Textbook Advances in Quantum Chemistry Ratner Volume John R Sabin and Erkki J Brandas Eds Ebook All Chapter PDF

Advances in Quantum Chemistry
Ratner Volume John R. Sabin And Erkki

J. Brändas (Eds.)
Visit to download the full and correct content document:
https://textbookfull.com/product/advances-in-quantum-chemistry-ratner-volume-john-r
-sabin-and-erkki-j-brandas-eds/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Advances in the Theory of Quantum Systems in Chemistry

and Physics 1st Edition Erkki J. Brändas (Auth.)
https://textbookfull.com/product/advances-in-the-theory-of-
quantum-systems-in-chemistry-and-physics-1st-edition-erkki-j-
brandas-auth/
Quantum Systems in Chemistry and Physics Progress in

Methods and Applications 1st Edition Erkki J. Brändas
(Auth.)
https://textbookfull.com/product/quantum-systems-in-chemistry-
and-physics-progress-in-methods-and-applications-1st-edition-
erkki-j-brandas-auth/
Advances in Organometallic Chemistry Volume 65 Pedro J.

Pérez (Eds.)
https://textbookfull.com/product/advances-in-organometallic-
chemistry-volume-65-pedro-j-perez-eds/

Pérez (Eds.)
Pérez (Eds.)
Advances in Heterocyclic Chemistry Volume 113 1st

Edition Alan R. Katritzky
https://textbookfull.com/product/advances-in-heterocyclic-
chemistry-volume-113-1st-edition-alan-r-katritzky/
Quantum Mechanics for Chemistry Seogjoo J. Jang
https://textbookfull.com/product/quantum-mechanics-for-chemistry-
seogjoo-j-jang/
Ideas of Quantum Chemistry: Volume 1: From Quantum

Physics to Chemistry 3rd Edition Lucjan Piela
https://textbookfull.com/product/ideas-of-quantum-chemistry-
volume-1-from-quantum-physics-to-chemistry-3rd-edition-lucjan-
piela/
Advances in Organometallic Chemistry 1st Edition Pedro

J. Pérez
chemistry-1st-edition-pedro-j-perez/
EDITORIAL BOARD
Remigio Cabrera-Trujillo (UNAM, Mexico)

Hazel Cox (UK)
Frank Jensen (Aarhus, Denmark)
Mel Levy (Durham, NC, USA)
Jan Linderberg (Aarhus, Denmark)
Svetlana A. Malinovskaya (Hoboken, NJ, USA)
William H. Miller (Berkeley, CA, USA)
John W. Mintmire (Stillwater, OK, USA)
Manoj K. Mishra (Mumbai, India)
Jens Oddershede (Odense, Denmark)
Josef Paldus (Waterloo, Canada)
Pekka Pyykko (Helsinki, Finland)
Mark Ratner (Evanston, IL, USA)
Dennis R. Salahub (Calgary, Canada)
Henry F. Schaefer III (Athens, GA, USA)
John Stanton (Austin, TX, USA)
Alia Tadjer (Sofia, Bulgaria)
Harel Weinstein (New York, NY, USA)
Academic Press is an imprint of Elsevier
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
525 B Street, Suite 1800, San Diego, CA 92101-4495, United States
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
125 London Wall, London, EC2Y 5AS, United Kingdom
First edition 2017
Copyright © 2017 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency, can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by
the Publisher (other than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices,
or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described
herein. In using such information or methods they should be mindful of their own safety and
the safety of others, including parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.
ISBN: 978-0-12-812888-6
ISSN: 0065-3276
For information on all Academic Press publications

visit our website at https://www.elsevier.com/books-and-journals
Publisher: Zoe Kruze

Acquisition Editor: Kirsten Shankland
Editorial Project Manager: Shellie Bryant
Production Project Manager: Vignesh Tamil
Cover Designer: Christian J. Bilbow
Typeset by SPi Global, India
CONTRIBUTORS
Adam P. Ashwell
Northwestern University, Evanston, IL, United States
Asbjørn Bols
Department of Chemistry, University of Copenhagen, Copenhagen, Denmark
Eric A. Buchanan
University of Colorado, Boulder, CO, United States
R. Benny Gerber
The Hebrew University of Jerusalem, Jerusalem, Israel; University of California, Irvine, CA,
United States
Thorsten Hansen
Zdeněk Havlas
University of Colorado, Boulder, CO, United States; Institute of Organic Chemistry and
Biochemistry, Academy of Sciences of the Czech Republic, Prague, Czech Republic
Barak Hirshberg
The Hebrew University of Jerusalem, Jerusalem, Israel
Joshua Jortner
School of Chemistry, Tel Aviv University, Tel Aviv, Israel
Isidore Last
School of Chemistry, Tel Aviv University, Tel Aviv, Israel
Jan Linderberg
Aarhus University, Aarhus C, Denmark; Henry Eyring Center for Theoretical Chemistry,
University of Utah, Salt Lake City, UT; Quantum Theory Project, University of Florida,
Gainesville, FL, United States
Josef Michl
University of Colorado, Boulder, CO, United States; Institute of Organic Chemistry and
Biochemistry, Academy of Sciences of the Czech Republic, Prague, Czech Republic
Kurt V. Mikkelsen
Jens Oddershede
University of Southern Denmark, Odense, Denmark; University of Florida, Gainesville, FL,
United States
John F. Ogilvie
University of Southern Denmark, Odense, Denmark; Simon Fraser University, Burnaby,
BC, Canada; Universidad de Costa Rica, San Jose, Costa Rica
ix
x Contributors
Stine T. Olsen
John W. Perram
School of Mathematics and Statistics, University of New South Wales, Kensington, NSW,
Australia
Mark A. Ratner
John R. Sabin
University of Southern Denmark, Odense, Denmark; University of Florida, Gainesville, FL,
United States
Michael Sabio
Stevens Institute of Technology, Center for Healthcare Innovation, Hoboken; 3D-2Drug,
Fair Lawn, NJ, United States
Stephan P.A. Sauer
University of Copenhagen, Copenhagen, Denmark
George C. Schatz
Sid Topiol
Stevens Institute of Technology, Center for Healthcare Innovation, Hoboken; 3D-2Drug,
Fair Lawn, NJ, United States
PREFACE
It is with the greatest pleasure that we present this volume in tribute to Mark
Ratner on the occasion of his 75th birthday for his seminal contributions to
theoretical chemistry during the last half century.
Mark Ratner began his career as an undergraduate in chemistry at
Harvard University, where he earned the AB degree in 1964. He then
did a PhD in chemistry at Northwestern University in 1969, working with
Prof. G. Ludwig Hofacker, followed by postdoctoral work with Prof. Jan
Linderberg in Århus, Denmark. Mark took his first job at New York
University, and in 1975 moved to Northwestern University where he
remains. Presently, Mark Ratner is Lawrence B. Dumas Distinguished
University Professor Emeritus Departments of Chemistry, Materials Science,
and Engineering at Northwestern University. While at Northwestern, he has
held several positions such as Chairman of the Chemistry Department and
Associate Dean of the College of Arts and Sciences.
Mark Ratner is a very personable fellow, which makes him an ideal
colleague to work with. He is friendly and helpful, and, perhaps most impor-
tant, an original and deep thinker in many aspects of theoretical chemistry;
he leads a scientific life dominated by a quest for understanding of chemical
structure and function at the nanoscale and the theory that will describe and
predict such in many nanoscale subfields. These characteristics have led to
a world-class research program under Mark Ratner’s direction. Mark Ratner
is not only a producer of world-class research, but he is prolific. Presently
he has produced over 800 technical papers, 5 edited volumes, and 4 books.
xi
xii Preface
It should be noted that of the books, two of them are coauthored by his son,
Daniel Ratner, and the other two are written together with George Schatz.
In addition, he has given over 100 named lectures around the world from
Denmark, to Israel, to Japan, and on.
Perhaps Mark Ratner’s best known work is in the field of molecular
electronics. In 1974, Mark Ratner and his at that time student, Ari Aviram,
produced a work on molecular rectifiers [Bull. Am. Phys. Soc. 19, 341
(1974); Chem. Phys. Lett. 29, 277 (1974)] which is probably his best known
work. This was the first time it had been shown, as inconceivable as it was by
most experts in the field, that a single molecule could act as a rectifier. From
their work, a whole field of molecular electronics has blossomed. However,
Mark Ratner’s work has not been confined to this particular subfield, as he
has significantly contributed in many other related areas. For instance, he
has contributed substantially to work on molecular electron transfer and
electron transport in organic polymers as well as to various aspects of
molecular dynamics and relaxation.
Mark Ratner’s achievements have won him many honors. Most presti-
gious, he is a member of the National Academy of Science (2002) and a
foreign member of the Royal Danish Academy of Science (2004). In
addition he has received honorary Doctor of Science degrees from the
Hebrew University of Jerusalem (2004) and the University of Copenhagen
(2010). In 2010, Mark Ratner was also celebrated and honored with a fest-
schrift published in the Journal of Physical Chemistry C (volume 114, #48).
It has certainly been a great honor for those of us whom have had
interaction, scientific and social, with Mark Ratner, to collect this tribute
to him. He has been a vibrant member of the Quantum Chemistry Club of
those individuals who have practiced the science for nearly a half century,
and we wish him the very best for the future.
The papers presented here are written by Mark Ratner’s friends and
coworkers to demonstrate their respect for him and his work.
Best wishes, Mark!
By: JOHN R. SABIN
€
ERKKI J. BRANDAS
CHAPTER ONE
Mean-Field Methods for Time-

Dependent Quantum Dynamics of
Many-Atom Systems
Barak Hirshberg*, R. Benny Gerber*,†,1
*The Hebrew University of Jerusalem, Jerusalem, Israel
†
University of California, Irvine, CA, United States
1
Corresponding author: e-mail address: bgerber@uci.edu
Contents
1. Introduction 2
2. Theoretical Framework, Advantages, and Limitations 5
2.1 Vibrational Self-Consistent Field 5
2.2 Time-Dependent Self-Consistent Field 6
2.3 Mixed Quantum–Classical Methods 8
2.4 Classical Separable Potentials 10
3. Recent Developments and Applications 12
3.1 Ab Initio Classical Separable Potentials 12
3.2 Mixed Quantum–Classical Methods 18
4. Summary and Future Directions 20
Acknowledgments 22
References 22
Abstract
Methods that can accurately describe the quantum dynamics of large molecular sys-
tems have many potential applications. Since numerical solution of the time-dependent
Schro€dinger equation is only possible for systems with very few atoms, approximate
methods are essential. This paper describes the development of such methods for this
challenging time-dependent many-body quantum mechanical problem. Specifically,
we focus on the development of mean-field theories, to which Mark Ratner has contrib-
uted greatly over the years, such as the time-dependent self-consistent field method,
mixed quantum–classical methods, and the classical separable potentials method.
The advantages and limitations of the different variants of mean-field theories are
highlighted. Recent developments, aimed at applying mean-field methods for large sys-
tems, and their applications are presented. Issues where further methodological
advancement is desirable are discussed. Examining the tools available so far, and the
Advances in Quantum Chemistry, Volume 75 # 2017 Elsevier Inc. 1

