Discrete-continuous and classical-quantum

A discussion concerning the opposition between discretness and continuum in quantum mechanics is presented. In particular this duality is shown to be present not only in the early days of the theory, but remains actual, featuring different aspects of discretization. In particular discreteness of quantum mechanics is a key-stone for quantum information and quantum computation. A conclusion involving a concept of completeness linking discreteness and continuum is proposed. 1. Discrete-continuous in the old quantum theory Discretness is obviously a fundamental aspect of Quantum Mechanics. When you send white light, that is a continuous spectrum of colors, on a mono-atomic gas, you get back precise line spectra and only Quantum Mechanics can explain this phenomenon. In fact the birth of quantum physics involves more discretization than discretness : in the famous Max Planck’s paper of 1900, and even more explicitly in the 1905 paper by Einstein about the photo-electric effect, what is really done is what a computer does: an integral is replaced by a discrete sum. And the discretization length is given by the Planck’s constant. This idea will be later on applied by Niels Bohr during the years 1910’s to the atomic model. Already one has to notice that during that time another form of discretization had appeared : the atomic model. It is astonishing to notice how long it took to accept the atomic hypothesis (Perrin 1905). Nevertheless Bohr proposed a quantum atomic model : the atom is a classical one, that is a nucleus with electrons ”turning” around, but instead of considering the continuous family of possible trajectories, Bohr proposed to select only those for which the action (the enclosed area inside the planar trajectory) would by a multiple of the Planck’s constant. The ratios by the Planck’s constant of the differences of the corresponding energies give precisely the frequencies of the spectral lines of the hydrogen atom as observed experimentally. Extending this ”algorithm” to more general (non-integrable) situation will give rise, through the work of Born and Heisenberg and Schródinger, to the birth of quantum mechanics (Paul 2006), a much more conceptual and fundamental theory, from which the old Bohr theory can be deduced.in the limit where the Planck’s constant ~ → 0. But the old ”phenomenological” theory remain accurate even nowadays for systems for which quanDiscrete-continuous and classical-quantum 1 tum mechanics pains to predict numbers. In chemistry or atomic physics for example, the large number of degrees of freedom make often quantum mechanics difficult to operate explicitly: solving Schrödinger equation is hard, and its semicassical approximation (see below) difficult to handle when the system is not integrable. Roughly speaking let us say that the temptation to apply the old Bohr’s rule, that is to make discrete (and multiple of the an effective small Planck’s constant) any quantity which look ”like” an action, is big....and without any justification is sometimes very efficient. In fact the old quantum rules are a little bit like a visa† which permits to enter the quantum world, at the condition of staying close to the classical border. But not only this visa helps to compute numbers, it is many time the only way of getting effective computations. It is even now a great subject of inspiration for those who want to understand the semiclassical limit of non-integrable systems. Let us remark also that for this (pre-quantum mechanics) period the quantum theory appears twice as being a discretization of the classical situation: atomic hypothesis as discretization of continuous matter, and Bohr theory as discretization of the continuous classical mechanics. We would like to end up this introduction by mentioning that discreteness is one of the key ingredient in quantum computation (the other one being the superposition principle). The fact that a single particle (qubit) can support a binary information is typically quantum.. Moreover modern experiments in atomic physics reveal more and more that quantum discreteness can be observed in real situations. Indeed, although it has always been considered that, due to the smallness of th Planck constant and the large number of particles involved in experiments, discreteness would be hardly shown in nature, (a little bit like statistical (discrete) phenomena are usually well handled by continuous situations). modern physics is able now to provide real situations with very few atoms, and to let quantum discreteness appear. This new situation should inspire different ways of thinking about the alternative discrete/continuous, at least in its relationship with natural sciences. 2. Discrete-continuous and the Heisenberg-Schödinger pictures Separated in time only by a few months the two births of quantum mechanics (the preceding period is more referred as quantum theory) by Heisenberg first and then Schrödinger reveals also, according to us, the opposition discrete-continuous. The matrix theory by Heisenberg is a radical change of paradigm (although it is also founded on discretization of perturbation theory (Paul 2006)): all the classical quantities become matrix, that is discrete. The Schrödinger vision seems more ”classical”: the Schrödinger equation is a partial differential equation-and this explain certainly the success of this theory. It also reflects the opposition between the young Heisenberg and the already well established Schrödinger. It would certainly be very interesting to study historically