A theory of transformation is presented for the diagonalization of a Hamiltonian that is quadratic in creation and annihilation operators or in coordinates and momenta. It is the systemization and theorization of Dirac and Bogoliubov-Valatin transformations, and thus provides us an operational procedure to answer, in a direct manner, the questions as to whether a quadratic Hamiltonian is diagonalizable, whether the diagonalization is unique, and how the transformation can be constructed if the diagonalization exists. The underlying idea is to consider the dynamic matrix. Each quadratic Hamiltonian has a dynamic matrix of its own. The eigenvalue problem of the dynamic matrix determines the diagonalizability of the quadratic Hamiltonian completely. In brief, the theory ascribes the diagonalization of a quadratic Hamiltonian to the eigenvalue problem of its dynamic matrix, which is familiar to all of us. That makes it much easy to use. Applications to various physical systems are discussed, with especial emphasis on the quantum fields, such as Klein-Gordon field, phonon field, etc..
Diagonalisation of quadratic multi-mode, time-dependent Hamiltonians can be fraught with difficulties. We set about diagonalising Hamiltonians from quantum optics. The Hamiltonians we choose represent Raman scattering, parametric amplification and the frequency converter. We present a systematic approach to this problem. The process of diagonalisation allows us to find invariants of the Hamiltonians under time evolution, find ground-state energies, apply the theory of Greens functions to find the solution to Schrodinger's equation, examine the conditions for oscillation or growth, and squeezing.
It is well known that the spectrum of a homogeneous quadratic Hamiltonian in m fermion construction-operator pairs is characterised by m mode energies. In this paper it is proved that the spectrum of a quadratic Hamiltonian with a linear part presents the same feature. Simple methods are given for the calculation of the mode energies and of the ground-state energy of the Hamiltonian. It is shown that, in contrast to the boson case, the diagonalisation cannot be carried out by a linear transformation of the fermion construction operators. Application of such a transformation can only result in a diagonalisation after the introduction of a 'ghost' particle (fermion) and a corresponding pair of Hermitian-conjugate construction operators.
As is well-known, in Bogoliubov's theory of an interacting Bose gas the ground state of the Hamiltonian
H^=∑k=0H^k is found by diagonalizing each of the Hamiltonians
H^k corresponding to a given momentum mode
k independently of the Hamiltonians
H^k′(=k) of the remaining modes. We argue that this way of diagonalizing
H^ may not be adequate, since the Hilbert spaces where the single-mode Hamiltonians
H^k are diagonalized are not disjoint, but have the
k=0 in common. A number-conserving generalization of Bogoliubov's method is presented where the total Hamiltonian
H^ is diagonalized directly. When this is done, the spectrum of excitations changes from a gapless one, as predicted by Bogoliubov's method, to one which has a finite gap in the
k→0 limit.
Here we present a problem related to the local Hamiltonian problem (identifying whether the ground-state energy falls within one of two ranges) which is restricted to being translationally invariant. We prove that for Hamiltonians with a fixed local dimension and
O(log(N))-body local terms, or local dimension
N and two-body terms, there are instances where finding the ground-state energy is quantum-Merlin-Arthur-complete and simulating the dynamics is BQP-complete (BQP denotes ``bounded error, quantum polynomial time''). We discuss the implications for the computational complexity of finding ground states of these systems and hence for any classical approximation techniques that one could apply including density-matrix renormalization group, matrix product states, and multiscale entanglement renormalization ansatz. One important example is a one-dimensional lattice of bosons with nearest-neighbor hopping at constant filling fraction---i.e., a generalization of the Bose-Hubbard model.