ISSN 0065-3276 All rights reserved.
http://dx.doi.org/10.1016/bs.aiq.2017.01.002
2 Barak Hirshberg and R. Benny Gerber
recent progress, we conclude there are promising perspectives for future development
of mean-field theories for quantum dynamics with applications to realistic systems in
important chemical and physical processes.
1. INTRODUCTION
Classical molecular dynamics, in which the motion of the atoms is
described using Newton’s equations of motion, is one of the most widely
applied methods in chemistry. This includes methods in which the interac-
tions between the atoms are described using empirical force-fields1 as well as
ab initio molecular dynamics (AIMD),2–5 which directly employs potential
energy surfaces (PES) obtained from quantum chemical methods to the clas-
sical nuclear equations of motion. The scope of applications is overwhelm-
ing and ranges from the study of biological molecules,6,7 properties of water
and ice,8–10 materials science,11 chemical reaction dynamics,12 and many
more. Describing the motion of the nuclei using classical mechanics is often
an excellent approximation at typical chemical conditions, e.g., room tem-
perature. However, in many cases, nuclear quantum effects (NQE) such as
zero-point energy (ZPE), tunneling, and quantum interference may play a
significant role, especially for light atoms. Important example is liquid water
and ice, and a detailed review on the importance of NQE for water and ice
was published recently by Ceriotti et al.13 Other examples include molecules
embedded in superfluid helium droplets,14,15 the kinetic isotope effect in
enzyme catalysis,16–18 and more. As a result, for these processes and others,
a quantum description of the nuclei is essential.
Ideally, one would like to numerically solve the time-dependent
Schr€ odinger equation (TDSE) for the nuclei using a suitable Born–
Oppenheimer (BO) PES. However, due to the exponential scaling of the
computational effort with the number of degrees of freedom,19 this is an
extremely difficult task. State-of-the-art calculations in full dimensionality
are currently limited to systems with approximately 4–6 atoms,20 such as
the recent study of electron photodetachment from F(H2O).21 Since most
chemical and physical processes of interest occur in larger systems, approx-
imate methods for quantum nuclear dynamics are necessary.
Several different approximate methods are available which address, con-
ceptually, NQE. This includes semiclassical methods such as the semiclassi-
cal initial value representation method by Miller22 and the Gaussian
wavepacket method by Heller23,24 as well as methods based on the
Mean-Field Methods for Time-Dependent Quantum Dynamics of Many-Atom Systems 3
Path-integral formulation of quantum mechanics, such as centroid molecu-

lar dynamics25,26 and ring-polymer molecular dynamics (RPMD).27 This
overview will focus on a particular class of approaches, mean-field methods
for quantum nuclear dynamics, such as the time-dependent self-consistent
field method (TDSCF),28–30 the classical separable potentials (CSP)
method,31 and mixed quantum–classical methods.28,32 The reason for this
focus is that mean-field methods have been very successful in describing cer-
tain time-independent many body problems, such as vibrational spectros-
copy. Unlike time-dependent methods, mean-field theories such as the
vibrational self-consistent field (VSCF) method33,34 are now routinely
applied using ab initio PES35 to many-atom systems36 including, recently,
even peptides.37,38 This suggests that the same may be possible for mean-
field time-dependent methods, in spite of the added computational effort
due to the time-dependence.
Several review articles were previously published on approximate theo-
ries for quantum nuclear dynamics: A masterful review on many of the
methods discussed in this paper was written by Makri39 over 15 years
ago. Excellent reviews focusing on specific methods not discussed in this
paper were also written recently, such as the reviews by Habershon et al.
on RPMD27 and by Meyer on the multiconfigurational time-dependent
Hartree method.40 It is the purpose of this paper to give an updated account
on the state of the field, with particular emphasis on mean-field methods, on
their implementation to systems of substantial numbers of degrees of free-
dom, on the possibility of using directly ab initio potentials in these methods
and on the connection between time-dependent and time-independent
mean-field methods. In addition, we limit the scope of this paper to situa-
tions in which the BO approximation is adequate and a single adiabatic PES
is considered.
Perhaps the first mean-field theory used to describe time-dependent
quantum dynamics is the TDSCF method which originates from the work
of Dirac.41 One of its first applications to chemical physics was to the vibra-
tional predissociation of a van der Waals molecule, I2Ne.28 In this method, a
separable ansatz for the nuclear wave function is assumed, typically using
normal coordinates. Applying the Dirac–Frenkel time-dependent varia-
tional principle42,43 to this ansatz results in a set of coupled equations for
the vibrational modes. Each mode evolves in time under the influence of
a mean potential which is the quantum mechanical average of the full
PES over all other degrees of freedom. Classical and semiclassical variants
of TDSCF were also introduced.28,44 TDSCF was shown to give good
agreement with numerical solution of the nuclear TDSE for several pro-
cesses such as vibrational predissociation of I2He45 as well as photodissoci-
ation of Xe-HI46 and Ar-HCl.47 While the computational complexity of the
TDSCF method is significantly reduced in comparison to numerical solu-
tion of the TDSE, the evaluation of the mean potentials involves multi-
dimensional integrals over all modes of vibration except one, which are
the computational bottleneck of TDSCF. As a result, TDSCF was only
applied to relatively small systems.
An approximation which, when justified, can significantly decrease the
computational complexity of TDSCF is the treatment of a small subset of the
system using TDSCF, while treating the other degrees of freedom in the
classical limit of TDSCF. This family of methods is often referred to as mixed
quantum–classical methods.29 More precisely, the quantum degrees of free-
dom evolve in time according to a TDSCF-like equation; however, the mean
potential for the quantum subsystem is obtained from classical trajectories.
Accordingly, the classical degrees of freedom are not described by a wave
function, and instead evolve in time classically under the influence of a force
which is the quantum mechanical average of the full potential over the quan-
tum degrees of freedom. Using quantum–classical TDSCF and employing an
additional approximation, describing each mode using Gaussian wavepackets,
low temperature clusters such as Ar13 and (H2O)548 and inelastic scattering of
argon off (H2O)1149 were studied.
Another approach which was successful in incorporating quantum effects
in significantly larger systems is the CSP method.31 In this method, which is
based on the TDSCF method, the quantum mechanical mean potential
governing the time-dependent dynamics of each mode is replaced by an
average potential obtained from classical trajectories simulations, using prop-
erly sampled initial conditions. A key point here, as well as in mixed
quantum–classical dynamics, is that the initial conditions sampled for the
classical trajectory simulations represent the classical analog for the quantum
mechanical process studied. The main difference between CSP and mixed
quantum–classical methods is that in CSP while all degrees of freedom are
described quantum mechanically, there is no quantum force acting on the
nuclei in the classical trajectories used to build the effective potentials for
each mode. The average potentials obtained from classical molecular
dynamics simulations implicitly couple the different modes of vibration to
each other through the time-dependence of the potential. As a result, the
single-mode equations are not solved self-consistently in the CSP method.
While the lack of self-consistency may give rise to errors, it was shown that,
for several test cases, CSP gives results in good agreement with TDSCF
results.31 Using CSP, much larger systems were studied such as the electron
photodetachment in C60 .50 In this paper, we will also present a recent
extension of CSP, denoted ab initio CSP (AICSP),51 which uses potentials
directly obtained from quantum chemical methods. Recent application to
test the validity of AICSP to vibrational spectroscopy of realistic systems will
be discussed.
The remainder of the paper is organized as follows: Section 2 presents the
theory of VSCF, TDSCF, mixed quantum–classical methods, and the CSP
method. Section 3 describes new methodological developments, which are
aimed at describing large systems using ab initio PES, and their recent appli-
cations. Section 4 of the paper summarizes and presents current challenges of
mean-field methods and possible future directions.
2. THEORETICAL FRAMEWORK, ADVANTAGES,

AND LIMITATIONS
2.1 Vibrational Self-Consistent Field
We start by briefly discussing the VSCF method and several issues involved in
its practical application, which are also relevant for the development of time-
dependent mean-field methods.The time-independent nuclear Schr€ odinger
equation is given by
^ ðnÞ ¼ E ðnÞ XðnÞ
HX (1)
where Χ (n) and E(n) are the total nuclear wave functions and energies, respec-
tively, corresponding to state n and H ^ is the Hamiltonian of the system. In
the framework of VSCF, neglecting rotational–vibrational coupling, the
Hamiltonian is written in mass-weighted normal coordinates
2X
^ ¼ ℏ @2
N
H + V ðQ1 , …, QN Þ (2)
2 j¼1
@Qj2
where Q1, …, QN are the mass-weighted normal coordinates, V(Q1, …, QN)

is the full PES, and N represents the number of vibrational degrees of free-
dom. While normal coordinates are often a good choice for describing
vibrational spectroscopy, other coordinate systems can be beneficial, such
as local coordinates52 or internal coordinates.53 For a detailed discussion
of the optimal coordinates choice for VSCF, see the review by Roy and
Gerber36 and references therein.
Next, it is assumed that the total nuclear wave function can be represen-
ted by a simple product of single-mode functions
Y
N
Χ ðnÞ ðQ1 , …, QN Þ ¼ χ j ðnÞ Qj (3)
j¼1
Applying the time-independent variational principle results in the VSCF

single-mode equations
ℏ2 @ 2 χ k ðnÞ ðQk Þ ðnÞ

+ V k ðQk Þχ k ðnÞ ðQk Þ ¼ εk ðnÞ χ k ðnÞ ðQk Þ (4)
2 @Qk2
where the effective single-mode potential is given by the quantum mechan-
ical average of the full PES over all other modes of vibration
* +
Y N Y N
ð Þ ð Þ ð Þ
V k ðQk Þ ¼
n n n
χ j Qj V ðQ1 , …, QN Þ χ j Qj (5)
j6¼k
j6¼k
Since the effective potential for each mode depends on all other modes,
the VSCF equations are solved self-consistently. The main computational
difficulty in applying VSCF lies in the evaluation of the multidimensional
integrals involved in calculating the effective potentials. This computational
complexity is significantly reduced if the potential can be approximated, to a
reasonable extent, by a pairwise approximation:
X
N
diag XN X
N
coup
V ðQ1 , …, QN Þ ¼ Vj Qj + Vij Qi , Qj (6)
j¼1 i j>i
The effective potentials then only involve integrals over 2 degrees of

freedom. For many cases, the level of agreement obtained using this approx-
imation with experimental results is very good.54 If needed,55 higher order
coupling between specific modes can also be introduced.
2.2 Time-Dependent Self-Consistent Field