in the XXth century the penetration of theses two points of view in the scientific community. † we take this image form a talk by J.M Levy-Leblond T. PaulC.N.R.S., D.M.A., Ecole Normale Supérieure 2 The way how Heisenberg picture is included in the Schrödinger one is interesting by itself as it shows a situation where the continuum generate the discrete‡. To understand this let us consider the analogy between quantum mechanics and a drum. A drum is a membrane which vibrates: the vibrations are encoded in a partial differential equation (wave equation). What is a partial differential equation? It is a way of going from local to global: knowing the position of the membrane in a very tiny piece of the surface will determine the position everywhere thanks to the propagation driven by the PDE. But a drum contains also something else: the boundary, on which the vibration must cancel. If one gives an initial position, in the given tiny part, and propagate it trough the PDE the result is that the vibration will never cancel at the boundary, never except for a discrete set of initial positions for which the miracle will happen. This discrete set is called the spectrum of the drum and is an example of discreteness generated by the property of continuity of the PDE. The drum selects, in-between the continuum of vibrations, those which will cancel on the border. And, whatever is the kick you give to the drum, the spectrum is (almost) the same. Heisenberg appears from Schrödinger in the same way: a PDE embedded in the continuum gives rise to discretness thanks to the boundary condition. But this time this set of frequencies is more complicated, less harmonic that for the drum (from this point of view it would be amusing to analyze carefully the analogy between this loss of harmonicity, and the almost contemporary birth of non-tonal music). But the idea is the same: continuum generates discretness. By saying this we are a little bit abusing language. Indeed if it is true that the PDE lives definitively in the continuum, the problem of eigenvalues for the Schrödinger equation is a little more tricky. Indeed the way the spectrum exhibits itself is a mixture of 1) the PDE (Schrödinger equation), 2) the boundary condition (border of a box), and the space of solution where one look to solve the problem. This last part took a certain time to be established carefully (by von Neumann). The space of solution has the nowadays magic name of Hilbert space. And a Hilbert space can be seen (this way seems old now, but on can find it in textbooks until the late 40s) as a limit of finite dimensional spaces together with a concept of completeness: the quantum continuum is indeed a passage to infinity with completeness, and this is crucial when recovering the (discrete) spectrum. We will discuss this fact later. 3. Discrete-continuous and the Bohr-Sommerfeld semiclassical formula The big success of Heisenberg and Schrödinger was to recover the spectrum of simple systems (such as Hydrogen atom or oscillators) in accordance with the old Bohr theory (actually the very big success was to predict the 2 of the harmonic oscillator that wasn’t given by Bohr). But exact solution are rare and soon semiclassical methods gave rise to results where the old Bohr law was recovered from Schrödinger equation in the limit ‡ Let us remark that in his original paper Schrödinger thanks...Hermann Weyl for help concerning the resolution of his eigenvalue equation Discrete-continuous and classical-quantum 3 ~ → 0. The so-called WKB method used here, inherited from optics (the semiclassical limit is equivalent to the passage from physical to geometrical optics) gives a precise prescription to order the spectrum by the set of natural. But labeling, in a natural and explicit way, an (a priori unordered) discrete set by integers is a very ambitious task. Therefore it is not surprising that it works only for few systems (the ones called integrable) and one can say that the extension of such a procedure to the non-integrable situation is not nowadays fully understood. The principal fact because of which the WKB-Bohr-Sommerfeld theory doesn’t work in general is that it relies on the existence of so-called invariant (by the classical flow) tori, a geometrical object which ceases to exist for non-integrable systems. Nevertheless if one gives up the ambition to associate to EACH eigenvalue a natural number (given by classical mechanics) but considers globally the set of eigenvalues (spectrum) there is a natural discrete set of objects which always exists: the set of periodic trajectories. We will see in the next section the link between these two discrete sets. 4. Discrete-continuous and the trace formula The helium atom is a 3-body system. But at the contrary of the celestial 3-body problem it is not perturbative. What count for the Schrödinger equation are the charges and not the masses. And the charge of each electrons is half he one of the kernel, that is of the same order. Therefore, after the great success of quantum mechanics, the helium atom (and more generally all quantum systems far for integrable) has remained as a challenging system, a system for which the regular methods of computing eigenvalues fail. The situation changed radically in 1971 when Gutzwiller published a fascinating paper whose contents is now called the Gutzwiller trace-formula. The idea is that the trace of the resolvent at energy E of the Schrödinger operator is determined semiclassically by the set of periodic orbits of the classical system of energy E. Mathematically it reads: ∑