Professional Documents
Culture Documents
PDF Conical Intersections in Physics An Introduction To Synthetic Gauge Theories Lecture Notes in Physics 965 Jonas Larson Ebook Full Chapter
PDF Conical Intersections in Physics An Introduction To Synthetic Gauge Theories Lecture Notes in Physics 965 Jonas Larson Ebook Full Chapter
https://textbookfull.com/product/conductors-semiconductors-
superconductors-an-introduction-to-solid-state-physics-
undergraduate-lecture-notes-in-physics-rudolf-p-huebener/
https://textbookfull.com/product/an-introduction-to-gauge-
theories-1st-edition-nicola-cabibbo/
https://textbookfull.com/product/introduction-to-muon-spin-
spectroscopy-applications-to-solid-state-and-material-sciences-
lecture-notes-in-physics-961-1st-edition-amato/
https://textbookfull.com/product/surveys-in-theoretical-high-
energy-physics-2-lecture-notes-from-serc-schools-1st-edition-
raghavan-rangarajan/
An Introduction to Geometrical Physics Second Edition
Ruben Aldrovandi
https://textbookfull.com/product/an-introduction-to-geometrical-
physics-second-edition-ruben-aldrovandi/
https://textbookfull.com/product/an-introduction-to-quantum-
physics-first-edition-anthony-p-french/
https://textbookfull.com/product/math-8-lecture-notes-
introduction-to-advanced-mathematics-xin-zhou/
https://textbookfull.com/product/lie-algebras-in-particle-
physics-from-isospin-to-unified-theories-first-edition-howard-
georgi/
https://textbookfull.com/product/statistical-field-theory-an-
introduction-to-exactly-solved-models-in-statistical-physics-2nd-
edition-giuseppe-mussardo/
Lecture Notes in Physics 965
Jonas Larson
Erik Sjöqvist
Patrik Öhberg
Conical
Intersections
in Physics
An Introduction to Synthetic Gauge
Theories
Lecture Notes in Physics
Volume 965
Founding Editors
Wolf Beiglböck, Heidelberg, Germany
Jürgen Ehlers, Potsdam, Germany
Klaus Hepp, Zürich, Switzerland
Hans-Arwed Weidenmüller, Heidelberg, Germany
Series Editors
Matthias Bartelmann, Heidelberg, Germany
Roberta Citro, Salerno, Italy
Peter Hänggi, Augsburg, Germany
Morten Hjorth-Jensen, Oslo, Norway
Maciej Lewenstein, Barcelona, Spain
Angel Rubio, Hamburg, Germany
Manfred Salmhofer, Heidelberg, Germany
Wolfgang Schleich, Ulm, Germany
Stefan Theisen, Potsdam, Germany
James D. Wells, Ann Arbor, MI, USA
Gary P. Zank, Huntsville, AL, USA
The Lecture Notes in Physics
The series Lecture Notes in Physics (LNP), founded in 1969, reports new devel-
opments in physics research and teaching-quickly and informally, but with a high
quality and the explicit aim to summarize and communicate current knowledge in
an accessible way. Books published in this series are conceived as bridging material
between advanced graduate textbooks and the forefront of research and to serve
three purposes:
• to be a compact and modern up-to-date source of reference on a well-defined
topic
• to serve as an accessible introduction to the field to postgraduate students and
nonspecialist researchers from related areas
• to be a source of advanced teaching material for specialized seminars, courses
and schools
Both monographs and multi-author volumes will be considered for publication.
Edited volumes should, however, consist of a very limited number of contributions
only. Proceedings will not be considered for LNP.
Volumes published in LNP are disseminated both in print and in electronic for-
mats, the electronic archive being available at springerlink.com. The series content
is indexed, abstracted and referenced by many abstracting and information services,
bibliographic networks, subscription agencies, library networks, and consortia.
Proposals should be sent to a member of the Editorial Board, or directly to the
managing editor at Springer:
Dr Lisa Scalone
Springer Nature
Physics Editorial Department
Tiergartenstraße 17
69121 Heidelberg, Germany
[email protected]
Conical Intersections
in Physics
An Introduction to Synthetic Gauge Theories
123
Jonas Larson Erik Sjöqvist
Department of Physics Department of Physics and Astronomy
Stockholm University Uppsala University
Stockholm, Sweden Uppsala, Sweden
Patrik Öhberg
IPaQS/EPS
Heriot-Watt University
Edinburgh, UK
This Springer imprint is published by the registered company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To the memory of Stig Stenholm.
Preface
In the history of science we find many examples where the wheel has been
reinvented. It may be that some work, for some reason, fell into oblivion or it was
simply never recognised by the community. In Einstein’s two papers Quantentheorie
des einatomigen idealen Gases from 1924 and 1925, he essentially predicted
condensation. Einstein realised that there is a critical temperature below which
a single state becomes macroscopically populated; he calls it ‘condensation’.