In essence, TDSCF is the time-dependent analog of VSCF. Accordingly, the
starting point for deriving the TDSCF equations is the time-dependent
nuclear Schr€odinger equation
@Χ ^
iℏ ¼ HΧ (7)
@t
where H, ^ as in the previous section, is the Hamiltonian of the system and Χ

is the total nuclear wave function. The question arises, which coordinates
should be used to describe the time-dependent dynamics of the system?
For VSCF, which is mainly designed for vibrational spectroscopy, normal
coordinates are often a very suitable choice, despite the fact other choices
may improve the accuracy for specific modes. However, the choice of coor-
dinates for TDSCF is not as straightforward and the selection of optimal
coordinates for TDSCF has been far less studied than for VSCF. To start
with, it is not generally obvious that the dynamics can be restricted to the
vicinity of a single minimum. Jungwirth et al. suggested56 a simple proce-
dure, based on a set of classical trajectories, to evaluate the suitability of dif-
ferent coordinates for TDSCF and applied it to the relaxation dynamics
following electron photodetachment in clusters such as IAr12. The coor-
dinates used should be those that maximize the separability of the different
modes during the time-dependent process. For processes which are domi-
nated by a single minimum on the PES, one expects normal coordinates to
be a reasonable choice, similarly to VSCF. In addition, one can also use nor-
mal modes at the transition state57–59 which can provide a good description
of the transition state dynamics. There is no general method to determine
optimal coordinates for all cases.
We confine our attention to dynamics around a single minimum or sad-
dle point. For convenience, we will use normal coordinates Q1, …, QN to
present the TDSCF equations. The Hamiltonian is then given by Eq. (2).
Assuming that the total nuclear wave function Χ(Q1, …, QN, t) can be rep-
resented by a single product of time-dependent single-mode functions
χ j(Qj, t)
Y
N
ΧðQ1 , …, QN , t Þ ¼ χ j Qj , t (8)
j¼1
And applying the Dirac–Frenkel variational principle42,43
^ iℏ @
hδΧjH jΧi ¼ 0 (9)
@t
where δΧ is the variation in Χ with respect to the single-mode function χ k,
we get the following TDSCF single-mode equations
@ϕk ðQk , t Þ ℏ2 @ 2 ϕk ðQk , tÞ

iℏ ¼ + V TDSCF ðQk , t Þϕk ðQk , t Þ (10)
@t 2 @Qk2 k
where ϕk(Qk, t) differs from χ k(Qk, t) by a physically insignificant purely

time-dependent phase factor and the single-mode effective potential
V TDSCF
k ðQk , t Þ is given by
* +
Y N Y

N
V k
TDSCF
ðQk , tÞ χ j Qj , t V ðQ1 , …, QN Þ χ j Qj , t (11)
j6¼k
j6¼k
Similarly to VSCF, the main computational bottleneck for applying

TDSCF is the evaluation of the multidimensional integrals that appear in
the expression for the effective single-mode potentials. The complexity,
however, is greater than for VSCF, since the integrals have to be reevaluated
at every time step. Similarly to VSCF, the computational demand can be
significantly reduced by using the pairwise approximation, expanding the
potential according to Eq. (6). This approach for reducing the computa-
tional complexity of the TDSCF method has not been explored extensively
so far. It is likely this will be somewhat less successful than in VSCF, because
the configurational space relevant for dynamics is in practice larger. While
the integrals that appear in the TDSCF equations still need to be evaluated at
every time step, it will undoubtedly significantly increase the size of systems
that can be studied using TDSCF. We note that while the validity of the
pairwise approximation for time-dependent processes needs to be
reevaluated, the success of this approximation within the framework of
VSCF suggests it may be useful in time-dependent problems which are
dominated by a single minimum on the PES.
2.3 Mixed Quantum–Classical Methods

In cases when the density of states is high for certain modes, it is possible to
describe these modes, to a good approximation, by classical mechanics while
retaining the quantum description for all other degrees of freedom. This is
the central idea behind mixed quantum–classical methods. It is clear that
using this approximation, the computational complexity can be reduced,
since the quantum mechanical problem will be of lower dimension. Here,
we briefly highlight some central aspects of the theory. For an authoritative
review on mixed quantum–classical methods, the reader is referred to the
review by Tully.32
One possible starting point for mixed quantum–classical methods is the
TDSCF equations given by Eqs. (10) and (11). Then, for a mode which can
be approximately described by classical mechanics, we follow the usual
procedure60 of writing the single-mode wave function using the polar rep-
resentation (given in Eq. (12)), deriving the equations for the amplitude
Aj(Qj, t) and phase Sj(Qj, t) and taking the classical limit in which ℏ ! 0.
i
ϕj Qj , t ¼ Aj Qj , t eℏSj ðQj , tÞ (12)
Plugging Eq. (12) into Eq. (10) and separating the real and imaginary
parts, we get the equations for the amplitude and phase
2
@Sj Qj , t 1 @Sj Qj , t
+ + V TDSCF Qj , t
@t 2 @Qj j

ℏ2 @ 2 A j Qj , t
¼ (13)
2Aj Qj , t @Qj2

@Aj Qj , t @Aj Qj , t @Sj Qj , t
¼
@t @Qj @Qj

1 @ 2 Sj Qj , t
Aj Qj , t (14)
2 @Qj2
So far, Eqs. (13) and (14) are completely analogous to the TDSCF equa-
tions. Now, however, we can take the classical limit to get an equation for
the classical phase SCj (Qj, t)
!2
@SjC Qj , t 1 @Sj Qj , t
C

+ + V TDSCF Qj , t ¼ 0 (15)
@t 2 @Qj j
The equation for the classical amplitude is identical to Eq. (14), except
that the classical phase appears instead of the quantum mechanical phase in
the equation. Eq. (15) is simply the Hamilton–Jacobi equation for a particle
evolving in time classically under the influence of the potential

V TDSCF
j Qj , t . We note that V TDSCF
j Qj , t is, as before, the quantum
mechanical average of the full potential over all other degrees of freedom.
Thus, the classical modes are evolved under the influence of a force which
originates from the other quantum degrees of freedom. Accordingly, the
TDSCF equations for the quantum mechanical modes are not changed,
but the averaging of the potential over the classical modes is now done using
classical trajectories. We note that in the derivation given here, starting from
the TDSCF equations and taking the classical limit, complete separability is
assumed between all modes, including between one classical mode to

another, for example. This is not a general requirement of mixed
quantum–classical methods. An alternative derivation which is often used
to present mixed quantum–classical methods only assumes general separabil-
ity between the classical and quantum degrees of freedom, which results in
the following equations of motion for the classical nuclei, written in Carte-
sian coordinates, and can be viewed as a specialization of the Ehrenfest
theorem
!
Z
!
MI R€I ðtÞ ¼ rI drψ*ðr, tÞV ðr, RÞψ ðr, tÞ (16)
Here, ψ(r, t) is the nonseparable wavefunction for the quantum nuclei

!
and R I ðtÞ are the instantaneous position vectors of the classical nuclei.
The equations of motion for the quantum nuclei are given by
@ψ ðr, tÞ ^
iℏ ¼ T r + V^ ðr, RðtÞÞ ψ ðr, tÞ (17)
@t
where T^ r is the kinetic energy operator for the quantum nuclei and V^ is the
potential energy operator.
In both derivations, the quantum mechanical equations are solved self-
consistently with the equations of motion for the classical subsystem. We
note that while mixed quantum–classical methods allow simulations of large
systems, the quantum mechanical subsystem must still be small.
In cases where the energy or action for the classical subsystem is high,
describing some modes classically and others quantum mechanically works
very well, as shown for the photodissociation of HI in Xe clusters.61,62
However, due to the fact that some modes are not described by a wave func-
tion, significant errors can arise, such as negative absorption cross sec-
tions.63,64 Other possible discrepancies can results from the fact that the
energy for the classical modes can drop below the ZPE of the mode.
2.4 Classical Separable Potentials

The CSP method was suggested as an alternative procedure to mixed
quantum–classical dynamics in order to describe large systems. The main
assumption is that the effective single-mode potentials that appear in the
TDSCF equations can be replaced by an average potential obtained from
classical trajectories simulations. The potential is then given by
X
Ntraj

V CSP
k ðQk , t Þ ¼ Wα V Q1α ðtÞ,…, Qk , …, QN
α
ðt Þ (18)
α¼1
where Ntraj is the number of trajectories performed, Wα is the averaging

weight given to each trajectory, and Qαj (t) is the value of normal mode
Qj at time t from trajectory α. The CSP equations are then given by
@ϕk ðQk , tÞ ℏ2 @ 2 ϕk ðQk , tÞ CSP

iℏ ¼ + V k ðQk , tÞϕk ðQk , tÞ (19)
@t 2 @Qk2
In essence, this approximation means that each degree of freedom is

described quantum mechanically, evolving in time under the influence of
a mean-field potential which is generated as if all other modes of vibration
are classical.
The main difference between CSP and mixed quantum–classical dynam-
ics is the absence of a self-consistent quantum force in the classical trajecto-
ries used to build the effective potentials for the quantum degrees of
freedom. Because the effective single-mode potentials no longer depend
explicitly on all other normal modes, the CSP equations are not solved
self-consistently. While this may give rise to errors it was shown, as was
mentioned previously, that for several test cases CSP gives results in good
agreement with TDSCF,31 at least for short timescales. Another important
difference is that all degrees of freedom in CSP are treated quantum
mechanically.
A key point in the CSP method, similarly to mixed quantum–classical
dynamics, is that the initial conditions used to generate the classical trajec-
tories must represent the classical analog to the initial quantum state of the
system. For example, it is often possible to approximate the initial quantum
state using a product of harmonic single-mode functions obtained from the
PES at t ¼ 0. Then, the initial coordinates are sampled proportionally to the
square of the wave function of each normal mode. The initial momenta can
be determined from the harmonic ZPE of each mode. In other cases, if one is
interested in representing a specific quantum state, which is not approxi-
mated well by a product of harmonic functions, the initial coordinates
and momenta can be sampled from the Wigner distribution65 obtained from
the state. If the initial quantum state represents a single highly excited mode,
while all others are in their vibrational ground state, the initial conditions for
the excited mode can be sampled using the classical probability distribution
for the harmonic oscillator.
The main limitation of the CSP method, compared to TDSCF and

mixed quantum–classical methods, is the lack of self-consistency. However,
the agreement with the TDSCF results previously mentioned suggests that is
does not introduce significant errors. Its main advantage is that it is probably
the only method presented in this paper which allows a quantum mechanical
description of all nuclear modes, while being applicable to systems of realistic
size. The fact that on the course to a quantum mechanical treatment one can
obtain insights from the classical trajectories may also be useful.
3. RECENT DEVELOPMENTS AND APPLICATIONS

