by. 11 –using Eqs. C 1 i RAH t qi t . the phase is known for individual recurrences only up to a constant phase factor. Similar discussion is possible for the time derivative of the autocorrelation phase. we obtain ln C t 1 i IAH d nV dq n n P n 2 n! n p . H p2 /2 q2 /2. An alternative to the WPS approximation. 21 . ˆ H i AH qi t . when only the first term of the series for IAH in Eq. see http://jcp. and 19 we obtain the WPS approximation of the nominator. A. 23 is obtained from Eq. the phase derivative between the recurrences cannot be calculated for the reason of numerical inaccuracy.org/jcp/copyright.aip. 23 is obtained from Eq. 18 IAH 1 2n 1 ! By comparison of Eqs. Eq. 17 . 6 . 6 . C 2 . which coincides with the autocorrelation function at a single point within each recurrence. 25 results in values of C / t which are incorrect for harmonic oscillator. while Eq. 1 exp F/4 q q0 2 2 i q"p0 .J. Vol. In order to clarify the role of the quadratic term in calculation of the phase derivative. Eq. 19 . 8 is used for the denominator. The existence of different formulas for same quantities. Eq.29.g. dp q 24 1 2 qi t . using of Eq. all consistent with the WPS assumption Eq. AT i n 0 n d nT dp n n P Another way to obtain the time derivative of the autocorrelation amplitude. by comparison of Eqs. No. 18 .. 20 The phase of the autocorrelation function is calculated using Eq. 22 ˆ The Wigner-Weyl transform of A T is obtained in the same manner using the momentum representation. One can see that Eq. 23.jsp . Downloaded 14 Jan 2005 to 132. 16 qi t . 25 is given by the linear approximation to Eq.pi t i . represents the inherent error of the WPS approximation. 13. 2 n n! dqn qn d n V 14 correlation function.pi t iA 2 n! n qn . C iH 2n d 0 4 V 1 2n 1 2n 1 P 1 d 2n 1 T 1 2n 1 P 1 t . 25 dq2n p2n dp2n q2n ˆ We obtain the WPS approximation for H by substitutions into Eq. where we define and evaluate a cross- The proof of Eq.pi t P qi t . As a consequence.pi t P qi t . In cases where the autocorrelation function consists of well separated recurrences see. 25 can be found in Ref. 25 is obtained from Eq. 19 is included. is the expression given by.pi t . Fig. 22 and 23 one can see that Eq. IV. 5 . when only the first term of the serie for RAH . 21 . 19 qi t . let us consider the case of a coherent Gaussian wave packet given by. 23 ˆ The Wigner-Weyl transform AH of A H is a complex function given by the series for its real and imaginary parts. e. 21 and 25 one can see that Eq. 2 n RAH n 0 4 d 2n V dq 2n 2n 1 2n ! P d 2n T dp 2n 2n P p 2n 2 n q 2n .pi t i . We use the following property of Fourier transform.68. Chem. ln C t where A dV P dq p dT P . Eq.. 21 . The derivation of the phase derivative. 2 . Therefore in such cases. Redistribution subject to AIP license or copyright. is included.1. 22 . Eq. which are derived using different algebraical approaches. On the other hand. 22 using the quadratic approximation. 121. Using Eq. Again.pi t i .pi t P qi t . 1 October 2004 Energy spectrum by classical trajectories 6177 V q q/2 n 0 .