Interestingly, at that time the concept of an order parameter had not been introduced,
but Einstein notices ‘One can assign a scalar wave field to such a gas’. What Einstein
was hinting at was what today is known as the Gross–Pitaevskii equation. It took,
however, Gross and Pitaevskii more than 30 years to write down the equation for
the classical field. Another example is that of the Aharonov–Bohm effect presented
in 1959. However, less known is that the effect was predicted already some 10 years
earlier by Ehrenberg and Siday.
The above gives two examples where a result in its full glory has not been
recognised (or it is simply not known). This often happens when essentially the
same phenomenon is rediscovered in different communities. The 1984 seminal work
by Berry on ‘phase factors accompanying adiabatic changes’ had two precursors.
Already in 1956, Pancharatnam demonstrated how the interference of two polarised
light beams depends on a geometric phase. While in 1958, Longuet-Higgins et
al. showed that adiabatic following around a conical intersection between two
electronic molecular potential surfaces resulted in a sign change of the wave
function. Berry’s work showed the generality of this phenomenon; whenever the
wave function adiabatically encircles some sort of singularity in parameter space it
acquires a geometric phase factor. Such singularity appears at a conical intersection
which is characterised by a point degeneracy of at least two energy surfaces in some
parameter space.
Conical intersections are to be found in a range of different physical systems, and
it seems that the importance of them has often been analysed independently in the
different communities. Thus, the physics of conical intersections can be regarded as
yet another example for where the wheel has been reinvented. Even if the concept
of conical intersections as such is the same in the different communities, there are
vii
viii Preface
Acknowledgements
Over the years, there have been many people who we have collaborated or discussed
with, people whom, in one way or another, have had an influence on this book. We
are especially thankful to Alexander Altland, Emil Bergholtz, Jean Dalibard, Marie
Ericsson, Barry Garraway, Gonzalo García de Polavieja, Nathan Goldman, Osvaldo
Goscinski, Hans Hansson, Niklas Johansson, Gediminas Juzeliūnas, Thomas Klein
Kvorning, Åsa Larson, Maciej Lewenstein, Jani-Petri Martikainen, Luis Santos, Ian
Spielman, Stig Stenholm, Robert Thomson, Dianmin Tong, Manuel Valiente, and
Johan Åberg.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 4
2 Theory of Adiabatic Evolution . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 5
2.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 5
2.2 Adiabatic Time-Evolution.. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 6
2.2.1 Adiabatic Theorem . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 6
2.2.2 Adiabatic Approximation . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 7
2.2.3 The Marzlin–Sanders Paradox .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 9
2.2.4 The Importance of the Energy Gap: Local Adiabatic
Quantum Search . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 11
2.3 Gauge Structure of Time-Dependent Adiabatic Systems .. . . . . . . . . . . . 13
2.3.1 The Wilczek–Zee Holonomy . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 13
2.3.2 Adiabatic Evolution of a Tripod . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
2.3.3 Closing the Energy Gap: Abelian Magnetic Monopole
in Adiabatic Evolution . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 20
2.4 Born–Oppenheimer Theory .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
2.4.1 Synthetic Gauge Structure of Born–Oppenheimer
Theory.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
2.4.2 Adiabatic Versus Diabatic Representations . . . . . . . . . . . . . . . . . . 25
2.4.3 Born–Oppenheimer Approximation . . . . . .. . . . . . . . . . . . . . . . . . . . 27
2.4.4 Synthetic Gauge Structure of an Atom in an
Inhomogeneous Magnetic Field . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 28
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 31
3 Conical Intersections in Molecular Physics . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 33
3.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 33
3.2 Where Electronic Adiabatic Potential Surfaces Cross:
Intersection Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 35
3.2.1 The Existence of Intersections . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 35
3.2.2 Topological Tests for Intersections . . . . . . .. . . . . . . . . . . . . . . . . . . . 38
xi
xii Contents
3.2.3
The Molecular Aharonov–Bohm Effect on the
Nuclear Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 42
3.3 The Jahn–Teller Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 44
3.3.1 Spontaneous Breaking of Molecular Symmetry: The
Jahn–Teller Theorem . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 44
3.3.2 The E × JT Model . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 46
3.4 Dynamical Manifestation of Conical Intersections . . . . . . . . . . . . . . . . . . . 49
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 53
4 Conical Intersections in Condensed Matter Physics .. . . . . . . . . . . . . . . . . . . . 55
4.1 Band Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 55
4.1.1 Bloch’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 56
4.1.2 Tight-Binding Model .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 57
4.1.3 Bloch and Wannier Functions .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 59
4.1.4 Single Particle Lattice Models and Bloch Hamiltonians . . . . 61
4.1.5 Symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
4.1.6 Topological Invariant .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 67
4.2 Spin–Orbit Couplings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 73