Despite the various methods described in the previous section, there is
no method available today which routinely allows quantum mechanical sim-
ulations of nuclear dynamics for large molecules, certainly when ab initio
PES are used. Here, we describe recent theoretical developments and appli-
cations with particular aim to treat large systems using directly potentials
from quantum chemical methods. We present detailed results only for work
done recently in our group, but also summarize promising recent advance-
ments done by other groups.
3.1 Ab Initio Classical Separable Potentials

3.1.1 Treatment of Stationary States
Classical trajectories simulations, using either empirical force-fields36 or
ab initio potentials,66,67 are often used to calculate the vibrational spectra
of molecules and extended systems. The main advantages of classical MD
simulations are that they are applicable to large systems, including condensed
phases, as well as for long simulation times. However, empirical force fields
often do not provide the necessary accuracy, for example, to be able to
determine different amino acids conformers using vibrational spectros-
copy.36 While AIMD simulations often perform better, their range of appli-
cability for calculating vibrational spectra is limited to relatively high
temperatures and it is not clear that their relative success does not result from
a fortuitous cancelation of errors. As a result, it is desirable to have a quantum
mechanical method that can describe stationary vibrational states and is
applicable to large systems using ab initio PES. In addition, we stress that
stationary vibrational spectroscopy offers a very strict test for quantum
mechanical methods, due to the highly accurate experimental results, and
the availability of data for comparison.
In order to accurately treat large systems, we recently developed51 the

AICSP method which combines CSP with potentials obtained directly from
quantum chemical methods. While the method is generally developed for
time-dependent nonequilibrium processes, it is also useful in calculating sta-
tionary states. In order to describe stationary states using the AICSP method,
one must determine a procedure to calculate the appropriate effective poten-
tials for each mode. The key point here is to select the initial conditions for
the classical trajectories simulations in a way which, at least approximately,
will be representative of a stationary state of the system at T ¼ 0 K. In the
vibrational ground state, the system has energy equal to the quantum
mechanical ZPE and no energy flow is possible between the modes. One
way to choose the initial conditions for the classical trajectories, used to build
the effective potentials, is to simply give each mode a classical energy which
is equal to its quantum mechanical ZPE. However, while in quantum
mechanical simulations this ZPE cannot flow between the modes, classically,
due to the energy differences between the modes, the energy will be trans-
ferred between the modes to a significant extent. Since these dynamics will
be significantly different than found using a quantum mechanical descrip-
tion, this procedure seems inadequate for describing a stationary state.
Alternatively, if one is interested in a specific vibrational mode, it is pos-
sible to choose the initial conditions such as to give only this mode its ZPE
and keep all other modes initially at the classical equilibrium positions. This
procedure, however, greatly underestimates the dynamical coupling
between the different vibrational modes. As a result, in describing station-
ary states within the framework of AICSP, a better approach seems to be to
choose an initial effective “classical ZPE” for each mode. This initial clas-
sical ZPE is chosen to be identical for all vibrational modes, in the spirit of
the equipartition of the energy, to avoid, at least initially, significant energy
flow between the modes. This, however, is not sufficient since this initial
total energy must be small enough as to avoid unphysical processes at zero
temperature, such as isomerization or chemical transformations. As a
result, the effective classical ZPE, given initially to each mode, was chosen
to be the harmonic ZPE for the softest vibrational mode of the system. The
use of the harmonic approximation at this point is not essential or unique
and we stress that the classical trajectories still include all anharmonic
effects.
After sampling initial coordinates and momenta using the procedure
described above, which approximately describe a stationary state, classical
MD simulations are performed and the effective single-mode potentials
are determined from Eq. (18). In order to obtain stationary frequencies and
wavefunctions for each mode, a reasonable approach seems to be to average
the time-dependent effective potentials over a time period τ, to obtain effec-
tive time-independent potentials, given by Eq. (20). τ should obviously be
long compared with the time-scale of the relevant mode
Zτ
e CSP 1
V k ð Qk Þ ¼ V CSP
k ðQk , t Þ: (20)
τ
0
Then, the one-dimensional time-independent Schr€ odinger for each

mode can be solved on a grid to obtain the anharmonic frequencies and
wavefunctions
e k ðQk Þ
ℏ2 @ 2 ϕ
e CSP e e
+V k ðQk Þϕk ðQk Þ ¼ Ek ϕk ðQk Þ: (21)
2 @Qk 2
3.1.2 Dynamical States

TDSCF, CSP, and mixed quantum–classical methods have mostly been
used to describe, naturally, the dynamics of nonequilibrium processes.
While AICSP has not yet been applied to nonstationary states, it is an inter-
esting challenge to apply it to time-dependent problems, e.g., calculating
vibrational linewidths and lineshapes. We note here that if the initial state
is not a stationary state, a very different procedure must be used for the sam-
pling of initial conditions. The main idea is again to translate the initial quan-
tum state to a set of corresponding initial coordinates and momenta for the
classical trajectories. As was mentioned previously, there are several possible
procedures to do so, depending on the process described. For situations in
which a well-defined initial quantum state can be specified, such as photo-
dissociation or electron photodetachment processes in which the initial state
is determined by the Franck–Condon principle, the Wigner representa-
tion65 of the initial state can be used. Alternatively, in cases where the process
is dominated by a single minimum on the PES, a separable approximation
(typically the harmonic approximation) can be used for the initial state. In
this procedure, initial coordinates can be sampled in proportion to the square
of the harmonic single-mode functions, while the initial momenta can be
determined from the single-mode energies.
3.1.3 Combining CSP With Ab Initio Potentials

In all previous applications of CSP, empirical potentials were used. While for
some specific processes, accurate potentials of this kind may exist, for the
majority of problems, no such reliable potentials exist and the construction
of accurate empirical potentials for complex systems is a difficult task. One
exception is the application of CSP using an approximate density functional
theory (DFT) method based on the local density approximation by Knospe
and Jungwirth50 to the electron photodetachment from C 60. The recent
development of AICSP51 allows for combining the CSP method with much
better potentials, for example, potentials based on second-order M€ oller–
Plesset perturbation theory68 (MP2) or state-of-the-art long range corrected
hybrid DFT functionals with empirical dispersion corrections, such as
ωB97X-D.69 The ability to use ab initio PES increases the accuracy of
the CSP method, as well as the applicability of CSP to systems for which
empirical potentials are not available.
As explained previously, the application of AICSP involves several steps,
mapping the initial quantum state to a set of initial coordinates and momenta
for the classical simulations, performing the classical trajectories, building the
effective potentials, and solving the single-mode time-dependent equation for
each mode. When combining CSP with ab initio potentials, the classical tra-
jectories and potential building steps become significantly more computation-
ally intensive than when empirical potentials are used. However, the potential
building step is entirely independent for each mode. As a result, parallelizing
the potential building step on many CPUs is a trivial task, which suggests that
this may allow application of AICSP to very large systems. Application of
AICSP to increasingly large systems is limited to cases for which AIMD sim-
ulations can be performed. However, classical trajectories are required only for
time scales involved in the quantum process under study, which are usually
relatively short. In addition, the power of the AIMD method and codes for
its efficient implementation are becoming increasingly available.
3.1.4 Applications
The recent application of AICSP51 focused on CH, NH and OH stretching
modes in amino acids such as glycine, alanine, and proline as well as the
guanine–cytosine (G–C) pair of nucleobases. We note that while these sys-
tems are already nontrivial in size (for example, the G–C complex has
81 vibrational modes), admittedly, the AICSP method was developed with
the intent of applying it to much larger systems, such as peptides. However,
if the method performs well for amino acids, the building blocks of peptides
and proteins, this will encourage applications also for other biological mol-
ecules. The anharmonic frequencies obtained using AICSP were in very
good agreement with experimentally measured values.51 Using MP2 poten-
tials, the deviation from experimental values for nonhydrogen-bonded OH
and NH stretching frequencies was smaller than 1%. For the hydrogen-
bonded OH stretches, the agreement was good, with deviations from exper-
imental values being approximately between 2% and 4%. In this overview,
we focus on features seen in the time-dependent effective potentials as well
as on the properties of the effective potential averaged over time, from
which the stationary vibrational frequencies are obtained. The differences
between the effective single-mode potentials, the harmonic approximation,
and the diagonal anharmonic potential, given by Eq. (22), are also discussed.
We present specific results for conformer I of alanine, shown in Fig. 1, as a
representative example
Fig. 1 Time-dependent single-mode effective potentials for different vibrational modes

of conformer I of alanine. Panel (A) presents results for the symmetric NH2 stretch, panel
(B) presents results for the asymmetric NH2 stretch, and panel (C) presents results for the
OH stretch. Panel (D) shows the optimized geometry for conformer I of alanine, carbon
atoms are shown in gray, oxygen atoms in red, nitrogen atoms in blue, and hydrogen
atoms in white. Calculations are performed using potentials at the MP2/aug-cc-pVDZ
level of theory.
diag
Vk ðQk Þ ¼ V ð0, …, Qk , …, 0Þ (22)
While the first application of AICSP was to stationary states, as was men-
tioned earlier, analyzing the time-dependent potentials still provides dynam-
ical information on classical vibrational energy flow in the system. Fig. 1
shows the different time-dependent single-mode effective potentials for sev-
eral modes of conformer I of alanine. Several interesting features can be seen:
First, fluctuations in the time-dependent effective potentials are seen initially
for all modes. However, for approximately 600 fs these fluctuations are
rather mild. Then, for the symmetric NH2 stretching mode and the OH
stretching mode, the fluctuations seem to increase and the potential becomes
softer than at t ¼ 0. This suggests that for short timescales, due to the initial
conditions chosen, energy flow between the modes is indeed minor. How-
ever, after a certain period of time, energy starts to flow significantly
between some of the modes, which results in larger fluctuations of the effec-
tive potentials. These increased fluctuations are not seen for the asymmetric
NH2 stretching mode, which fluctuates at a similar amplitude up to a sim-
ulation time of 1 ps. This may suggest that the OH stretching mode and the
symmetric NH2 stretch are more significantly coupled to other modes in the
system. The significant energy flow between the modes may also provide
information on vibrational linewidths, and we intend to pursue such analysis
in the near future.
Fig. 2 shows different time-averaged effective potentials obtained using
Eq. (20) from the time-dependent single-mode potentials presented in
Fig. 1. The shape of time-averaged effective potentials provides information
on the nature of the different modes. For example, it shows that the OH
stretch is significantly more anharmonic than the NH2 stretches. This agrees
with experimental measurements70 since a significant redshift of approxi-
mately 170 cm1 between the harmonic frequencies calculated at the
MP2 level of theory and the experimental value is seen for the OH stretching
mode, while a much lower shift of approximately 30–55 cm1 is seen for the
NH2 stretching modes. As was mentioned previously, the stationary fre-
quencies obtained using the time-averaged potentials are in very good agree-
ment with experimental measurements.
A comparison of the time-averaged effective potentials with the har-
monic potential and the diagonal anharmonic potential also provides addi-
tional information on the different modes. The comparison for the different
modes of conformer I of alanine is given in Fig. 3. The main feature seen is
that the time-averaged potentials are often very similar to the diagonal
Fig. 2 Time-averaged effective single-mode potentials obtained using Eq. (20) for dif-
ferent vibrational modes of conformer I of alanine. Blue diamonds represent the sym-
metric NH2 stretch, orange circles the asymmetric NH2 stretch, and yellow squares
present the OH stretching mode. Results were obtained using MP2/aug-cc-pVDZ level
of theory.
potential. This is not surprising since, especially for fundamentals, the diag-
onal accounts for a significant portion of the anharmonic correction, which
is often called the intrinsic anharmonicity. The time averaging over the clas-
sical trajectories accounts for the different interactions between the modes.
This effect is subtle, but very important in order to obtain spectroscopically
accurate results for modes with significant anharmonicity, such as OH
stretches. A comparison to the harmonic potentials shows that indeed,
the asymmetric NH2 stretch can be well approximated by the harmonic
approximation even for energies well above the fundamental transition. This
agrees with the small deviation of the harmonic approximation from the
experiment for this mode which is about 1% at the MP2 level of theory.
On the other hand, this analysis also emphasizes the significant anharmonic
character of the OH stretching mode for which anharmonic treatment is
essential.
3.2 Mixed Quantum–Classical Methods