pi T i P qi T . the periodic contour can be arbitrary.p i (t) E 0 E. 121. in contrast to the most common definition of crosscorrelation function. 10 and 11 .6178 J. to give a large preference to those contributing to a given autocorrelation recurrence. however.. 25 . The idea is to define the new function (t) such that (t NT) and therefore. Let us consider a quasiperiodic motion with the approximate period T. It is evident from this discussion that the second-order term is responsible for one half of the zero-point energy shift in the case of coherent harmonic motion. ii The second condition is that the contour must form a closed loop..68.pi t P qi t .1. Chem. Classical periodic contour F 2 1 2 p p0 2 q q0 2 . 2 32 which is in aqreement with the quantum-mechanical definition given by Eqs. RAH PH PH (2) . 27 The Hamiltonian H and the Wigner distribution P are substituted into Eq.pc (T) . RAH can be written as the sum of zero-order and second-order terms. Further we confine to trajectories falling within a small energy interval E 0 E H q i (t). here we suggest the following procedure: i positions and momenta of evolving trajectories are weighted by P qi (T). It is thus expected that an error of about one half of the zero-point energy is caused by omitting the quadratic term in series Eq. Then we may assume that a recurrence in autocorrelation function occurs for time t T. (t) is given by t 1 exp F/4 q q0 qc 0 2 pc t i qc t 2 The idea is to define (t) in such a way that there are no revivals in X(t) and therefore unlike C(t). 2 29 The contribution of the second term is calculated using the equation.. which is accomplished by using the back-andforth Fourier transform: qc t exp n T i 2 nt T 2 nt T qc0 t . where H (2) is given by H 2 packet moving according to the classical periodic contour. Moiseyev P 1 exp F q q0 p p0 .and second-order contributions leads to the correct result. Phys. 34 that represents a frozen wave packet which keeps the shape of the initial wave function while its position is changed according to the classical periodic contour. qc (t). qc0 (t).p i (t) E 0 E and where i 1 for E 0 0 otherwise. 18 . X(t) does not drop to zero between the recurrences. C/ 28 t. (NT) .29. X(t) represents a cross-correlation function. 36 dt exp i 0 In this section we show how over-all phase factors of autocorrelation recurrences can be obtained. Zdanska and N. B. CROSS-CORRELATION FUNCTION E H q i (t). etc. Redistribution subject to AIP license or copyright. notice that in this case the left vector (t) depends on t parametrically. such that qc (0).pi t i P qi t ..pc0 (t) satisfy the condition of api proximately following the classical phase-space packet.pi T i i . 33 where X(t) is defined as X t t e . The first estimate of the classical contour according to this procedure is given by qc0 t pc0 t i qi t H 2 F . 37 N 0. The sum of the zero. A.. is given by C t H F 4 q0 p0 .pi (T) . 4 31 t P qi T .pc (0) qc (T). this is especially reasonable in cases when the wave function bifurcates due to time evolution.. Except for approximately following the motion of the classical phase-space density. periods the values of C(t NT) are equal to X(t NT). 18 .pc (t) . it is reasonable to define (t) as the frozen wave i q• p0 pc 0 p0 • qc 0 qc t 38 Downloaded 14 Jan 2005 to 132. after T.pi T t P qi T .pi T i P qi T . C i pi i i t F 2 q0 p0 . 2T. see http://jcp. ˆ iH t/ where n acquires small integer values pc (t) is obtained similarly. In this way we avoid numerical difficulties in extracting time-dependent phase derivatives between recurrences where C(t) 0.1. 13. 1 October 2004 2 2 ˘ˇ´ ´ P. C iH H 2 2 t qi t . Vol. 30 and is given by C We define an approximate classical periodic contour orbit . qc (t). No..pc (t) . Frozen wave packet We define a wave function t e i/ pc t ˆ pc 0 q (t) given by qc t ˆ qc 0 p e i/ .pi t . To assure this condition.org/jcp/copyright.aip.jsp . as de- The contribution of the zero-order term to fined by Eq. 35 . IV.. C NT X NT . R. The wave packet (t) acquires the important property.