4.2.1 Rashba and Dresselhaus Spin–Orbit Couplings.. . . . . . . . . . . . . 74
4.2.2 Intrinsic Spin Hall Effect .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
4.3 Superconductors .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 78
4.4 Graphene .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
4.4.1 Tight-Binding Band Spectrum . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
4.4.2 Relativity at Almost ‘Zero’ . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 85
4.4.3 The Haldane Model . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 86
4.5 Weyl Semimetals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 87
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
5 Conical Intersections in Cold Atom Physics . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
5.1 Introduction .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
5.2 Light–Matter Interactions and Optical Forces . . . .. . . . . . . . . . . . . . . . . . . . 94
5.3 Adiabatic Dynamics and Synthetic Gauge Potentials . . . . . . . . . . . . . . . . 99
5.3.1 The Adiabatic Principle and Dressed States. . . . . . . . . . . . . . . . . . 100
5.3.2 A Pedagogical Example: The Two-Level System.. . . . . . . . . . . 101
5.4 Spin–Orbit Coupling and Non-Abelian Phenomena .. . . . . . . . . . . . . . . . . 103
5.4.1 Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 107
5.4.2 A Quasi-Relativistic Example: The Atomic
Zitterbewegung . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 109
5.5 Cold Atoms and the Bose–Einstein Condensate . .. . . . . . . . . . . . . . . . . . . . 118
5.5.1 The Description of a Condensate . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 119
5.5.2 Conical Intersections and the Gross–Pitaevskii Equation . . . 122
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 124
6 Conical Intersections in Other Physical Systems . . . . .. . . . . . . . . . . . . . . . . . . . 127
6.1 Cavity Quantum Electrodynamics .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
6.1.1 The Jaynes–Cummings and Quantum Rabi Models . . . . . . . . . 128
Contents xiii
objective of the Lecture Notes is to emphasis the unifying aspect of CIs and synthetic
gauge structures in quantum physics.
Chapter 2 is devoted to the concept of adiabaticity in quantum systems. In
Sect. 2.2, we discuss the theory of adiabatic time-evolution and examine some of
its subtleties. The synthetic gauge structure associated with parameter-dependent
energy subspaces is described in Sect. 2.3. This gauge structure is generally non-
Abelian, meaning that it gives rise to matrix-valued ‘phase factors’ (more precisely,
unitary matrices), but becomes Abelian in the case of non-degenerate energies.
Here, we see the first example where CIs play a central role, namely that these
intersections are sources of magnetic monopoles, which are missing entities in
electromagnetism, but now can be studied experimentally in adiabatic systems.
In Sect. 2.4, we discuss the time-independent counterpart of adiabatic systems.
Here, the parameters are themselves quantum variables, such as the nuclear degrees
of freedom in a molecule or crystal. In this setting, the slow coordinates move on
adiabatic potential energy surfaces (APSs), being parameter dependent solutions
of the time-independent Schrödinger equation for the fast variables. Such ‘fast–
slow’ quantum systems are called Born–Oppenheimer systems. Just as in the time-
dependent case, the Born–Oppenheimer setting gives rise to synthetic gauge fields,
but now having a direct physical effect on the motion of the slow system.
In Chap. 3, we further explore the synthetic gauge structure in Born–
Oppenheimer systems, by considering the molecular case in more detail. As
mentioned above, the nuclear variables in a molecule play the role of slow
parameters coupled to the fast electronic degrees of freedom. In Sect. 3.2.1,
we examine different types of intersections between electronic APSs and in
Sect. 3.2.2, we discuss tests to find intersection points by using tools from algebraic
topology.
Sections 3.3 and 3.4 focus on the Jahn–Teller effect and its relation to CIs
between symmetry adapted electronic states. We discuss the linear+quadratic E ×
Jahn–Teller system, which, despite its simplicity, is associated with a quite intricate
CI structure. We examine the Berry phase as well as the dynamical implications
of the CI structure for the nuclear pseudo-rotational motion. In particular, we
demonstrate that this system provides a realisation of a spin Hall effect that is
manifest in the dynamics of the nuclei.
In Chap. 4, we turn our attention to condensed matter systems. An essential
conceptual difference between the CIs considered up to this point and those in
condensed matter systems is that the latter appear in momentum space. In other
words, a CI is a point or line in the Brillouin zone where two or more energy bands
cross. The perhaps most well-known example of such a crossing is for graphene,
i.e., a hexagonal monolayer of carbon atoms for which the Fermi surface is exactly
at the CIs. By linearising the dispersion, the electrons close to the Fermi level are
described by a Dirac-like equation in 2 + 1 dimensions, but with an effective ‘speed
of light’ being considerably smaller than c. CIs in the context of condensed matter
systems are therefore called Dirac points or Dirac CIs.
Dirac CIs are also of importance in spin–orbit coupled systems and in the BCS
theory of superconductivity. Another central concept, used in order to analyse con-
1 Introduction 3
densed matter systems with non-trivial topological features, is the Chern number.
This quantity is a topological invariant of an energy band in two dimensions and is
defined as the Berry phase integrated over the whole Brillouin zone. Its importance
is evident as it is directly related to the quantised Hall conductivity.