In recent years, several important advancements were made in applying
mixed quantum–classical methods to large molecules, some using
ab initio PES. Here, we briefly discuss these developments and their
applications.
Fig. 3 Time-averaged single-mode effective potentials (blue circles), the diagonal

anharmonic potential given by Eq. (22) (orange squares), and the harmonic potential
(yellow diamonds) for different normal modes of alanine I. Panel (A) presents results
for the symmetric NH2 stretch, panel (B) presents results for the asymmetric NH2 stretch,
and panel (C) presents results for the OH Stretch. Results were obtained using MP2/
aug-cc-pVDZ level of theory.
3.2.1 Quantum Wavepacket AIMD

The quantum wavepacket AIMD (QWAIMD) method was developed by
Iyengar and coworkers.71–73 In essence, it combines the mixed quantum–
classical method described in Section 2 with ab initio PES. Using this
method, the system is divided into three subsystems: the electrons, which
are treated quantum mechanically, most of the nuclei, which are treated
classically; and selected nuclei, which are treated quantum mechanically.

Similarly to the AICSP method, the BO approximation is used and the elec-
trons are treated through the time-independent Schr€ odinger equation. The
equations of motion for the classical nuclei are solved using the atom-
centered density matrix propagation scheme,74–76 which is computationally
efficient, with the aim of describing large systems. Initial applications treated
the classical nuclei using pure quantum chemical potentials such as B3LYP
for relatively small systems. However, later the method was also combined
with a QM/MM description77 of the classical nuclei in order to perform effi-
cient simulations for large systems. Applications of the method were per-
formed for calculating vibrational spectra in small molecules, such as
ClHCl,72 for hydrogen tunneling in a model for an active site of a protein78
and for shared proton dynamics in a complex between phenol and
trimethylamine.71 More recently, QWAIMD was introduced for periodic
systems,79 with application to a two-dimensional model of concentrated
solid hydrochloric acid. While the method seems very promising for
describing large systems, the general difficulties which are common to all
mixed quantum–classical methods may also be relevant to QWAIMD.
Mainly, since only few atoms in the system are treated quantum mechani-
cally, errors may arise when calculating observables, such absorption cross
sections.63,64
Another development direction in the field of mixed quantum–classical
dynamics is the treatment of open quantum systems.80–82 For example, sev-
eral methods were developed based on the quantum–classical Louiville
equation. For a detailed description of the different methods, see the reviews
by Karpal.32,80 Very recently, Kananenka et al. suggested an efficient
method to describe the long-time dynamics of an open quantum system
using the transfer tensor method. QWAIMD was also recently extended
to treat coupled electron-nuclear dynamics of open systems through a par-
titioning of the total Schr€odinger equation into local domains that interact
with each other through absorbing or emitting potentials.83 This method
was then applied to a model of a molecular wire.
4. SUMMARY AND FUTURE DIRECTIONS