68. Vol. 26 . 43 V q 2 /2 q 2 /2 1 2 T 44 p 2 /2 p 2 /2.t . C.p. 46 It represents two nonlinearly coupled harmonic oscillators. The initial wave packet 0.pi t iP iP qi t . as long as it does not include revivals nor drops to zero. 1 2 q 2q 2.u. 1 2 48 ˆ where the notation a ˆ a ˆ ˆ ˆ H . which is given by Eqs.. the system For energies beyond a given energy limit becomes mixed regular and chaotic. The wave packet. this is not true for the autocorrelation function C(t) which can be evaluated from classical trajectory calculations for individual recurrences only up to the overall phase factors. The quantum-mechanical propagation.t d qe i/ p• q 2 i P qi t . The initial position of in the phase space is q0 4 a. from the calculated classical cross-correlation function. we derive the expression for the time derivative of X given by X To summarize. 33 . The system is illustrated in Fig.957 a. 6 is correct. but stays localized. is evaluated using P ˆ Eq. 13. Aq . .org/jcp/copyright. As discussed before. p0 0. P a . the WPS approximation given by Eq. 26 . Redistribution subject to AIP license or copyright. where all trajectories acquire the same Maslov indices at a time. 7 for this time period. ˆ The WPS approximations for the integrals R H ˆ and R p . 1 October 2004 Energy spectrum by classical trajectories 6179 for the case of initial Gaussian wave packet defined by Eq. 47 Calculation of complex cross-correlation function X(t) by the WPS approach is similar as in the case of the autocorrelation function. ˆ P t t t . X 2 . as shown in Ref.05. 40 . Thus the missing overall phases of autocorrelation recurrences are determined and the classical complex autocorrelation function is obtained. the energy limit is given by the Gaussian function. for one classical period. Phys.. and the denominator. 23. qi t .pi t qi t . for three classical periods. 42 By differentiating the identity X(t)/X * (t) exp 2i X(t) . 45 The Wigner-Weyl transforms. 43 and 47 . q/2. AH .. 2T.R q . P p2n q2n .q. Once this condition is satisfied. as shown in Fig. stands for . is evaluated using Eq.29. while the phase X is obtained by numerical integration of its time derivative. 2. we obtain the correct phase of the autocorrelation function at T. 40 .pi t qi t . The corresponding energy expectation value is 21 a. a ˆ iH t/ The Pullen-Edmonds Hamiltonian is given by pc t . and Ap . 6 .u. 23.t . ILLUSTRATION FOR REGULAR SYSTEM 1 t ˆ R H ˙ ˆ q• R p ˙ ˆ pc •R q pc 0 ˆ . ˆ a i Aa iP qi t . This is approximately the Ehrnfest time for the studied system.pi t . 39 ˆ R q ˆ R p qi t . of the cross-correlation function is deX t X t ei . V.aip. as shown in Ref. The prevailed motion for this wave packet is regular. 1.J.0 . 121.jsp . The autocorrelation function C(t) includes three well distinguished recurrences.pi t .pi t pi t . approximately. where A ˆ Aa is given by the Wigner-Weyl transform of A a ˆ a . i P qi t . 2. deforms and starts to develop nodes.u. The crosscorrelation function X(t) can be evaluated numerically in the time interval from t 0 to t NT.pi t qi t . X / t. which is evolved. are given by ˆ R H i RAH iP iP ˆ . Downloaded 14 Jan 2005 to 132.. . The absolute value X is obtained using Eq.t 41 * q The phase. No.pi t . ˆ ˆ ˆ e iH t/ P a e X 2 ˆ The WPS approximation to a is obtained as follows: the ˆ ˆ nominator. allows us to calculate the exact autocorrelation function Eq. which is set far in classical mixed regular-chaotic regime. Eq.u. are complex functions whose only real parts are used to calculate the cross-correlation phase Eq. 43 . These are listed below: 2 RAH P H n 1 2n 4 P 1 2n ! d 2n T dp2n 2n d 2n V dq2n RAq RAp P q... 10 a. The contour plot of potential illustrates its nonlinearity which displays itself by deviation from circular curvature. The absolute value square X 2 is obtained as the expectation value of the density operator. which serves as the benchmark for our study. Chem.1.p . P p. Absolute value and phase of cross-correlation function qi t . fined as X t X. For the chosen. see http://jcp.23 Using the equivalence Eq.pi t such that X t where P q. the complex cross-correlation function X(t) is calculated using the WPS approach. 40 q q/2. We confine our study to a short-time propagation.