The motion of cold atoms in slowly varying inhomogeneous laser fields is
governed by artificial gauge fields that arise when the lasers induce fast transitions
between the internal atomic levels. These gauge fields can be Abelian or non-
Abelian. In Chap. 5, we discuss the slow motion of cold atoms in inhomogeneous
optical fields and some of the peculiar effects associated with the artificial gauge
structure. Standing laser wave fields can be used to create optical lattice that can
be used to simulate real crystals by injecting atoms into the periodic structure and
thereafter studying their motion.
Sections 5.2 and 5.3 contain a description of light–atom interaction and the
adiabatic dynamics associated with the induced synthetic gauge fields. The idea is
to view the atom as a Born–Oppenheimer-like system for which the centre of mass
and the internal energy levels play the role as the slow and fast degrees of freedom,
respectively. These ‘subsystems’ are coupled via the inhomogeneity of the laser
fields, which are assumed to vary sufficiently slowly for the system to be adiabatic.
Just as in the spin systems discussed in Chap. 2, the induced gauge fields take the
form of magnetic monopoles located at CIs, now in the space defined by the laser
parameters. We discuss, in Sect. 5.4, how non-Abelian synthetic gauge fields can be
obtained if the lasers couple three ground state levels to an excited stated, forming
a tripod system. These gauge fields can be used to simulate relativistic effects, such
as Zitterbewegung, which otherwise require relativistic particles.
The synthetic gauge fields are not limited to single atom systems. In Sect. 5.5, we
discuss the theory of Bose–Einstein condensates, in which a cloud of atoms undergo
a phase transition so as to behave as a macroscopically occupied single quantum
state at sufficiently low temperatures. The description of a condensate differs
significantly from fundamental quantum systems in that the atom–atom collisions
give rise to non-linear effects, as described by the Gross–Pitaevskii equation. The
non-linearity implies that CIs can have completely different shapes in parameter
space, as discussed in Sect. 5.5.2.
In Chap. 6, we examine CIs in some other system. We first consider cavity
electrodynamics in Sect. 6.1. Here, the internal levels of an atom are coupled to
a quantised cavity field. This system is described by the Rabi or Jaynes–Cummings
models. In the semi-classical regime, one obtains CIs in the quadrature phase space.
In particular, we demonstrate how such a system can give rise to an intrinsic
anomalous Hall effect.
In Sect. 6.2, we discuss a system of trapped ions interacting with the vibrations
in the trap. The ion trap and cavity settings both describe a discrete level system
interacting with a harmonic oscillator system, in the former case the oscillations in
the trap while in the latter case the oscillations of the photon standing wave in the
cavity. We show how the ion trap system can be used to realise different types of
Jahn–Teller systems.
4 1 Introduction
The relevance of CIs is not restricted to quantum systems. They may occur also
in classical wave systems. In Sect. 6.3, we discuss how such intersections may show
up for classical light of sufficiently long wavelength moving through materials with
a small space-dependent refractive index and how lattice models can be emulated in
such settings. These systems may realise Dirac cones by appropriately choosing the
lattice structure. Among other conical singularity structures in classical systems, but
not covered by these Lecture Notes, are found in Einstein’s theory of gravity. These
structures occur along the world sheet of cosmic strings [9] and for black holes [10].
They differ conceptually from those considered here in the sense that they contain a
single cone and not two intersecting ones.
In Sect. 6.4, we end our expose over CIs in quantum physics by considering what
happens if the system is open. An open system is defined as a quantum system that
interacts with some kind of ‘environment’. This means that we need to replace the
Schrödinger equation by an effective description, which we obtain by tracing over
the environmental degrees of freedom. Under certain approximations (essentially
the Markov limit), this is described by the Lindblad equation, which is a master
equation that models the evolution of the density operator. The Liouvillian of a
Lindblad equation is not Hermitian, and the energies typically become complex-
valued. Thereby, the concept of CIs drastically change. Still there is a notion of
energy level crossings in parameter space, called exceptional points, which are
essentially CIs where both the real and imaginary parts of the spectrum cross
simultaneously.
References
1. Berry, M.V.: Wave geometry: a plurality of singularities. In: Anandan, J.S. (ed.) Quantum
Coherence, pp. 92–98. World Scientific, Singapore (1991)
2. Jahn, H.A., Teller, E.: Stability of polyatomic molecules in degenerate electronic states. I.
Orbital degeneracy. Proc. R. Soc. A. 161, 220 (1937)
3. Longuet-Higgins, H.C., Öpik, U., Pryce, M.H.L., Sack, R.A.: Studies of the Jahn–Teller effect.
II. The dynamical problem. Proc. R. Soc. Lond. Ser. A 244, 1 (1958)
4. Mead, C.A.: The molecular Aharonov-Bohm effect in bound states. Chem. Phys. 49, 23 (1980)
5. Berry, M.V.: Quantal phase factors accompanying adiabatic evolution. Proc. R. Soc. Lond. Ser.
A 392, 45 (1984)
6. Bernevig, B.A.: Topological Insulators and Topological Superconductors. Princeton University
Press, Princeton and Oxford (2013)
7. Dalibard, J., Gerbier, F., Juzeliūnas, G., Öhberg, P.: Colloquium: artificial gauge potentials for
neutral atoms. Rev. Mod. Phys. 83, 1523 (2011)
8. Carollo, A.C.M., Vedral, V.: Holonomic quantum computation. In: Bruß, D., Leuchs, G. (eds.)
Quantum Information: From Foundations to Quantum Technology Applications, pp. 475–482.