Mean-field methods for quantum molecular dynamics have gained
renewed interest in recent years. One important reason for this is that they
allow a computationally efficient framework to describe the interactions
between the nuclei using ab initio potentials directly. Such potentials are
required in order to accurately describe complex systems, for which empir-
ical potentials are not generally available. Several developments in this spirit,
presented in this paper are (1) The AICSP method, which combines the
CSP method for quantum nuclear dynamics with ab initio PES, and (2)
the QWAIMD method, which is a mixed quantum–classical method where
the trajectories for the classical subsystem are performed using potentials
obtained directly from quantum chemical methods. AICSP was applied
recently to the calculation of stationary anharmonic vibrational frequencies
of amino acids and the G–C pair of nucleobases,51 for which ab initio poten-
tials are essential in order to obtain quantitative agreement with experimen-
tal measurements. QWAIMD was applied, for example, to study hydrogen
tunneling in a model for an enzyme active site.78
These recent developments open a way to several exciting future
directions and to a wide range of applications, such as ultrafast time-
domain spectroscopy measurements, the calculation of vibrational lin-
eshapes, electron photodetachment in large systems, and transition state
dynamics. For example, while several applications, mainly by McCoy
and coworkers,57–59 were performed using mixed quantum–semiclassical
TDSCF to study transition state dynamics of ClHCl in Ar clusters, the
application of mean-field methods for transition state dynamics has so
far been limited mainly to systems for which an accurate empirical poten-
tial in the vicinity of the transition state was available. The direct use of
ab initio potentials can significantly increase the range of applicability of
mean-field methods to study transition state dynamics in different systems.
This may also allow quantitative comparison of mean-field methods with
experimental measurements, which has not been done extensively in ear-
lier studies. At the same time, a major challenge that remains is finding
extensions of the method. Currently, methods such as AICSP are very
much limited to cases for which normal coordinates can be used and
the dynamics are dominated by a single minimum on the potential. Appli-
cation of these methods for cases in which more than one local minimum is
involved is problematic, since the separable ansatz will no longer be ade-
quate. It is an interesting challenge to develop methods that use other coor-
dinates which maintain the separability of the nuclear wavefunction for
longer timescales and more complicated dynamics. In these cases, tech-
niques used for stationary vibrational spectroscopy, such as the use of inter-
nal coordinates53 or local coordinates52 may prove very useful.
ACKNOWLEDGMENTS
We dedicate this paper to Mark Ratner. R.B.G. gratefully acknowledges very stimulating
collaborations with Mark, including in the field of this paper, and many illuminating
discussions and exchanges of ideas. B.H. is supported through an Adams Fellowship of the
Israel Academy of Sciences and Humanities. We thank Dr. Laura McCaslin for her
comments on the manuscript.
REFERENCES
[1] Rapaport, D. C. The Art of Molecular Dynamics Simulation, Cambridge University Press:
Cambridge, 2004.
[2] Marx, D.; Hutter, J. Ab Initio Molecular Dynamics: Theory and Implementation. In
Grotendorst, J., Ed.; Modern Methods and Algorithms of Quantum Chemistry, 1,
2000; pp 301–449.
[3] Car, R.; Parrinello, M. Unified Approach for Molecular Dynamics and Density-
Functional Theory. Phys. Rev. Lett. 1985, 55(22), 2471–2474.
[4] Hutter, J. Car-Parrinello Molecular Dynamics. Wiley Interdiscip. Rev. Comput. Mol. Sci.
2012, 2(4), 604–612.
[5] K€uhne, T. D. Second Generation Car-Parrinello Molecular Dynamics. Wiley Interdiscip.
Rev. Comput. Mol. Sci. 2014, 4(4), 391–406.
[6] Karplus, M.; McCammon, J. A. Molecular Dynamics Simulations of Biomolecules.
Nat. Struct. Biol. 2002, 9(9), 646–652.
[7] Ufimtsev, I. S.; Luehr, N.; Martinez, T. J. Charge Transfer and Polarization in Solvated
Proteins from Ab Initio Molecular Dynamics. J. Phys. Chem. Lett. 2011, 2(14),
1789–1793.
[8] Hassanali, A. A.; Cuny, J.; Verdolino, V.; Parrinello, M. Aqueous Solutions: State of the
Art in Ab Initio Molecular Dynamics. Philos. Trans. A. Math. Phys. Eng. Sci. 2014,
372(2011), 20120482.
[9] Lin, I.-C.; Seitsonen, A. P.; Tavernelli, I.; Rothlisberger, U. Structure and Dynamics of
Liquid Water from Ab Initio Molecular Dynamics—Comparison of BLYP, PBE, and
revPBE Density Functionals with and without van Der Waals Corrections. J. Chem.
Theory Comput. 2012, 8(10), 3902–3910.
[10] Matsumoto, M.; Saito, S.; Ohmine, I. Molecular Dynamics Simulation of the Ice
Nucleation and Growth Process Leading to Water Freezing. Nature 2002,
416(6879), 409–413.
[11] Hafner, J. Ab-Initio Simulations of Materials Using VASP: Density-Functional Theory
and beyond. J. Comput. Chem. 2008, 29(13), 2044–2078.
[12] Gerber, R. B.; Shemesh, D.; Varner, M. E. E.; Kalinowski, J.; Hirshberg, B. Ab Initio
and Semi-Empirical Molecular Dynamics Simulations of Chemical Reactions in Iso-
lated Molecules and in Clusters. Phys. Chem. Chem. Phys. 2014, 16, 9760–9775.
[13] Ceriotti, M.; Fang, W.; Kusalik, P. G.; McKenzie, R. H.; Michaelides, A.;
Morales, M. A.; Markland, T. E. Nuclear Quantum Effects in Water and Aqueous Sys-
tems: Experiment, Theory, and Current Challenges. Chem. Rev. 2016, 116(13),
7529–7550.
[14] Paesani, F.; Viel, A.; Gianturco, F. A.; Whaley, K. B. Transition from Molecular Com-
plex to Quantum Solvation in 4He_{N}OCS. Phys. Rev. Lett. 2003, 90(7), 73401.
[15] Grebenev, S.; Sartakov, B. G.; Toennies, J. P.; Vilesov, A. F. Spectroscopic Investiga-
tion of OCS (P-H2)n (n ¼ 1–16) Complexes inside Helium Droplets: Evidence for
Superfluid Behavior. J. Chem. Phys. 2010, 132(6), 064501.
[16] Gao, J.; Truhlar, D. G. Qunatum Mechanical Methods for Enzyme Kinetics. Annu.
Rev. Phys. Chem. 2002, 53(1), 467–505.
[17] Vardi-Kilshtain, A.; Nitoker, N.; Major, D. T. Nuclear Quantum Effects and Kinetic
Isotope Effects in Enzyme Reactions. Arch. Biochem. Biophys. 2015, 582, 18–27.
[18] Wang, L.; Isborn, C. M.; Markland, T. E. Chapter Fifteen—Simulating Nuclear
and Electronic Quantum Effects in Enzymes. In: Gregory A., Voth Ed.; Methods in
Enzymology, Academic Press, 2016, Vol. 577, pp 389–418, ISSN 0076-6879, ISBN
9780128053478, http://dx.doi.org/10.1016/bs.mie.2016.05.047.
[19] Kosloff, R. Time-Dependent Quantum-Mechanical Methods for Molecular Dynam-
ics. J. Phys. Chem. 1988, 92(8), 2087–2100.
[20] Zhang, D. H.; Guo, H. Recent Advances in Quantum Dynamics of Bimolecular Reac-
tions. Annu. Rev. Phys. Chem. 2016, 67(1), 135–158.
[21] Otto, R.; Ma, J.; Ray, A. W.; Daluz, J. S.; Li, J.; Guo, H.; Continetti, R. E. Imaging
Dynamics on the F + H2O ! HF + OH Potential Energy Surfaces from Wells to Bar-
riers. Science 2014, 343, 396–399.
[22] Miller, W. H. The Semiclassical Initial Value Representation: A Potentially Practical
Way for Adding Quantum Effects to Classical Molecular Dynamics Simulations.
J. Phys. Chem. A 2001, 105(13), 2942–2955.
[23] Heller, E. J. Time-Dependent Approach to Semiclassical Dynamics. J. Chem. Phys.
1975, 62(4), 1544.
[24] Heller, E. J. Frozen Gaussians:A Very Simple Semiclassical Approximation. J. Chem.
Phys. 1981, 75(6), 2923.
[25] Cao, J.; Voth, G. A. The Formulation of Quantum Statistical Mechanics Based on the
Feynman Path Centroid Density. I. Equilibrium Properties. J. Chem. Phys. 1994,
100(7), 5093–5105.
[26] Cao, J.; Voth, G. A. The Formulation of Quantum Statistical Mechanics Path
Centroid Density. II. Dynamical Properties. J. Chem. Phys. 1993, 1994(100),
5106–5117.
[27] Habershon, S.; Manolopoulos, D. E.; Markland, T. E.; Miller, T. F. Ring-Polymer
Molecular Dynamics: Quantum Effects in Chemical Dynamics from Classical Trajec-
tories in an Extended Phase Space. Annu. Rev. Phys. Chem. 2013, 64, 387–413.
[28] Gerber, R. B.; Buch, V.; Ranter, M. A. Time-Dependent Self-Consistent Field
Approximation for Intramolecular Energy Transfer. I. Formulation and Application
to Dissociation of van Der Waals Molecules. J. Chem. Phys. 1982, 1982(77), 3022.
[29] Messina, M.; Coalson, R. D. Time-Dependent Hartree Wave Packet Dynamical Tech-
niques for Computation of Electronically Excited State Optical Spectra of Many-Body
Quantum Systems. J. Chem. Phys. 1989, 90(8), 4015.
[30] Makri, N.; Miller, W. H. Time-Dependent Self-Consistent Field (TDSCF) Approxi-
mation for a Reaction Coordinate Coupled to a Harmonic Bath: Single and Multiple
Configuration Treatments. J. Chem. Phys. 1987, 87(1987), 5781.
[31] Jungwirth, P.; Gerber, R. B. Quantum Dynamics of Large Polyatomic Systems Using a
Classical Based Separable Potential Method. J. Chem. Phys. 1995, 102, 6046.
[32] Tully, J. C. Mixed Quantum-Classical Dynamics. Faraday Discuss. 1998, 110, 407–419.
[33] Gerber, R. B.; Ratner, M. A. A Semiclassical Self-Consistent Field (SC SCF) Approx-
imation for Eigenvalues of Coupled-Vibration Systems. Chem. Phys. Lett. 1979, 68(1),
195–198.
[34] Bowman, J. M. Self-Consistent Field Energies and Wavefunctions for Coupled Oscil-
lators. J. Chem. Phys. 1978, 68(2), 608.
[35] Chaban, G. M.; Jung, J. O.; Gerber, R. B. Ab Initio Calculation of Anharmonic Vibra-
tional States of Polyatomic Systems: Electronic Structure Combined with Vibrational
Self-Consistent Field. J. Chem. Phys. 1999, 111(5), 1823.
[36] Roy, T. K.; Gerber, R. B. Vibrational Self-Consistent Field Calculations for Spectros-
copy of Biological Molecules: New Algorithmic Developments and Applications. Phys.
Chem. Chem. Phys. 2013, 15(24), 9468–9492.
Another random document with
no related content on Scribd:
have been developed in membrane overlying one of the dentigerous
bones mentioned, without having become anchylosed to the bone.
When the tooth is fixed to the bone the attachment has generally
been effected by the ossification of the bone of the tooth, but in
some fishes a process of the bone projects into the cavity of the
tooth; in others the teeth are implanted in alveoli. In these, again,
frequently a process of bone rises from the bottom, on which the
tooth rests.
Many fishes, especially predatory fishes with long, lancet-shaped
teeth, have all or some of the teeth capable of being bent inwards
towards the mouth. Such “hinged” teeth resume at once the upright
position when pressure is removed from them. They are, however,
depressible in one direction only, thus offering no obstacle to the
ingress, but opposing the egress of prey. Mr. C. S. Tomes has shown
that the means by which this mechanism is worked are different in
different fishes; for whilst, in the Pediculati and Gadoids (Hake) the
elasticity resides solely in the tissue of the hinge (the tooth being as
resilient as ever after everything else is severed), in the Pike the
hinge is not in the least endowed with elasticity, but the bundles of
fibres proceeding from the interior of the dentine cap are exceedingly
elastic.
As regards texture the teeth of fishes show considerable
variation. The conical teeth of the Cyclostomes and the setiform
teeth of many Teleosteans consist of a horny albuminous substance.
The principal substance of the teeth of other fishes consists of
dentine, with numerous dividing and anastomosing tubercles,
sometimes covered by a stratum of unvascular dentine. An enamel-
like substance has been observed on the crown of the teeth of
Sargus and Balistes, and an ossification of the capsule of their
matrix covers the enamel with a thin coating of cement. The teeth
either possess a cavity in which the matrix is received, or, more
frequently, they are solid, in which case vascular canals of the
underlying bone are continued into the substance of the tooth. In the
teeth of some fishes numerous sets of canals and tubes are so
arranged that they do not anastomose with one another, each set
being surrounded by a layer of dentine and cement. These
apparently simple teeth are evidently composed of numerous small
teeth, and called compound teeth.
The teeth may be, and generally are, very different as regards
size or form in the different parts of the mouth; they may be also
different according to the age or sex of the fish (Raja). The teeth may
be few in number and isolated, or placed in a single, double, or triple
series, distant from one another or closely set; they may form narrow
or broad bands, or patches of various forms. As regards form, they
may be cylindrical or conical, pointed, straight, or curved, with or
without an angular bent near their base; some are compressed
laterally or from the front backwards; the latter may be triangular in
shape, or truncated at the top like incisors of mammals; they may
have one apex (cusp) only, or be bi- or tri-lobate (bi- or tri-cuspid); or
have the margins denticulated or serrated. Compressed teeth may
be confluent, and form a cutting edge in both jaws, which assume
the shape of a parrot’s beak (Fig. 53). In some the apex is hooked or
provided with barbs. Again, some teeth are broad, with flat or convex
surface, like molar teeth. With regard to size, the finest teeth are like
fine flexible bristles, ciliiform or setiform; or, if very short and
anchylosed to the bone, they appear only as inconspicuous
asperities of the bone. Very fine conical teeth arranged in a band are
termed villiform teeth; when they are coarser, or mixed with coarser
teeth, they are card-like (dents en rape or en cardes) (Fig. 54);
molar-like teeth of very small size are termed granular.
Fig. 53.—Jaws of Calliodon.
In all fishes the teeth are constantly shed or renewed during the
whole course of their life. In fishes which have compound teeth, as
the Dipnoi, Chimæroids, Scari,[14] Gymnodonts, as well as in those
which have apparently permanent teeth, as in the saw of Pristis, the
detrition of the surface is made up by a constant growth of the tooth
from its base. When the teeth are implanted in alveoli, they are
generally succeeded by others in the vertical direction, but in others
they succeed one another, side by side. In the majority of fishes the
new tooth is not developed (as in reptiles and mammals) in a
diverticulum of the sack of its predecessor, but like this from the free
surface of the buccal membrane. Generally there are more than one
tooth growing, which are in various stages of development, and
destined to replace the one in function. This is very conspicuous in
Sharks, in which the whole phalanx of their numerous teeth is ever
marching slowly forwards (or in some backwards), in rotatory
progress, over the alveolar border of the jaw, the teeth being
successively cast off after having reached the outer margin, and
fulfilled for a longer and shorter period their special function.
[The richest materials for our knowledge of the teeth of fishes are
contained in Owen’s “Odontography.” Lond. 1840. 8vo.]
Fig. 54.—Cardlike teeth of
Plectropoma dentex, with
canines.
The intestinal tract is divided into four portions: œsophagus,
stomach, small and large intestine; two or more of these divisions
may coalesce in fishes and become indistinguishable. But it is
characteristic of the class that the urinary apertures are constantly
situated behind the termination of the intestinal tract.
In Branchiostoma the whole intestinal tract is straight, and coated
with a ciliated mucous membrane. The wide pharynx passes into a
narrow œsophagus, this into a gastric cavity, the remainder being
again narrower and terminating in the anal aperture, which lies
somewhat to the left of the median line. The liver is represented by a
green coloured cœcal diverticulum of the stomachic dilatation. A
mesenterium is absent.
In the Cyclostomi the intestinal tract is likewise straight, and
without clearly defined divisions; however, in Petromyzon the
œsophagus shows numerous longitudinal folds, and the intestine
proper is provided with a single longitudinal fold. A mesentery, which
is present in the Myxinoids, is represented by a short median fold
only, by means of which the hindmost part of the intestine is fixed.
The Palæichthyes show differences in the structure of their
intestinal tract as considerable as are found among the Teleostei, but
they have that in common that the absorbent surface of their
intestine is enlarged by the development of a spiral valve, evidence
of the presence of which in extinct Palæichthyes is still preserved in
the fossilised fæces or coproliths, so abundant in some of the older
strata.
In Chondropterygians (Fig. 55) the stomach is divided into a
cardiac and pyloric portion, the former frequently terminating in a
blind sac, and the latter varying in length. The pyloric portion is bent
at its origin and end, and separated from the short duodenum (called
Bursa entiana in these fishes) by a valve; the ductus hepaticus and
pancreaticus enter the duodenum. This is succeeded by the straight
intestine provided with the spiral valve, the coils of which may be
either longitudinal and wound vertically about the axis of the
intestine, as in Carcharias, Galeocerdo, Thalassorhinus, and
Zygœna, or they may be transverse to that axis, as in the other
genera. The number of gyrations in the latter case varies: there may
be as many as forty. The short rectum passes into a cloaca, which
contains also the orifices of the urogenital ducts. Only the
commencement and end of the intestinal tract are fixed by
mesenterial folds.
In the Holocephali and Dipnoi, the intestinal tract is short,
straight, and wide, without stomachic dilatation, a pyloric valve, close
to which the ductus choledochus enters, indicating the boundary of
the intestine proper (Fig. 57, p). The spiral valve is perfect, and
makes from three (Chimæra) to nine (Ceratodus) gyrations. A cloaca
is present, as in Chondropterygians. A mesentery fixing the dorsal
side of the intestine is absent.
Fig. 55.—Siphonal stomach and spiral valve of
Basking-Shark (Selache). (After Home and Owen.)
a, Œsophagus; b, Cardiac portion of stomach; c, pyloric
portion; d, pouch intermediate between stomach
and duodenum, with circular valves at both ends; e,
Duodenum; f, Valve of intestine; g, Ductus
hepaticus; h, Spleen.
The other Ganoids resemble again more the Chondropterygians
in the structure of their intestinal tract. The stomach has always a
distinct pyloric portion, and has a still more complicated structure in
Acipenser. The duodenal portion receives the contents of
Appendices pyloricæ, which are confluent into a gland-like mass in
Acipenser, but separate in Polyodon, and numerous and short in
Lepidosteus, whilst Polypterus possesses one such appendage only.
A spiral valve is developed in the Sturgeons and Polypterus, but in
Amia, in which the intestine performs several convolutions, the four
gyrations of the valve are situated far back towards the end of the
intestine. In Lepidosteus the valve is rudimentary, and indicated only
by three raised lines crossing the terminal portion of the intestine. In
all these Ganoids the rectum has a separate opening, without
cloaca.
The structure of the intestinal tract of Teleosteous fishes is
subject to so numerous modifications that we should go beyond the
limits of the present work if we would attempt to enter into details.
Great differences in this respect may be found even in groups of the
same natural families. Frequently the intestinal tract remains of
nearly the same width throughout its course, and only the entrance
of the various ducts serves as a guide for the distinction of its
divisions. An intestine of such uniform width may be straight and
short, as in Scombresocidæ, Symbranchidæ, or it may be more or
less convoluted and long, as in many Cyprinidæ, Doradina, etc. On
the whole, carnivorous fishes have a much shorter and simpler
intestinal tract than herbivorous.
In the majority of Teleosteans, however, œsophagus, stomach,
duodenum, small intestine and rectum, can be more or less
distinctly, even externally distinguished.
There are two predominant forms of the stomach, intermediate
forms being, however, numerous. In the first, the siphonal, it
presents the form of a bent tube or canal, one-half of the horse-shoe
being the cardiac, the other the pyloric portion. In the second, the
cæcal, the cardiac division is prolonged into a long descending blind
sac, the cardiac and pyloric openings of the stomach lying close
together (Clupea, Scomber, Thynnus, etc.)
Fig. 56.—Siphonal Stomach and Pyloric
Appendages of a Female Salmon, 3⅓ feet
long. a a a, Pyloric appendages; ch, ductus
choledochus; oe, œsophagus; st, lower end of
stomach; p, pyloric region; i, ascending; and í’,
descending portion of intestine.
The duodenum receives always the hepatic and pancreatic
secretions, and, besides, those of the appendices pyloricæ, which, in
varying numbers (from 1 to 200), are of very common occurrence in
Teleosteans (Fig. 56). They vary also in length and width, and whilst
the narrowest serve only as secretory organs, the widest are
frequently found filled with the same contents as the intestine. When
few in number, each opens by a separate duct into the duodenum;
when their number is greater two or more coalesce into a common
duct; in the latter case the appendages cease to be free, and are
connected with one another by a more or less firm tissue.
Cœcal appendages at the end of the intestinal tract are of
exceedingly rare occurrence in fishes (Box). There is no cloaca in
Teleosteans.
In the majority of Teleosteous fishes the vent is situated on the
boundary between trunk and tail, behind the ventral fins. In a few it
lies farther backwards, not far from the caudal fin; more frequently it
is advanced forwards, under the middle of the abdomen or to the
scapular arch. In two fishes, Aphredoderus and Amblyopsis, it lies
before the pectoral fins.
A peritoneum envelops all the divisions of the intestinal tract
within the abdominal cavity. A broad, well-developed omentum has
hitherto been found in Gobiesox cephalus only.
Liver.—The existence of a liver in Branchiostoma as a long
diverticulum of the intestine has been mentioned above. In the
Myxinoids the liver is divided into two glandular bodies, an anterior
rounded smaller one, and a posterior larger one of an elongate
shape. The gall-bladder lies between both, and receives a cystic
duct from each of them. In the other fishes the proportionally large
liver is a single large gland, from which only now and then small
portions are found to be detached. It is either simple, or with a right
and left lobe, or with a third lobe in the middle; each lobe may have
incisions or subdivisions, which, however, are very inconstant. The
liver of fishes is distinguished by the great quantity of fluid fat (oil)
which it contains. The gall-bladder is but rarely absent, and attached
to the right lobe, or towards the centre; however, in some fishes it is
detached from the liver and connected with it by the cystic duct only.
The bile may be conveyed by one or more hepatic ducts into a
common duct which is continued towards the gall-bladder as ductus
cysticus, and towards the duodenum as ductus choledochus; or
some of the hepatic ducts enter directly the gall-bladder, or directly
the duodenum, without communicating with the common duct.
Individual variations in this respect are of common occurrence.
A pancreas has been found hitherto in all Chondropterygians,
Acipenser, and many Teleosteans. In the first it is a glandular mass
of considerable size behind the stomach, close to the spleen; its duct
leads into the duodenum. In the Sturgeons the pancreas is attached
to the duodenum, and opens close to the ductus choledochus. In
Silurus glanis it is very large, and the ductus choledochus passes
through its substance; it is smaller in Belone and Pleuronectes, and
situated in the mesentery; its duct accompanies the terminal portion
of the ductus choledochus. In the Salmon, which possesses a large
lobed pancreas, the duct is so intimately connected with the ductus
choledochus that both appear externally as a single duct only.
The spleen, which is substantially a lymphatic gland, may be
mentioned here, as it is constantly situated in the immediate vicinity
of the stomach, generally near its cardiac portion. With the exception
of Branchiostoma, it is found in all fishes, and appears as a rounded
or oblong organ of dark-red colour. In the Sharks frequently one or
more smaller pieces are detached from the principal body. In the
Dipnoi a thin layer of a very soft substance of brownish-black colour
below the mucous membrane of the stomach and upper part of the
intestine has been regarded as the homologue of the spleen (Fig.
57, m). In most Teleostei the spleen is undivided, and appended by
its vessels and a fold of the peritoneum to the pyloric bend of the
stomach or the beginning of the intestine.
Fig. 57.—Upper part of Intestine of Ceratodus.
The anterior wall of the intestine is opened, the liver
(c) and gall-bladder (e) being drawn forward. A slit is
made at n, through which part of the next
compartment of the spirally wound intestine may be
seen.
é, Mouth of ductus choledochus; f, stomach; i,
adipose agglomeration; l, first compartment of
intestinal spire; m, spleen; oe, lower part of
œsophagus, opened; p, double pyloric fold; q q,
glandular patches.
CHAPTER IX.
ORGANS OF RESPIRATION.
Fishes breathe the air dissolved in water by means of gills or

