1: \section{Introduction}
2: \label{intro}
3: %
4: Since their introduction in the field of quantum chaos by Kottos and
5: Smilansky \cite{paper:kottossmilansky2,paper:kottossmilansky}, quantum
6: graphs have played an important role in efforts towards a deeper
7: understanding of correlations in spectra of classically chaotic quantum
8: systems. According to the conjecture of Bohigas, Giannoni, and Schmit
9: \cite{paper:bohigasgiannonischmit}, these correlations can be described
10: by random matrix theory (RMT) \cite{book:mehta}. Although overwhelming
11: evidence supports this conjecture, until recently the theoretical
12: understanding of the RMT connection remained rather poor.
13:
14: Spectral two-point correlations are often measured in terms of the form
15: factor $K(\tau)$. Its small-$\tau$ asymptotics have been the subject of
16: recent
17: studies, producing a considerably improved understanding of spectral
18: correlations in terms of correlations
19: among classical periodic orbits
20: \cite{paper:sieberrichter,paper:sieber,paper:mulleretal}.
21: In this context quantum graphs have proven
22: ideal models for detailed investigations into the duality between
23: eigenvalue and periodic orbit correlations
24: \cite{paper:berkolaikoschanzwhitney,paper:berkolaikoschanzwhitney2,
25: paper:berkolaiko}.
26:
27: A guiding principle of the RMT conjecture is that symmetries of the
28: quantum system determine the universality class of spectral correlations.
29: Usually, the presence or absence of time-reversal invariance in quantum
30: systems with integer or half-integer total spin is taken as an indicator
31: whether an orthogonal, a unitary, or a symplectic universality class
32: is appropriate. Since Berry's pioneering
33: semiclassical analysis of the form factor \cite{paper:berry},
34: however, a vast majority of
35: investigations have concentrated on the orthogonal and the unitary case.
36: (See \cite{paper:scharfdietzkushaakeberry,paper:boltekeppeler3,
37: paper:keppelermarklofmezzadri} for exceptions where the symplectic case
38: is considered.) Quantum graphs with spin 1/2 were
39: introduced in \cite{paper:bolteharrison} in the context of a realization
40: of a Dirac operator on graphs. We demonstrated that in this way a
41: quantum system in the symplectic universality class can be realized,
42: and analyzed the form factor in the light of the recent developments
43: \cite{paper:bolteharrison2}.
44:
45: Here we extend our previous studies of spin-orbit coupling on quantum
46: graphs to the case of arbitrary values of the spin quantum number $s$.
47: Since the use of Dirac operators is restricted to the case of $s=1/2$,
48: this extension is most conveniently performed with a Pauli operator.
49: Starting from the case of spin zero, we introduce a spin-orbit coupling
50: that is localized at the vertices, in terms of boundary conditions
51: describing spin rotations at the vertices. We then continue to
52: investigate the form factor for the spectrum of the bond $S$-matrix
53: for a Pauli operator, following the diagrammatic method introduced in
54: \cite{paper:berkolaikoschanzwhitney,paper:berkolaikoschanzwhitney2}.
55: We analyze the group of spin rotations on the graph generated by those
56: at the vertices and show that properties of its irreducible
57: representations fix the spin contribution to the form factor.
58: In particular, it is found that a change from integer to half-integer
59: spin induces the same transformation of the form factor as exists between
60: the Circular Orthogonal Ensemble (COE) and Circular Symplectic Ensemble
61: (CSE) in RMT.
62:
63: This paper is organized as follows: After an introduction we
64: recall basic facts about graphs in section~\ref{1sec}. The construction
65: of Pauli operators on graphs and the structure of the corresponding
66: bond $S$-matrices is explained in section~\ref{2sec}. The definition
67: of the form factor and the removal of Kramers' degeneracy is presented
68: in section~\ref{s:form factor}. Section~\ref{s:spin contribution} is
69: then devoted to discussing the spin contribution to the form factor and
70: in section~\ref{s:correlations} we evaluate the spin contribution.
71: Group theoretic properties of representations used in the calculation
72: of the spin contribution are explained in an appendix.
73:
74:
75:
76: