A program of proofs is outlined to show how the linear structure of quantum mechanics can be derived from basic principles. For the dynamics part, three alternatives are reviewed briefly. They include the Kadison theorem for evolution of density matrices and the Segal–Simon theorem for evolution of observables in the Heisenberg picture. The focus is on Wigner’s theorem that gives linear operators for evolution of state vectors. New insights are described that motivate the key assumption about probabilities in that theorem. One is that it is implied by the second law of thermodynamics. Another is that it follows from assumptions that evolution in time preserves mixtures of states and preserves probabilities for what is in the mixture.
A program of proofs is outlined to show how the linear structure of quantum mechanics can be derived from basic principles. For the dynamics part, three alternatives are reviewed briefly. They include the Kadison theorem for evolution of density matrices and the Segal–Simon theorem for evolution of observables in the Heisenberg picture. The focus is on Wigner’s theorem that gives linear operators for evolution of state vectors. New insights are described that motivate the key assumption about probabilities in that theorem. One is that it is implied by the second law of thermodynamics. Another is that it follows from assumptions that evolution in time preserves mixtures of states and preserves probabilities for what is in the mixture.
A basic linearity of quantum dynamics, that density matrices are mapped linearly to density matrices, is proven very simply for a system that does not interact with anything else. It is assumed that at each time the physical quantities and states are described by the usual linear structures of quantum mechanics. Beyond that, the proof assumes only that the dynamics does not depend on anything outside the system but must allow the system to be described as part of a larger system. The basic linearity is linked with previously established results to complete a simple derivation of the linear Schr\"odinger equation. For this it is assumed that density matrices are mapped one-to-one onto density matrices. An alternative is to assume that pure states are mapped one-to-one onto pure states and that entropy does not decrease.
A consistent description of the physical state of a system is provided by its wavefunction whose time-evolution is described by the Schrödinger equation The wavefunction, is the coordinate representation of the state vector in the Hilbert space that describes the system. Unitary transformations of an orthonormal basis set that spans the Hilbert space amount to rotation of the basis set. An example of this is the unitary transformation from basis to using the Clebsch– Gordan coefficients that we discussed in Chapter 4. Such unitary transformations preserve the norm of the state vectors, and alternative basis sets connected by them provide mathematically equivalent descriptions of quantum mechanics. Preference of using one basis over another is dictated by algebraic elegance. Alternative descriptions of the temporal evolution of a physical system are possible using generalized rotations that leave the physics invariant, but alter the description of the temporal evolution of the system. Whereas the Schrödinger picture describes temporal evolution of a physical system using methodologies in which the operators are independent of time, and all the time-dependence appears in the wavefunctions, in the Heisenberg picture all the time-dependence is contained in the operators, while the wavefunctions are considered independent of time. In the Dirac picture (also called as the Interaction picture ), both the wavefunctions and the operators are considered to be time-dependent. The three pictures are essentially equivalent, and transformations from any one of them to any other are effected using generalized rotations brought about by unitary transformations described below. The time-evolution operator in the Schrödinger picture is given by Eq. 1.86b (Chapter 1). If the reference time t 0 is considered to be zero, the time evolution of a physical state is described by where the exponential operator is which gives the dynamical phase (Eq. 1.114b in Chapter 1) of a stationary state: Operators like q, p, H are independent of time in the Schrödinger picture.
In classical mechanics the state of a system is defined by the numerical values of a set of dynamical variables such as the position coordinates and the velocities. In quantum mechanics the state of a system at an instant is represented by a unit vector in a Hilbert space, where a set of axes are the eigenvectors of a complete set of observables. The description of quantum dynamics where the wave function is time-dependent and the operators describing the observables are time-independent is known as the Schrodinger picture. The use of total time derivative of operator that depends on the time evolution of the wave function as well as on any intrinsic time-dependence in the operators simplifies the development of quantum mechanics and allows the development to more closely follow the corresponding development of classical mechanics. The difference between the Schrodinger and Heisenberg pictures is best understood as the difference in the basic set of observables.
A seed George planted 45 years ago is still producing fruit now. In 1961, George set out the fundamental proposition that quantum dynamics is described most generally by linear maps of density matrices. Since the first sprout from George's seed appeared in 1962, we have known that George's fundamental proposition can be used to derive the linear Schrodinger equation in cases where it can be expected to apply. Now we have a proof of George's proposition that density matrices are mapped linearly to density matrices, that there can be no nonlinear generalization of this. That completes the derivation of the linear Schrodinger equation. The proof of George's proposition replaces Wigner's theorem that a symmetry transformation is represented by a linear or antilinear operator. The assumption needed to prove George's proposition is just that the dynamics does not depend on anything outside the system but must allow the system to be described as part of a larger system. This replaces the physically less compelling assumption of Wigner's theorem that absolute values of inner products are preserved. The history of this question is reviewed. Nonlinear generalizations of quantum mechanics have been proposed. They predict small but clear nonlinear effects, which very accurate experiments have not seen. This begs the question. Is there a reason in principle why nonlinearity is not found? Is it impossible? Does quantum dynamics have to be linear? Attempts to prove this have not been decisive, because either their assumptions are not compelling or their arguments are not conclusive. The question has been left unsettled. The simple answer, based on a simple assumption, was found in two steps separated by 44 years.
In this chapter, we finally introduce time evolution. The development is carried out for finite, enumerable systems (i.e., systems whose states live in a finite dimensional, inner product space) which, of course, allows a straight forward extension to the continuum and infinite dimensional state spaces. The necessary mathematical background of unitary operators, and their properties relevant to quantum mechanics are first introduced. In particular, the fact that preservation of norm under the action of an operator implies unitarity, and that every unitary operator can be expressed as the exponential of i timesa Hermitian operator are introduced. Time evolution is then introduced in its integral form through the time evolution operator, and the Hamiltonian operator is introduced in this context. This approach, however, necessitates the restriction to conservative systems. Next, the differential form of the law of time evolution, i.e., Schrodinger equation, is derived from the time evolution operator, and this is postulated to be the general law of time evolution. We then demonstrate how, for conservative systems, the solution of the Schrodinger equation essentially reduces to solving the eigenvalue problem of the Hamiltonian. Constants of motion and Heisenberg picture have been relegated to the problem section.
In 1926, the Austrian physicist Erwin Schrodinger derived an equation of waves from the Hamilton variational principle inspired by the analogy between Mechanics and Optics. This equation explained much of quantum phenomenology was known at that time. Physical laws must be invariant under certain symmetries, represented by the transformations of the mathematical object that define these laws. In particular in quantum mechanics one can ask for the transformations of states and observables. These transformations must fulfill what constitutes the essence of the so-called Wigner's theorem, that is, that the transformed observables must possess the same possible sets of eigenvalues as the old ones and that the transformations of the states must give the same probabilities. The temporal evolution of the states must always be given by unit operators that conserve scalar products.