branchiæ. The oxygen consumed by them is not that which forms
the chemical constituent of the water, but that contained in the air
which is dissolved in water. Hence fishes transferred into water from
which the air has been driven out by a high temperature, or in which
the air absorbed by them is not replaced, are speedily suffocated.
The absorption of oxygen by fishes is comparatively small, and it has
been calculated that a man consumes 50,000 times more than is
required by a Tench. However, some fishes evidently require a much
larger supply of oxygen than others: Eels and Carps, and other
fishes of similar low vitality, can survive the removal out of their
elements for days, the small quantity of moisture retained in their gill-
cavity being sufficient to sustain life, whilst other fishes, especially
such as have very wide gill-openings, are immediately suffocated
after being taken out of the water. In some fishes noted for their
muscular activity, like the Scombridæ, the respiratory process is so
energetic as to raise the temperature of their blood far beyond that of
the medium in which they live. A few fishes, especially such as are
periodically compelled to live in water thickened into mud by
desiccation and vitiated by decomposing substances, breathe
atmospheric air, and have generally special contrivances for this
purpose. These are so much habituated to breathing air that many of
them, even when brought into pure water of normal condition, are
obliged to rise to the surface at frequent intervals to take in a
quantity of air, and if they be kept beneath the surface by means of a
gauze net, they perish from suffocation. The special contrivances
consist of additional respiratory organs, lodged in cavities either
adjoining the gill-cavity or communicating with the ventral side of the
œsophagus, or of the air-bladder which enters upon respiratory
functions (Dipnoi, Lepidosteus, Amia).
The water used by fishes for respiration is received by the mouth,
and by an action similar to that of swallowing driven to the gills, and
expelled by the gill-openings, of which there may be one or several
on each side behind the head; rarely one only in the median line of
the ventral surface.
Fig. 58.—Fore-part of the body of an