3 and 5. qi t . cross-correlation function is a means to obtain the phase of the autocorrelation function at time points t NT. 5. where the two functions coincide. Classical trajectory calculation of cross-correlation function The classical trajectory simulation is performed for an ensemble of trajectories. 34 – 38 . The classical periodic contour is calculated using Eqs. Phys. The values for t T. see http://jcp. The time derivative of X is calculated from the ˆ ˆ ˆ integrals R H . we evaluate the crosscorrelation function using quantum mechanics. 46 and 47 .1. 2. Let us discuss in detail the classical approximation of the quantity R H in the light of the results obtained in Sec. 49 The absolute value as obtained from classical trajectory calculation is shown as part of Fig.2T. t 2 2. The phase of the cross-correlation function X is calculated using numerical integration of the time derivative. The Pullen-Edmonds potential is drawn by dashed contours. 50 where A H H H (2) .jsp . is the sum of the classical Hamiltonian H and the quadratic term H (2) .68.3T are emphasized by dots. and R p . FIG.6180 J. No. The comparison of quantum and classical results for X 2 . Clearly.u.u. The values at t T. the WPS approximations for R q ˆ and R p given by lines are in an excellent agreement with the quantum results given by points . The wave packet is shown at times t 1 0. 41 . For an illustration.pc (t) . In the classical trajectory method. Their quantum definition is given by Eq. 13. which approximately represents the classical periodic motion of the phase-space density. 3. This property is essential for its accessibility by the classical trajectory method. The cross-correlation function X(t) is defined by Eqs. Zdanska and N. X (t) does not climb down to zero. P . 1. P is defined by Eq. One-period wave packet evolution. P 1 q q0 qc 0 2 exp F qc t 2 The definition of the cross-correlation function is based on a notion of classical periodic contour orbit . These points are emphasized in both Figs. which is addressed below.. 38 to Eq.5 a. A. ˆ 4 a and 4 b .65 a. The evolved wave packet intersects the closed contour Eqs.pi t i P qi t .aip. which are chosen by means of preferential sampling of the initial phase-space density. which serves in the definition of the cross-correlation function. 1. 43 . Vol. The Wigner-Weyl transform of . nor creates recurrences. Downloaded 14 Jan 2005 to 132. 2–3. X / t. Eq. The quantum and ˆ ˆ classical results for R q and R p are displayed in Figs. Figs. 1 October 2004 ˘ˇ´ ´ P. 27 . The absolute value and the real part are denoted by black and gray lines.5 a. where the cross-correlation function coincides with the autocorrelation function. the p p0 pc 0 pc t . Chem.29. Fig.u. The absolute value and the real part are denoted by black and gray lines.pi t . The latter is given by. 47 . Cross-correlation function The absolute value of cross-correlation function is evaluated using Eq.2T. 44 .u.pi (t) . with the period T. Unlike the autocorrelation function.. is obtained by substitution of Eq.3 a.. 35 – 36 . 121.org/jcp/copyright. 3. qc (t). respectively. leads to the expression R H iA H FIG. III A.pi t ]P qi t . 35 and 36 . 40 . B. P. qc (t) is illustrated in Fig. R q . Redistribution subject to AIP license or copyright. R. respectively. are emphasized. The substitution of Pullen-Edmonds Hamiltonian Eq. qi (t).u. shows an excellent agreement. while their WPS approximations are given by Eqs. Moiseyev FIG. t 3 3.3T . Cross-correlation function by quantum propagation for the time interval 0 t 11 a. 48 to Eqs. Autocorrelation function calculated by exact quantum propagation for the time interval 0 t 15. and t 4 4.

err class quant /2 . R q . 6. To adjust the overall phase factors.440 0. solid line—classical and R p 2 ˆ triangles—quantum. 8 . The absolute value and the real part are denoted by black and gray lines. energy spectrum corresponding to the linear ap/(3T) proximation includes the error estimated by.967 0.226 0. The quantitative comparison of the quantum and classical cross-correlation functions is given in Table I. close to the quantum benchmark. FIG. In Fig. c R H circles— quantum. Redistribution subject to AIP license or copyright. 44 calculated by ˆ circles—quantum. A H H H (2) . A H H H (2) .u. 0. Phase of the cross-correlation function at points t T.304 ˆ ˆ ˆ FIG. which are used to obtain the phase of the autocorrelation function.1. Autocorrelation function obtained by classical trajectory calculations for the time interval 0 t 15. which appear in each classical period T. and A H can be substituted either by Hamiltonian only. 4.org/jcp/copyright.543 0.68. 5. quant –exact quantum values. as discussed in Sec. No.428 0. is evaluated from C / t by numerical integration and is known for individual recurrences up to overall phase factors.70% H class 0. t T 4. and R H . err-error of the classical values. class. It is clear that class is a H good approximation to quant . or by a series. which are obtained from the phase of the cross-correlation function.45% 0. solid line—twoterm classical approximation .u. Phys. this result is in agreement with the role of the quadratic term. see http://jcp. A H H. b R p 1 circles—quantum. where we list the phase of the cross-correlation functions at points t T. The absolute value and the real part are denoted by black and gray lines.. C . 13.jsp .2T.29. 6.60% 8. quant is the phase as obtained by quantum calculation. Classical trajectory calculation of autocorrelation function H 2 F 2 1 2 F f p p0 F 2 q q0 /2 pf q2 f 2 q2 f pf0 2 . As shown in Fig. The points t NT. it is necessary to determine the phase of the autocorrelation function for at least one time-point per FIG. respectively. C. are emphasized. respectively. defined by Eq. Convery large. are emphasized. 