Dissipation in Quantum Mechanics

 

Robert Dawes[1] and Ajay Patrikar[2]

 

 

Abstract - In this paper we propose a possible solution of the geometric quantization problem for dissipative systems and present the results of numerical simulations for the damped  harmonic oscillator. These simulations yield close agreement with the classical system, but only with an unexplained factor of 1/p that begs for a better derivation from the first principles. Our solution relies on the introduction of a nonlinear term into the potential energy field in a manner reminiscent of the way in which soliton solutions are obtained from the Schrödinger equation.

 

 

I. INTRODUCTION

 

The geometric quantization problem is the problem of obtaining quantum mechanical equivalents of given classical systems [1].  The quantization of systems subject to an external force has been a topic of investigation in physics literature [2]-[3].  Recently there has been considerable interest in quantization of control systems [4]-[5].  A standard approach to quantization is to construct a Lagrangian of the system from its dynamical equations, obtain the corresponding Hamiltonian, and use it in the Schrödinger equation [3, 5].  That approach, however, does not apply well to dissipative systems.  Ray [3] has shown that the quantum representation obtained using that method does not accurately describe the original dissipative system.  For example, when the procedure is applied to the damped harmonic oscillator, Ray shows that the resulting Schrödinger equation describes instead a non-dissipative system with time-varying mass m(t) = m0eat rather than a system whose amplitude damping results from the dissipation of its energy. Apparently, the proper treatment of dissipation in quantum mechanics remains an open question [2].

 

 

II. NONLINEAR SOLUTION

 

We have obtained a quantized representation of the damped harmonic oscillator not by starting with the Lagrangian for that system, but by incorporating nonlinear viscous damping into the potential energy field. We begin with a nonlinear Schrödinger equation

 

                                                         (1)

 

in which the potential energy U(x,Y,t) is the sum of two parts,

 

                                                                                            (2)

 

If  l = 0, and  V(x,t)  is given by

 

                                                                                                    (3)

 

then the controlled harmonic oscillator, whose classical dynamical equation is given by

 

                                                                                                                  (4)

 

results [5].

 

Let Q(Y) be an arbitrary quadratic form on Y.  The properties of a nonlinear Schrödinger equation in which Q(Y) is given by Q(Y)=G(|Y|2), for some function G, are described by Bialynicki-Birula and Mycielski [6].  We consider an equation in which Q(Y) is a flux potential for Y.  That is, for any given Y, if

 

                                 (5)

 

is the probability current density associated with Y, suppose that Q(Y) satisfies

 

                                                                                                              (6)

 

A potential term in this form induces a component of the Ehrenfest force that opposes the flow of the wave function in proportion to its current density. The coefficient l in (2) then determines the damping rate of the system. The properties of a nonlinear equation obtained with this choice of Q(Y) are studied through simulation in the following section.

 

 

III. SIMULATION RESULTS

 

We now compare the response of the classical damped harmonic oscillator with the response of the nonlinear Schrödinger equation obtained through numerical integration.  The equation of the classical system is given by

 

                                                                                                    (7)

 

where x is the damping constant, w is the natural angular frequency and f(t) is the external input.  The response of this system for x(0)=0.5, x = 0.1, w = 100 and f(t) = 0 was obtained analytically and is plotted in Fig. 1.

 

A numerical method proposed by Goldberg, Schey and Schwartz [7] was used to integrate the nonlinear Schrödinger equation. This method has been proven to be stable, unitary and second-order accurate in space and time.  The equation was integrated over a lattice of 1000 points in a spatial domain spanning the interval [-2, 2] leading to a spatial quantization step  e of 0.004.  The spatial boundary conditions were specified to be Y(‑2,t) = Y(2,t) =0 for all t.  The integration step size, d, in the time domain was set to 0.00005. The initial state of the particle was represented using a stationary Gaussian wave packet centered at x0 given by

 

                                                                                  (8)

  

where the initial velocity k0 was chosen to be 0 and x0 was chosen to be 0.5.  This form of  wave packet ensures that r(x,t) = |Y(x,t)|2 has unit integral. The variance of r(x,t), s02, was chosen to be 0.05.  This choice of s02 leads to a narrow wave packet and minimizes the effect the spatial boundary conditions may have on the integration.