embryon of Carcharias, showing the
branchial filaments (natural size).
The gills or branchiæ consist essentially of folds of the mucous
membrane of the gill-cavity (laminæ branchiales), in which the
capillary vessels are distributed. In all fishes the gills are lodged in a
cavity, but during the embryonic stage the Chondropterygians have
the gill-laminæ prolonged into long filaments projecting beyond the
gill-cavity (Fig. 58), and in a few young Ganoids external gills are
superadded to the internal.
In Branchiostoma the dilated pharynx is perforated by numerous
clefts, supported by cartilaginous rods (Fig. 29, h). The water passes
between these clefts into the peritoneal cavity, and makes its exit by
the porus abdominalis situated considerably in advance of the vent.
The water is propelled by cilia.
In the Cyclostomes the gills of each side are lodged in a series of
six or more antero-posteriorly compressed sacs, separated from
each other by intervening septa. Each sac communicates by an
inner duct with the œsophagus, the water being expelled by an outer
duct. In Bdellostoma each outer duct has a separate opening, but in
Myxine all the outer ducts pass outwards by one common gill-
opening on each side. In the Lampreys the ducts are short, the outer
ones having separate openings (Fig. 2, p. 39). The inner ducts lead
into a single diverticulum or bronchus, blind behind, situated below
the œsophagus, and communicating in front with the pharynx, where
it is provided with two valves by which the regurgitation of the water
into the buccal cavity is prevented.
The same type of branchial organs persists in
Chondropterygians, which possess five, rarely six or seven, flattened
pouches with transversely plaited walls. The septa between them are
supported by cartilaginous filaments rising from the hyoidean and
branchial arches. Each pouch opens by a cleft outwards, and by an
aperture into the pharynx, without intervening ducts. The anterior
wall of the first pouch is supported by the hyoidean arch. Between
the posterior wall of the first and the anterior of the second sac, and
between the adjacent walls of the succeeding, a branchial arch with
its two series of radiating cartilaginous filaments is interposed.
Consequently the first and last pouch have one set of gill-laminæ
only, viz. the first on its posterior and the last on its anterior wall. The
so-called spiracles on the upper surface of the head of
Chondropterygians are to be referred to in connection with the
respiratory organs. They are the external openings of a canal leading
on each side into the pharynx, and situated generally close to and
behind the orbit. They frequently possess valves or an irregularly
indented margin, and are found in all species during the embryonic
stage, but remaining persistent in a part only. The spiracles are the
remains of the first visceral cleft of the embryo, and in the fœtal state
long branchial filaments have been observed to protrude, as from the
other branchial clefts.
The Holocephali and Ganoidei show numerous deviations from
the Chondropterygian type, all leading in the direction towards the
Teleosteans. As a whole they take an intermediate position between
the preceding types and the Teleosteans, but they show a great
variation among themselves, and have in common only the imperfect
separation of the branchial sacs and the presence of a single outer
branchial aperture.
In Chimæra the septum separating the branchial sacs is
confluent with the wall of the gill-cavity in a part of its extent only, and
still more imperfect is the separation of those branchial divisions in
Ceratodus (Fig. 60). The other Ganoids show no such division
whatever. In Chimæra the first gill is incomplete (uniserial), and
belongs to the hyoid; then follow three complete gills; the last,
belonging to the fourth branchial arch, being again incomplete.
Acipenser, Scaphirhynchus, Lepidosiren, Protopterus, and
Lepidosteus, possess likewise an anterior incomplete gill (opercular
gill), followed by four complete gills in the Sturgeons and
Lepidosteus, whilst in Lepidosiren and Protopterus a part of the
branchial arches is gill-less. In Polyodon, Ceratodus, and Polypterus,
an opercular gill is absent, the two former having four complete gills,
the latter three and a half only. Spiracles are still in some Ganoids
present, viz. in the Sturgeons and Polypterus. In all the Ganoids an
osseous gill-cover is now developed.
In the Teleostei the gills with their supporting branchial arches lie
in one undivided cavity; more or less wide clefts between the arches
lead from the pharynx to the gills, and a more or less wide opening
gives exit to the water after it has washed the gills. The
interbranchial clefts have sometimes nearly the same extent as the
branchial arches; sometimes they are reduced to small openings, the
integuments stretching from one arch to the other. Sometimes there
is no cleft behind the fourth arch, in which case this arch has only an
uniserial gill developed. The gill-opening likewise varies much in its
extent, and when reduced to a foramen may be situated at any place
of the posterior boundary of the head. In the Symbranchidæ the gill-
openings coalesce into a single narrow slit in the median line of the
isthmus. In the majority of Teleosteans the integument of the
concave side of the branchial arches develops a series of horny
protuberances of various form, the so-called gill-rakers. They are
destined to catch any solid corpuscles or substances which would be
carried into the gill-cavity with the water. In some fishes they are
setiform, and form a complete sieve, whilst in others they are merely
rough tubercles, the action of which must be very incomplete if they
have any function at all.
Most Teleosteans possess four complete gills, but frequently the
fourth arch is provided with an uniserial gill only, as mentioned
above, or even entirely gill-less. The most imperfect gills are found in
Malthe, which has two and a half gills only, and in Amphipnous
cuchia, in which one small gill is fixed to the second arch.
The gills of the Teleosteans as well as of the Ganoids are
supported by a series of solid cartilaginous or horny pointed rods,
arranged along the convex edges of the branchial arches. Arches
bearing a complete gill have two series of those rods, one along
each edge; those with uniserial gills bear one row of rods only. The
rods are not part of the arch, but fixed in its integument, the several
rods of one row corresponding to those of the other, forming pairs
(feuillet, Cuvier) (Fig. 59). Each rod is covered by a loose mucous
membrane passing from one rod to its fellow opposite, which again
is finely transversely plaited, the general surface being greatly
increased by these plaits. In most Teleostei the branchial lamellæ
are compressed, and taper towards their free end, but in the
Lophobranchs their base is attenuated and the end enlarged. The
mucous membrane contains the finest terminations of the vessels,
which, being very superficial, impart the blood-red colour to living
gills. The Arteria branchialis, the course of which lies in the open
canal in the convexity of the branchial arch, emits a branch (a) for
every pair of lamellæ which ascends (b) along the inner edge of the
lamella, and supplies every one of the transverse plaits with a
branchlet. The latter break up into a fine net of capillaries, from
which the oxygenised blood is collected into venous branchlets,
returning by the venous branch (d), which occupies the outer edge of
the lamella.
Fig. 59.—A pair of branchial
lamellæ (magnified) of the Perch.
a, Branch of Arteria branchialis;
b, Ascending branch of the
same; c, Branch of Vena
branchialis; d, Descending
branch of the same; e,
Transverse section through
the branchial arch.
The so-called Pseudobranchiæ (Fig. 60) are the remains of an
anterior gill which had respiratory functions during the embryonic life
of the individuals. By a change in the circulatory system these
organs have lost those functions, and appear in the adult fish as retia
mirabilia, as they receive oxygenised blood, which, after having
passed through their capillary system, is carried to other parts of the
head. In Palæichthyes the pseudobranchia is a rete mirabile
caroticum for the brain and eye; in Teleosteans a rete mirabile
ophthalmicum only. Pseudobranchiæ are as frequently absent as
present in Chondropterygians as well as Teleosteans. As to the
Ganoids, they occur in Ceratodus, Acipenser, Polyodon, and
Lepidosteus, and are absent in Lepidosiren, Protopterus,
Scaphirhynchus, Polypterus, and Amia.
In Chondropterygians and Sturgeons the pseudobranchiæ are
situated within the spiracles; in those, in which spiracles have
become obliterated, the pseudobranchiæ lie on the suspensorium,
hidden below cellular tissue; but pseudobranchiæ are not
necessarily co-existent with spiracles. In the other Ganoids and
Teleosteans the pseudobranchiæ (Fig. 60, h) are within the gill-
cavity, near the base of the gill-cover; in Ceratodus even rudiments
of the gill-rakers (x’, x”) belonging to this embryonic gill are
preserved, part of them (x”) being attached to the hyoid arch.
Pseudobranchiæ are frequently hidden below the integuments of the
gill-cavity, and have the appearance of a glandular body rather than
of a gill.
[See Müller, “Vergleichende Anatomie des Gefäss-systems der
Myxinoiden;” and “Ueber den Bau und die Grenzen der Ganoiden.”]
Fig. 60.—Gills of Ceratodus.
x, Arcus aortæ; gl, Glossohyal; ch,
Ceratohyal; u, Attachment of the
first gill to the walls of the gill-
cavity; h, Pseudobranchia; x’, x”,
two series of gill-rakers
belonging to the
Pseudobranchia.
Accessory respiratory organs for retaining water or breathing air,
such as are found in the Labyrinthici, Ophiocephalidæ, certain
Siluridæ, and Lutodira, are structures so specialised that they are
better described in the accounts of the Fishes in which they have
been observed.

Textbook Advances in Quantum Chemistry Ratner Volume John R Sabin and Erkki J Brandas Eds Ebook All Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Textbook Advances in Quantum Chemistry Ratner Volume John R Sabin and Erkki J Brandas Eds Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Advances in Quantum Chemistry

Ratner Volume John R. Sabin And Erkki

Advances in the Theory of Quantum Systems in Chemistry

Quantum Systems in Chemistry and Physics Progress in

Advances in Organometallic Chemistry Volume 65 Pedro J.

Advances in Organometallic Chemistry Volume 66 Pedro J.

Advances in Heterocyclic Chemistry Volume 113 1st

Quantum Mechanics for Chemistry Seogjoo J. Jang

Ideas of Quantum Chemistry: Volume 1: From Quantum

Advances in Organometallic Chemistry 1st Edition Pedro

Remigio Cabrera-Trujillo (UNAM, Mexico)

First edition 2017

Copyright © 2017 Elsevier Inc. All rights reserved.

For information on all Academic Press publications

Publisher: Zoe Kruze

Mean-Field Methods for Time-

Advances in Quantum Chemistry, Volume 75 # 2017 Elsevier Inc. 1

Path-integral formulation of quantum mechanics, such as centroid molecu-

2. THEORETICAL FRAMEWORK, ADVANTAGES,

where Q1, …, QN are the mass-weighted normal coordinates, V(Q1, …, QN)

Applying the time-independent variational principle results in the VSCF

ℏ2 @ 2 χ k ðnÞ ðQk Þ ðnÞ

The effective potentials then only involve integrals over 2 degrees of

2.2 Time-Dependent Self-Consistent Field

where H, ^ as in the previous section, is the Hamiltonian of the system and Χ

And applying the Dirac–Frenkel variational principle42,43

@ϕk ðQk , t Þ ℏ2 @ 2 ϕk ðQk , tÞ

where ϕk(Qk, t) differs from χ k(Qk, t) by a physically insignificant purely

Similarly to VSCF, the main computational bottleneck for applying

2.3 Mixed Quantum–Classical Methods

assumed between all modes, including between one classical mode to

Here, ψ(r, t) is the nonseparable wavefunction for the quantum nuclei

2.4 Classical Separable Potentials

where Ntraj is the number of trajectories performed, Wα is the averaging

@ϕk ðQk , tÞ ℏ2 @ 2 ϕk ðQk , tÞ CSP

In essence, this approximation means that each degree of freedom is

The main limitation of the CSP method, compared to TDSCF and

3. RECENT DEVELOPMENTS AND APPLICATIONS

3.1 Ab Initio Classical Separable Potentials

In order to accurately treat large systems, we recently developed51 the

Then, the one-dimensional time-independent Schr€ odinger for each

3.1.2 Dynamical States

3.1.3 Combining CSP With Ab Initio Potentials

Fig. 1 Time-dependent single-mode effective potentials for different vibrational modes

3.2 Mixed Quantum–Classical Methods

Fig. 3 Time-averaged single-mode effective potentials (blue circles), the diagonal

3.2.1 Quantum Wavepacket AIMD

classically; and selected nuclei, which are treated quantum mechanically.

4. SUMMARY AND FUTURE DIRECTIONS

Fishes breathe the air dissolved in water by means of gills or

Fig. 58.—Fore-part of the body of an

You might also like