Wiley-VCH Verlag GmbH & Co. KGaA (2019)
9. Oliveira-Neto, G.: Identifying conical singularities. J. Math. Phys. 37, 4716 (1996)
10. Solodukhin, S.N.: Conical singularity and quantum corrections to the entropy of a black hole.
Phys. Rev. D 51, 609 (1995)
Chapter 2
Theory of Adiabatic Evolution
2.1 Introduction
We shall see that the notion of adiabaticity is associated with rich physical
properties and conceptual subtleties. In this chapter, we outline the basic theory
for adiabatic evolution and provide some examples. We start with time-dependent
systems in Sect. 2.2, where we delineate the conditions under which adiabatic
evolution occurs. Section 2.3 shows how a synthetic gauge structure, similar
to that of electromagnetism, appears in the adiabatic regime of time-dependent
systems. Section 2.4 addresses the time-independent situation, leading to the Born–
Oppenheimer approximation. It turns out that this approximation can be associated
with a similar kind of synthetic gauge theory as in the time-dependent case. In this
way, the synthetic gauge structure serves as a unifying aspect of adiabatic systems.
The precise meaning of ‘slowness’ is given by the adiabatic theorem. Here, we state
this important theorem and specify under which premises it holds.
The adiabatic theorem gives the rate at which transitions between the eigenspaces
of Ĥ (R(t/T )) tend to zero when T → ∞. Explicitly, by means of the parameter
change t → s = t/T , the adiabatic theorem can be stated as
ÛT (s, 0)P̂n (R(0)) − P̂n (R(s))ÛT (s, 0) = O(1/T ), T → ∞, ∀n, (2.1)
An implication of these conditions is that the rank of P̂n (R(s)) is not allowed to
change, not even at isolated points, on s ∈ [0, 1].
To see what the adiabatic theorem means, let us consider the simplest case where
the eigenprojector is one-dimensional (rank = 1) so that we can write P̂n (R(s)) =
|ψn (R(s)) ψn (R(s))| throughout the evolution. If the system starts in the state
|Φ(0) = |ψn (R(0)) and all the premises (i)–(iii) hold, then (2.1) implies
in the adiabatic T → ∞ limit. We shall have more to say about the phase κn (s) in
Sect. 2.3.3 below.
The adiabatic theorem gives the rate at which a given state starting in an eigen-
subspace of the system Hamiltonian remains in the eigensubspace when the
Hamiltonian change becomes increasingly slow. However, it should be clear that
the state never follows the instantaneous energy eigenstates perfectly for a large but
finite T ; there are always small but non-zero non-adiabatic transitions between the
eigensubspaces. Thus, perfect adiabatic evolution is an idealisation and in practical
applications it becomes essential to tell how well the evolution coincides with the
adiabatic one. Here, we find a useful condition that answers this question.
Consider a quantum system with Hilbert gspace H and parameter-dependent
Hamiltonian Ĥ (R(s)). Let P̂n (R(s)) = n
k=1 |ψn;k (R(s)) ψn;k (R(s))|, n =
1, . . . , K ≤ dim H , be the eigenprojectors of Ĥ (R(s)), gn being the rank (degree
of degeneracy) of the nth eigensubspace, with corresponding energies εn ((R(s))).
The eigensubspaces span the Hilbert space, meaning that any solution |Φ(s) of the
8 2 Theory of Adiabatic Evolution
K
gn s
εn (s )ds
|Φ(s) = e−iT 0 |ψn;k (R(s)) cn;k (s). (2.6)
n=1 k=1
K gn s
s
Δnm (R(s ))ds
− eiT 0 ψn,k (R(s ))|ψ̇m,l (R(s )) cm;l (s )ds (2.7)
m=n=1 l=1 0
with the energy gap functions Δnm = εn − εm . The third term on the right-hand side
of (2.7) describes non-adiabatic transitions between the energy eigensubspaces and
should be negligible in the adiabatic regime. It contains two competing time-scales:
the rate of transition between the eigensubspaces
1
ψn,k (R(s))|ψm,l (R(s + δs)) = ψn,k (R(s))|ψ̇m,l (R(s)) (2.8)
δs
and the ‘intrinsic’ oscillation frequency for transitions between the eigensubspaces
of the instantaneous Hamiltonian
s
d
T
ds Δnm (R(s )) = T |Δnm (R(s))| (2.9)
ds
0
s
associated with the phase factors eiT 0 Δnm (R(s ))ds . Intuitively, the non-adiabatic
term of (2.7) washes out due to these oscillatory phase factors when the intrinsic
frequency is much larger than the transition rate. This yields the adiabatic condition,
ψn,k (R(s))|ψ̇m,l (R(s))
T , ∀s ∈ [0, 1], ∀m. (2.10)
Δ (R(s))
nm
gn
|Φ(0) = |ψn;k (R(0)) → |Φ(s) ≈ |ψn;l (R(s)) cn;l (s)
l=1
ψn,k (R(s))|ψ̇m,l (R(s))
⇒T , ∀s ∈ [0, 1]. (2.12)
Δ (R(s))
nm
However, it is important to note that the reverse is not true since the relation between
the adiabatic condition and the adiabatic approximation is valid under one important
caveat: the adiabatic theorem should hold in the T → ∞ limit, which means, in
particular, that all the premises (i)–(iii) (see Sect. 2.2.1) must be satisfied by the
Hamiltonian of the system. The Marzlin–Sanders paradox [3], as will be discussed
in the following section, illustrates this point in a striking way.