121. The points t NT.295 0. R p . Downloaded 14 Jan 2005 to 132. The classical complex cross-correlation function is displayed in Fig. the WPS approximation for H is not unique.428 0. M class –classical quadratic and linear approximations. class includes the error sequently. III A.u. 5. 4 c we show these two approximations lines compared with the quantum result points .5 a.3T . a R q 1 ˆ solid line—classical and R q 2 triangles—quantum. given by Eq. dashed line—one-term classical approximation. This value is close to the half of zero-point energy. results. where the cross-correlation function coincides with the autocorrelation function in regular regime.602 0.255 3T 13. the autocorrelation function includes three recurrences.3T . quantum and classical trajectory methods. Vol. dashed line— ˆ ˆ classical . The contribution of H (2) is relatively small due to high expectation value of energy compared to the zero-point energy. while the error of class is H 2 at t 3T. We can conclude that both approximations provide similar The absolute value of the autocorrelation function C 2 (t) is obtained using Eq.45 a. III A. H and class is the phase as obtained by classical calculation in the linear approximation.625 2T 9. 21 .J. respectively.885 quant class err 3. Cross-correlation function obtained by classical trajectory calculations for the time interval 0 t 11 a. We are able to evaluate the time derivative of the phase C . using classical trajectory calculations. The phase. dashed line—classical . 1 October 2004 Energy spectrum by classical trajectories 6181 TABLE I. Chem. A H H. 51 f As discussed in Sec. class is the phase as obtained by classical calculation in the quadratic approximation.2T.aip. H (2) . represented here by the sum. in the case of harmonic oscillator.

Phase of the cross-correlation function at points t T 1 . 13.. its large portion occupies chaotic sea. shorter than expected. 27 is propagated. the complex autocorrelation function obtained entirely from classical trajectory calculations is shown in Fig.80 11.. 35 and 36 .u. the phase error FIG. indicating a bifurcation to three different quasiperiodic motions. 7.u. It is assumed that this error is caused by the interference of trajectories with different Maslov indices the Ehrnfest criterium is not perfectly satisfied . 6.jsp .u.. However.179 0. three small recurrences are found. recurrence by another calculation.3T are listed in Table I.u. Phys. 8. It is clear that the classical calculations provide a good approximation to the chaotic autocorrelation function with the exception of the second recurrence. . 33 . we use the same type of Hamiltonian Eq.u.T 5 .. VI. 7.16% 0.T 5 . Fig. Moiseyev FIG.592 0.T 5 for individual quasiperiodic motions are emphasized.2T. 1. Vol. In Table II we list the classical and quantum phase at time points t T 1 .0508 0.57 14. ILLUSTRATION FOR CHAOTIC REGIME Application of the new method in chaotic regime is possible with a small modification. where the initial Gaussian wave packet is centered at nonzero momentum position. on the other hand.65% 4.27 10. Quantum autocorrelation function for chaotic regime. The first recurrence appears for t 5 a.427 0. 48 .u. therefore recurrences are multiplied and appear at irregular time spacing.. TABLE II. as shown in this section. as can be shown with the ´ help of Poincare surface of section. where C (NT) X (NT). therefore the phase of each recurrence is obtained using a different cross-correlation function. R. Five cross-correlation functions are displayed. causing a sequence of equally spaced recurrences.87% 68. At t 14 a. The bifurcations of the time-evolved phase-space density must be reflected by different periodic contours for each recurrence. In place of a second recurrence at t 10 a. These points are determined by the cross-correlation function for the time points t NT. The total phase of recurrences is obtained for each recurrence separately. No.T 3 . For illustration.. Redistribution subject to AIP license or copyright. The chaotic motion.31 quant class err 1.. quant –exact quantum values.645 a. we define five periodic contours according to Eqs... is characterized by bifurcations of the phase-space density. Classical cross-correlation functions for chaotic regime. class –classical quadratic approximation...164 0.. we could identify a quasiperiodic motion with a period T. 4 a.277 0.294 0.u. and p0 0. err–error of the classical values.5 a. where the error of nearly 70% is obtained.. the period of which corresponds to three approximate classical periods. but the recurrence time is by 1 a.52% 5.377 Downloaded 14 Jan 2005 to 132..32 9.68. t T1 T2 T3 T4 T5 5.29. Within the time interval 0 t 15. see http://jcp. Each recurrence in the autocorrelation function is created by a different quasiperiodic motion.. there is a large recurrence. with periods T 1 . Zdanska and N. 8 a – 8 e .6182 J. Ref.11% 7... This becomes evident after a close look at quantum autocorrelation function.77 0. 23. 1 October 2004 ˘ˇ´ ´ P. Thus. In regular regime.141 0. When the corresponding phase-space packet Eq.org/jcp/copyright. Eq.. Classical periods T 1 .. from cross-correlation functions of their own periodic contours. Chem.u. here we choose q0 4 a.1.aip. err class quant /2 . where the cross-correlation functions coincide with the autocorrelation function in chaotic regime. The corresponding five classical cross-correlation functions are displayed in Figs. 121. The particular phase values at points t T.