 

The potential energy U(x,Y,t) was calculated as below:

 

                                                                     (9)

 

where jY(x,t) is given by (5).  The values of w and f(t) are the same as for the classical system.  In (5) the value of ÑY(x,t) was approximated by the central difference,

 

                                                                               (10)

 

The expectation of  x  was calculated as

 

                                                                                              (11)

 

Fig. 1 shows the expectation of x plotted against wt for l=2xw/p. The scaling factor of 1/p was empirically discovered in order to match the response of the quantum system with that of the classical system and clearly begs for a better derivation from the first principles.  Further simulations revealed that as the value of s0 is changed, the value of l also needs to be adjusted slightly to make the damping rates match. This indicates that the scaling factor is not equal to p for all values of s0 but remains within a vicinity of p.

Figure 1. Responses of the classical (outer plot) and quantized systems for a damped harmonic oscillator.

 

 

III. DISCUSSION

 

These simulation results indicate that it may be possible to model dissipative quantum systems with the help of a nonlinear Schrödinger equation. The proposed nonlinear solution is based on considerations similar to those described by Newell [8] for the use of the cubic nonlinearity in the Schrödinger equation in the form, Q(Y) = |Y|2.  In that form, with l < 0 in equation (2), the nonlinear term supplements the linear potential field with a potential well that shadows any wave packet and follows it wherever it goes.  The Ehrenfest force field of such a well counters the dispersive tendency of the linear time dependent Schrödinger equation.  In one spatial dimension, this cubic nonlinearity exactly offsets the dispersive force, resulting in soliton solutions (cf., Newell [8], pp. 24, 37, 46).

 

Similarly, the nonlinear term defined in (4) creates an Ehrenfest force field that is proportional to the probability current density and opposite to its direction.  This dissipates the energy of the wave function.  Note that this treatment of dissipation applies not only to the damping of the harmonic oscillator but to the general case as well.  It does not matter what form the linear part, V(x,t) of the scalar potential may be.     

 

 


REFERENCES

 

[1] D. J. Simms and N. M. J. Woodhouse, Lectures on Geometric Quantization, Lecture Notes in Physics, No. 53, Springer, New York, (1976).

 

[2] R. L. Anderson, Phenomenological quantum models for under and critically damped oscillators, American Journal of Physics, 61(4), April 1993.

 

[3] J. R. Ray, Lagrangians and systems they describe - how not to treat dissipation in quantum mechanics, American Journal of Physics, 47(7), July 1979.

 

[4] E. B. Lin, Quantum mechanical control systems, Mathematical and Computer Modelling, Vol. 12, No. 3, pp. 313-318, 1989.

 

[5] T. J. Tarn, G. Huang and J. W. Clark, Modelling of quantum mechanical control systems, Mathematical Modelling, Vol. 1, pp. 109-121, 1980.

 

[6] I. Bialynicki-Birula, and J. Mycielski, Nonlinear Wave Mechanics, Annals of Physics, 100, pp. 62-93, 1976.

 

[7] A. Goldberg, H. M. Schey, and J. L. Schwartz, Computer-generated motion pictures of one-dimensional quantum-mechanical transmission and reflection phenomena, American Journal of Physics, 35(3), pp. 177-186, March 1967.

 

[8] Newell, A., Solitons in Mathematics and Physics, Society for Industrial and Applied Mathematics, Philadelphia, PA, 1985.

 

 



[1] Martingale Research Corporation, 100 Allentown Parkway, Suite 211, Allen, Texas 75002

[2] Voice Control Systems, 14140 Midway Road, Suite 100, Dallas, Texas 75244