ÛT (s, 0) being the exact time-evolution operator associated with Ĥ (s). Note that,
unlike Ĥ , Hˆ does not define a fixed one-parameter family of Hamiltonians for all
ˆ is a function of both s and T . Conceptually, this means that the notion of
T , i.e. H
‘slowness’ becomes ambiguous; indeed, one can see from its definition that H ˆ (s, T )
10 2 Theory of Adiabatic Evolution
ˆ (s, 0) =
Let us test whether this holds for the mirror system. We note that U T
†
ÛT (s, 0), which implies
ˆ (s, 0)|ψ
n;k (0) †
ψm=n;l (s)|U T = ψm=n;l (s)|ÛT (s, 0)ÛT (s, 0)|ψn;k (0)
= ψm=n;l (s)|ψn;k (0) , (2.16)
which can take any value between 0 and 1. We thus arrive at the seemingly para-
doxical conclusion that although the mirror system satisfies the adiabatic condition,
it does not, in general, evolve adiabatically. That is, the adiabatic condition does not
imply the validity of the adiabatic approximation.
The origin of the paradox can be seen by noting that not all premises (i)–(iii) of
the adiabatic theorem (see Sect. 2.2.1) are satisfied by the mirror Hamiltonian [4].
Specifically, premises (i) and (ii) concern the energy gaps and are therefore equally
valid for Hˆ (s, T ) and Ĥ (s) since
εn (s) = −εn (s). The critical premise is instead
(iii), which holds, by assumption, for H (s) but not for H ˆ (s, T ). The reason is that
the eigenprojectors P ˆ (s, T ) = Û † (s, 0)P̂ (s)Û (s, 0) of the mirror Hamiltonian
n T n T
contain exponential factors, originating from ÛT (s, 0), whose argument tends to
infinity when T → ∞. These arguments make the derivatives of P ˆ (s, T ) undefined
n
2.2 Adiabatic Time-Evolution 11
in the adiabatic limit, thus invalidating the prerequisites for the adiabatic theorem.
One can therefore not expect that the adiabatic condition to have any relation to the
adiabatic approximation for this model system.
The energy gaps Δnm (s) are essential for the adiabatic approximation. These gaps
should be continuous functions, should never vanish, and should be sufficiently large
so as to prevent transitions between different energy subspaces during the evolution.
This latter aspect can be used for performing computation. As an illustration of the
role of the gap for adiabaticity, we shall consider one such computational problem,
which is how adiabatic evolution can be used to search for a marked item in an
unstructured database in an efficient way.
Suppose we wish to find a single marked item ν in an unstructured list of
N entries. This problem can be mapped to a fixed orthonormal basis {|k }N k=1
spanning an N-dimensional Hilbert space of a quantum system. We consider the
one-parameter family of Hamiltonians
1
N
Ĥ (s) = −(1 − s)|Ψ Ψ | − s|ν ν|, |Ψ = √ |k , (2.17)
N k=1
where s = t/T ∈ [0, 1], T being the run-time. Ĥ (s) has a decoupled (N − 2)-
fold degenerate zero energy excited eigensubspace and two lower non-degenerate
eigenstates |Eg (s) and |Ee (s) with corresponding energies Eg (s) < Ee (s) < 0,
defining the non-zero energy gap function,
1 + (N − 1)(2s − 1)2 1
Δ(s) = Ee (s) − Eg (s) = ≥√ , (2.18)
N N
| ν|Φ(1) |2 ≥ 1 − 2 , (2.19)
12 2 Theory of Adiabatic Evolution
where we have used (2.18). By diagonalising Ĥ (s), one can show that
Ee (s)| d Ĥ (s)|Eg (s) ≤ 1, (2.21)
ds
N
T ≥ . (2.22)
Thus, the time required to find the marked item with a given accuracy scales as N,
which coincides with the scaling of classical search.