u.aip. The spectral interval 18. a regular regime. Vol. VII. 9. III with no alteration for the chaotic regime. Moderate resolved spectra as obtained by Fourier transform of the classical gray and quantum black autocorrelation functions for a small energy interval. The evolution and overlap matrices are given by. 10 a and 10 b . 11 a and 11 b .. Chem. 10. 54 dt dt e i Ek E k t/ C t t . Redistribution subject to AIP license or copyright. 52 . t max k 0 dte ˆ i H E k t/ 53 are used as basis functions to find the eigenvalues of the ˆ ˆ evolution operator. which are shown by vertical lines.68.. U ( ) e iH t/ . is the time of the propagation. of energy grid points E k . has been calculated.29. t max 15. The actual energy levels can be obtained even from short-time propagations with the help of filterdiagonalization method.8. 1 October 2004 Energy spectrum by classical trajectories 6183 FIG. U kk S kk t max 0 t max 0 dt dt e i Ek E k t/ C t t .T 5 . 52 Limitation of our method to short-time propagation allows evaluation of moderate resolved energy spectra using Eq. for which the autocorrelation function 0. a regular regime. ENERGY SPECTRUM FROM CLASSICAL AUTOCORRELATION FUNCTION FIG.. is shown in Figs. Phys. The classical approximation is in qualitative agreement with quantum spectra in both regular and chaotic cases.jsp . see http://jcp. The actual quantum energy levels and Franck-Condon overlaps are shown by vertical lines.u. The agreement between the classical and quantum results Fig. does not take a great effect in the corresponding energy spectrum due to the relatively small amplitude of the second recurrence. Moderate resolved spectra as obtained by Fourier transform of the classical gray and quantum black autocorrelation functions for the whole range of occupied energies..5 a.u. The classical autocorrelation function for chaotic regime is shown in Fig.u.1. The present implementation of the filter-diagonalization method is described below: i The Fourier transforms of the time-dependent wave packet.5 a. FIG. The predominant role of the autocorrelation function in most calculations is that its Fourier transform gives energy spectra together with the corresponding Franck-Condon amplitudes I(E) as given by the equation. 11.u. 121.org/jcp/copyright. 9. at points t T 1 . Classical autocorrelation function—chaotic regime. The absolute phases of individual autocorrelation recurrences are obtained from the corresponding cross-correlation functions. k 1¯160. ii The basis functions k are defined by energy grid points E k . which are shown in Figs. No. b chaotic regime. we can see that the spectral peaks appear near the occupied energy levels.. The energy window width is given by the standard deviation 1. E 23.26 –28 Here we apply the filterdiagonalization method to classical autocorrelation functions in regular and chaotic regimes in order to demonstrate the ability of the WPS approach to approximate the exact quantum energy levels.. The precision of the calcu- Downloaded 14 Jan 2005 to 132. is a short-time interval. I E R 0 dt exp i/ Et C t . b chaotic regime. 7 is remarkable.J.67 a.05 a. Calculation of the classical autocorrelation amplitude and time derivative of the autocorrelation phase is done according to Sec. 13.5 a. which gives an optimal convergence of the filterdiagonalization method. as shown below. which are selected to follow a bell distribution around the energy window center E 0 .

