Static and dynamic variational principles for expectation values of observables (Q1107814)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Static and dynamic variational principles for expectation values of observables
scientific article

    Statements

    Static and dynamic variational principles for expectation values of observables (English)
    0 references
    0 references
    0 references
    1988
    0 references
    A general procedure is first reviewed for constructing variational principles (v.p.) suited to the optimization of a quantity of interest. The method is (enlightened) by a geometric interpretation. Conditions under which the solution may be obtained as a maximum rather than as a saddle-point are examined. Applications are then worked out, providing v.p. adapted to the evaluation of expectation values, fluctuations, or correlations of quantum observables, whether the system is at thermal equilibrium or whether it has evolved. In particular, a v.p. suited to the optimization of Tr \(Ae^{-\beta H}\) is built, generalizing to observables \(A\neq 1\) the usual free energy v.p.. To evaluate both the expectation value of an observable Q and its fluctuations (as well as correlations between \(Q_ 1\) and \(Q_ 2)\), the characteristic function, determined variationally by letting \(A=\exp [- \xi Q]\), is expanded in powers of \(\xi\). Within a given class of trial states, the best approximate state depends in general on the question raised. Another class of v.p. concerns dynamical problems. The general method allows one to recover a known time-dependent v.p. for a state and an observable [\textit{R. Balian} and \textit{V. Vénéroni}, Ann. Phys. 164, 334-410 (1985); and Phys. Rev. Lett. 47, 1353-1356, 1765 (1981)] which answers the following question. Given the state of the system at the time \(t_ 0\), what is the expectation value \(<A>\) at a later time \(t_ 1?\) A more general v.p. applies to situations in which the initial state is too complicated to be handled exactly. Both the approximate initial conditions and the approximate evolution are then determined so as to optimize \(<A>\). Finally, analogous v.p. are constructed in classical statistical mechanics and Hamiltonian dynamics. A recent formulation of classical mechanics in terms of covariant Poisson brackets [\textit{J. E. Marsden}, \textit{R. Montgomery}, \textit{P. J. Morrison} and \textit{W. B. Thompson}, Ann. Phys. 169, 29-47 (1986)] comes out naturally in this contact from the v.p. for the evaluation of a classical expectation value.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    variational principles
    0 references
    quantum observables
    0 references
    expectation value of an observable
    0 references
    classical statistical mechanics
    0 references
    Hamiltonian dynamics
    0 references
    covariant Poisson brackets
    0 references
    0 references