However, we can do better by optimising the adiabatic evolution. The idea is to
vary the speed so that the evolution slows down when the minimal gap between the
ground and first excited states is approached. In other words, one looks for time-
local reparametrisations t → s = s(t) ∈ [0, 1] that can improve how the run-time
T scales with the size N of the database. This is achieved by requiring the adiabatic
condition in (2.20) to hold locally at each infinitesimal time step, i.e.
d
Ee (s)| ds Ĥ (s)|Eg (s)
|ṡ(t)| ≤ . (2.23)
Δ2 (s)
Equation (2.21) implies that ṡ(t) = Δ2 (s) satisfies (2.23) and can be integrated to
yield
1 N √ √
t (s) = √ arctan( N − 1(2s − 1)) + arctan N − 1 . (2.24)
2 N − 1
It can be proved that (2.24) is the optimal choice of reparametrisation [5]. We show
the function s(t) on [0, T ] in Fig. 2.1 for N = 4, 16, 36, and 64 with 96% success
probability (i.e. = 0.2). The run-time of the search is
1 N √ π√
T = t (1) = √ arctan N − 1 ≥ N (2.25)
N −1 2
√
for large N. The linear scaling with N is clearly visible in Fig. 2.1. This quadratic
speed-up is achieved by utilising the central role of energy gaps in adiabatic
evolution.
2.3 Gauge Structure of Time-Dependent Adiabatic Systems 13
s
1
0.5
t
25 50
Fig. 2.1 Scaling of the run-time for searching a marked item out of N items, where N = 4
(dotted), N = 16 (dashed), N = 36 (dashed-dotted), and N = 64 (solid). The accuracy is set
to = 0.2, meaning √ that the success probability is 96%. We see that the run-time T = t (1)
scales linearly with N. Note how the system spends increasingly more time in the vicinity of the
minimum energy gap at s = 12 in order to satisfy the local adiabatic condition in (2.23)
gn R(s)
cn;k (s) = δkl − dR(s ) · ψn,k (R(s ))|∇R(s)ψn,l (R(s ))
l=1 R(0);C
gn
R(s)
+ dR(s ) · ψn,k (R(s ))|∇R(s ) ψn,l (R(s ))
l =1 R(0);C
R(s )
× dR(s ) · ψn,l (R(s ))|∇R(s ) ψn,l (R(s )) − · · · cn;l (0),
R(0);C
(2.27)
R(s )
where we have used the chain rule d/ds = Ṙ(s) · ∇R . The notation R(0);C dR(s )·
stands for a path integral along C , starting at R(0) and ending at a point R(s )
between the end-points of C . Since all path integrals take place over portions of C ,
it follows that one can formally rewrite (2.27) at s = 1 as
gn
Aαn (R)dRα
cn;k (1) = Pei C cn;l (0), (2.28)
kl
l=1
where P is path ordering along C and sum over repeated parameter indices is
understood. Here,
gn
|Φ(0) = |ψn;k (R(0)) cn;k (0) (2.30)
k=1
2.3 Gauge Structure of Time-Dependent Adiabatic Systems 15
transforms into
1
gn
|Φ(1) = e−iT 0 εn (R(s))ds
|ψn;k (R(0)) Pei C Aαn (R)dRα
cn;l (0) (2.31)
kl
k,l=1
α
after completing the loop. The unitary matrix Un;kl (C ) ≡ Pei C An (R)dRα
kl
is the Wilczek–Zee holonomy or non-Abelian geometric phase associated with
the adiabatic
1
loop C in parameter space. Note that the dynamical phase factor
−iT 0 εn (R(s))ds
e is factored out as a global phase with no observable effect on the
state change.
The above synthetic gauge connection (2.29) is a quantity that we will meet
throughout this book. It constitutes the basis for a synthetic gauge theory associated
with adiabatic evolution. To see this, we first state the meaning of gauge choice and
gauge transformations. The chosen basis of the eigensubspace is the gauge choice
and a gauge transformation is a smooth local basis change,
gn
n;k (R(s)) =
|ψn;k (R(s)) → |ψ |ψn;l (R(s)) Vlk (R(s)), (2.32)
l=1
under which the synthetic gauge connection transforms as a proper gauge potential,
α (R) = i ψ
Aαn;kl (R) → A n;k (R)|∂ α ψ
n;l (R)
n;kl
gn
†
gn
†
= Vkp (R)Aαn;pq (R)Vql (R) + i Vkp (R)∂ α Vpl (R). (2.33)
p,q=1 p=1
The inhomogeneous term on the right-hand side shows that the gauge connection
is not directly accessible experimentally, just as the vector potential in electro-
magnetism is a gauge dependent quantity and therefore is not uniquely defined by
the observable electromagnetic field. On the other hand, experimentally accessible
quantities should transform unitarily under gauge transformations in order for
expectation values and probabilities to be basis independent. Indeed, one finds that
the Wilczek–Zee holonomy transforms unitarily,
gn
†
Un;kl (C ) → Vkp (R(0))Un;pq (C )Vql (R(0)) (2.34)
p,q=1
under the smooth basis change in (2.32). Similarly, the anti-symmetric curvature
tensor, defined as
αβ
transforms unitarily under a gauge transformation. Fn is the non-Abelian analogue
of the electromagnetic curvature tensor, which defines the electromagnetic field.