aip. without further modification the classical Wigner method is not competitive with methods based on the VVG van VleckGutzwiller propagator expression. 1 October 2004 ˘ˇ´ ´ P. E quant –quantum results.700 21. –absolute difference between quantum and classical values.386 22. E quant –quantum results. 12 b .org/jcp/copyright. and proves to work in good quantitative agreement with quantum benchmarks.0006 0.u.u.530 22.0064 0.002 TABLE III.010 0. Chem.03 a.0005 E quant 19. Vol.0180 0. Energy levels obtained by the filter-diagonalization method from the quantum and classical short-time autocorrelation functions in regular regime.0015 0.1399 21..0145 0.5 a.0650 19.017 0.018 0.814 23.309 0.123 21.0025 0. Phys. 12.5 a.019 0.894 19. E 23. We formulate an approximation of complex autocorrelation function which is fully consistent with the Wigner phase-space approach.8121 22.018 0.. The finite width of peeks is given by the standard error calculated for multiple filterdiagonalization calculations.0224 0.01–0.03 a.0874 19. which arise using the presented method Ref.2 a. falling into interval 18.0001 0. Redistribution subject to AIP license or copyright.016 0. respectively. The obtained energies are listed in Table 18.. E class –classical results..023 0.032 0.0252 0.0012 0.8181 20. E class E quant .015 0.4288 21. a regular regime. No. as obtained by filter-diagonalization method bottom compared with Franck-Condon amplitudes top .022 0.005 0.020 0. b chaotic regime. namely.4512 21. E class E quant .0008 0.307 0.014 0.0021 0.2182 23.5291 23.0299 0. 23 .575 18. E class E quant 18.026 0. uses the rigorous quasiclassical expression for expectation values. We attempt the calculation of exact energy levels using the filter-diagonalization method in chaotic regime for the energy interval 18.020 0.021 IV and displayed in Fig. 19 a.015 0. which is an error significantly smaller than energy spacing between populated levels.’’ Italics added. in our experience the method is accurate only through intermediate times.0016 0. The seven significantly populated states are extracted both from quantum and classical autocorrelation functions. Zdanska and N.u. E class –classical results. 121. Moiseyev TABLE IV. –absolute difference between quantum and classical values. are listed in Table III and displayed in Fig.u.037 0.027 0.01–0. The method is tested for a model of two nonlinearly coupled harmonic oscillators for regular and chaotic regimes.557 18.361 19. The lost phase factors are obtained by calculation of a crosscorrelation function which serves as a link between t 0 and t T.4992 23.0007 0. VIII.036 0. We suggest a way to overcome a problem of lost overall phases of separate recurrences. E 0 23.9074 0.807 20. We may thus conclude that the WPS approximation has been capable of prediction of energy levels in regular regime with the absolute error being 0.024 0. Energy levels.u.778 20.041 0.999 22. We demonstrate that the Wigner phase-space approach is capable of prediction of energy levels of the given nonlinear system in regular regime with a very good precision.0007 0. Using the approximate complex autocorrelation function we attempt to calculate the energy spectrum of the system with the help of filter-diagonalization method. The standard deviations are of orders 10 3 – 10 4 a. and one of the main challenges of this approach can be expressed in words of Garashchuk and Tannor.0007 0.0323 0.5 a.8771 0. 12 a .027 0.21 ‘‘Although we have been able to calculate the phase to the same accuracy as the amplitude.8266 22.328 19. E 23.406 22.031 0.0224 0.212 20. Energy levels obtained by the filter-diagonalization method from the quantum and classical short-time autocorrelation functions in chaotic regime..5 a.u.248 20.025 0.u. The energy levels found in regular regime.u.29.1.7858 20. which signifies the incomplete representation of Hamiltonian due to short-time propagation. 13.jsp .2118 23.1651 21.0014 0.029 0. One can see that the standard error of the filter-diagonalization method 0. The quantum and classical results are marked by black and gray color. where T is the approximate classical period. R. We propose a quasiclassical method for calculation of complex autocorrelation functions within Ehrnfest time limit. however due to small energy spacing of occupied states the energy levels cannot be obtained.028 0.025 0. It is thus characteristic that the standard error of the obtained energy levels is almost one order of magnitude larger in chaotic case compared to regular regime.68.924 19.708 21. The error given by short-time propagation is also by one to two orders of magnitude smaller than the difference between the quantum and classical calculations given by the third column in Table III .103 21. E class 19. CONCLUSIONS FIG. One of the main obstacles for practical application of the Wigner phase-space method.008 0.003 0. The solution may hint toward separation of autocorrelation recur- Downloaded 14 Jan 2005 to 132.015 0.027 0.u.016 0.0103 0.0020 0.6184 J.020 0. is often larger than the energy spacing of the occupied levels in chaotic regime. see http://jcp.519 21.011 0.010 22.830 23.0303 lated energy levels is estimated as the standard deviation of few three to eight filter diagonalizations performed for different values of E 0 .

Y. B. which is funded by the Czech Ministry of Education Grant No. 324 1996 . Lett. 26 D.S. Narevicius and N. D.aip. 263. A. 1583 1999 . R. and L. Manopoulos. Kisil. J. Z4 055 905. B. S. U. 102. Chaban. Zdanska. M. 3578 1970 . 20 P. Phys. Prezhdo and V. 28 V. Manthe. Sun. Phys. J. 93. J. Chem. 12229 2002 . 6046 1995 . Chem. Chem.-D. J. Hanz. B. Gerber. Vol. Lett. Pang and D. Chem. R. A 56. 1979 1967 . A 105. H. Neuhauser. Jungwirth. 23 P. No. S. Phys. Phys. Korolkov. J. Phys. Horenko. 287. J. and N. 5733 2001 . S. 52. Slavicek. 19 E. Chem. Miller. H. Downloaded 14 Jan 2005 to 132. Meyer. 99. Lett. and C. Chem. J. J. Chem. 6246 1999 . Sci.org/jcp/copyright. B. J. Worth. Phys. Moiseyev. J. 3274 1997 . P. 12 O. Korsch. B. J. Phys. C. and D. Natl. Chem. Wang. Koppel. Phys. 250 1998 . Acad. Phys. and W. 65. 110.J. Phys. E. L. Rev. 13. Schmidt. Phys. Phys. Chem. Chem. 280. Chem. Chem. Rev. Washington. 14 J. 16 M.-D. J. A 14. 178 1928 . Zdanska and N. 53. Phys. Bargheer. Chem. Chem. 351 1996 . Tannor. Niv. Jungwirth. J. Chem. Rev. Zhao. Heller. S. Schmidt. H. and N. Neuhauser. 250 1997 . 9 M. 121. Phys. A 106. M. Zdanska. Cederbaum. Taylor. 7 M. Lett. 4190 1998 . H. 2611 1990 . 8 E.jsp . Chem. J. Garashchuk and D. Chem. Lett. 73 1990 . W. Phys. H. Moiseyev. 252. 106. Redistribution subject to AIP license or copyright. Van Vleck. S. A. 1 October 2004 Energy spectrum by classical trajectories 10 6185 rences using the calculation of their Maslov index. Yourshaw. Phys. J. Y. Chem. 4 P. Chem. 177 1997 . 21 S. 6 G. Proc. 1665 1984 . and P. 105. Phys. J. 78. Moiseyev. Edmonds. Heller. Narevicius. 13 I. M. 27 J. Phys. Lett. 165. 117. Europhys. Chem. 29 E. J. Neumark. Lett. Korsch. 17 J. Moiseyev. Chem. V. 2770 2001 . Schutte. Rev. and completed within the framework of the research project Grant No. Cederbaum. Haller. and L. 276. and P. ACKNOWLEDGMENTS This work has been supported by the Center for Complex Molecular Systems and Biomolecules. Jungwirth. Gerber. U. M. 15 W. Phys. Phys.68. 4412 1996 . 75. Roeselova. 112. the preliminary discussion of this problem is found in Ref. 115. J. Cederbaum. Phys. 11 G. 1 P.29. Brewer. Phys. J. Schmidt. Neuhauser. 2 I. and B. M. Jungwirth. and N. H. 14. 290 1998 . 24 R. Jungwirth and R. Schoendorff. D. 186 1981 . Chem.. 5 2002 . B. Jungwirth and R. Meyer. B. see http://jcp. 1289 1976 . V. H. L477 1981 . LNO0A032 . Phys. and D. Lett. J. 8. Pullen and A. and L.1. Chem. 105. 109. J. and P. Phys. H. Schmidt. Schmidt. Gerber. 173 1996 . Phys. H. 10608 2001 . Phys. 162 1997 . Mandelshtam and H. 3 P. Hulme. J. Moiseyev.C. V. 5 P. Chem. Lett.A. Gutzwiller. 10761 2000 . Phys. Miller. Phys. Schwentner. 115. 44. and N. 22 X. L. Guhr. J. 18 E. Mucha. 25 E. 23. 4832 1997 . Math. J.

Sign up to vote on this title
UsefulNot useful