In particular, note how the last commutator term vanishes identically in the non-
degenerate case gn = 1.
The curvature tensor tells us whether the gauge connection is curved or not, i.e.
whether the gauge structure can have a physical influence on the adiabatic evolution
or not. In particular, the curvature tensor has an immediate relation to the energy
gaps of the system, as can be seen by rewriting (2.35) by means of the instantaneous
Schrödinger equation H (R)|ψn;l (R) = εn (R)|ψn;l (R) yielding
αβ
Fn;kl (R)
K
gm
ψn;k (R)|∂ α H (R)|ψm;p (R) ψm;p (R)|∂ β H (R)|ψn;l (R)
=i
Δ2nm (R)
m=n=1 p=1
In other words, the curvature tensor is singular where the energy gaps close. We shall
see in the examples below that the vicinity regions of ‘intersection points’ where two
or more energies cross (become degenerate) are associated with a gauge structure
that resembles that of a magnetic monopole. Therefore, energy intersection points
play a central role as a source of the gauge field in the space of slow parameters in
the system.
Geometrically, the synthetic gauge connection can be understood in terms of
a natural concept of parallel transport over the eigensubspaces of the slowly
changing Hamiltonian. To see this, let us consider two nearby gn -dimensional
eigensubspaces L (R(s)) and L (R(s +δs)) along C with eigenprojectors P̂n (R(s))
gn
and P̂n (R(s + δs)), respectively. Let Wn (R(s)) = {|ψn;k (R(s)) }k=1 and Wn (R(s +
gn
δs)) = {|ψn;k (R(s + δs)) }k=1 be two frames (ordered bases) spanning the two
subspaces. These frames can be made parallel by minimising the function [8]
gn
D 2 (Wn (s), Wn (s + δs)) = |ψn;k (R(s)) − |ψn;k (R(s + δs)) 2
k=1
= 2gn − 2ReTrO (Wn (R(s)), Wn (R(s + δs))) (2.37)
over all possible choices of frame pairs. The minimum gives a gauge invariant notion
of distance between the eigensubspaces L (R(s)) and L (R(s + δs)), containing the
Fubini–Study distance [9] as the special case for which gn = 1. The key quantity in
the minimisation procedure is apparently the overlap matrix
Okl (Wn (R(s)), Wn (R(s + δs))) = ψn;k (R(s))|ψn;l (R(s + δs)) (2.38)
2.3 Gauge Structure of Time-Dependent Adiabatic Systems 17
between the two frames. This minimum is found by making the gauge transforma-
tion
gn
n;l (R(s)) =
|ψn;l (R(s)) → |ψ |ψn;j (R(s)) Vj l (R(s)) (2.39)
j =1
such that the corresponding overlap matrix ψ n;l (R(s + δs)) becomes
n;k (R(s))|ψ
positive definite to first order in δs. This yields the connection (rule for parallel
transport)
Thus, the synthetic gauge connection Aαn (R(s)) determines the rule for parallel
transport over the eigensubspaces along the path C in parameter space. The
Wilczek–Zee unitary Un (C ) is thereby the holonomy arising when the system is
parallel transported around a loop.
18. Kiviaika.
19. Kameleontti.
Heti kun talo oli tullut valmiiksi, sai äiti Rautakala kohteliaan
käskyn käydä sisään, minkä tämä tekikin vähän aikaa kursailtuaan.
Se laski pesään mätimunansa — sieviä pikku hiukkasia, eivät
unikukan siemeniä isompia — ja sitten se koputti vastapäiseen
seinään aukon, josta ui tiehensä. "Hän ei joutanut hoitamaan munia",
sanoi se, "sillä hänellä oli yllin kyllin muutakin työtä; mutta jos isä
Rautakala ottaisi niistä huolehtiakseen — niin olkaa hyvä vain!"
Ensimmäinen ääni.
Toinen ääni:
Kolmas ääni:
25. Mammutti.
26. Kääpiöpuita.
Japanissa ja myöskin Kiinassa on tapana kasvattaa kaikenlaisia
kääpiöpuita keskeyttämällä mahlan juoksu nuorissa taimissa. Muuan
Japanissa matkustellut englantilainen kasvientutkija kertoo siitä:
"Koska japanilaisten talojen ja myymälöiden alakerta on avoin
edestä ja takaa, voin katuja kulkeissani tuon tuostakin vilkaista
heidän sieviin pikku puutarhoihinsa; ja huomatessani jonkun toisia
paremmaksi en suinkaan laiminlyönyt käydä sitä lähemmältä
tarkastamassa. Monet näistä puutarhoista ovat erittäin pienet, jotkut
tuskin europalaista ruokasalia suuremmat; mutta niiden pinta on
saatu vaihtelevaksi pienillä turvepenkereillä, joille on istutettu
omituisiin muotoihin leikeltyjä kääpiöpuita, sekä pienoislammikoilla,
joissa kulta- ja hopeakalat pulahtelevat."