Mean-Field-Type Games For Engineers 1st Edition Barreiro-Gomez Download PDF
Mean-Field-Type Games For Engineers 1st Edition Barreiro-Gomez Download PDF
Mean-Field-Type Games For Engineers 1st Edition Barreiro-Gomez Download PDF
com
https://ebookmeta.com/product/mean-field-type-
games-for-engineers-1st-edition-barreiro-gomez/
OR CLICK BUTTON
DOWLOAD EBOOK
https://ebookmeta.com/product/mean-field-games-cetraro-
italy-2019-lecture-notes-in-mathematics-yves-achdou/
https://ebookmeta.com/product/fun-and-games-field-day-
understanding-length-chryste-l-berda/
https://ebookmeta.com/product/mean-field-game-and-its-
applications-in-wireless-networks-1st-edition-reginald-a-banez-
lixin-li-chungang-yang-zhu-han/
https://ebookmeta.com/product/mean-field-game-and-its-
applications-in-wireless-networks-1st-edition-reginald-a-banez-
lixin-li-chungang-yang-zhu-han-2/
Antimicrobial Therapies Carlos Barreiro José Luis
Barredo
https://ebookmeta.com/product/antimicrobial-therapies-carlos-
barreiro-jose-luis-barredo/
https://ebookmeta.com/product/point-cloud-technologies-for-
geomorphologists-from-data-acquisition-to-processing-1st-edition-
christopher-gomez/
https://ebookmeta.com/product/my-war-memoirs-1939-1945-210-field-
company-royal-engineers-44th-division-bef-1st-edition-don-mason/
https://ebookmeta.com/product/aerodynamics-for-engineers-6th-
edition-bertin/
https://ebookmeta.com/product/shrinkage-estimation-for-mean-and-
covariance-matrices-hisayuki-tsukuma-tatsuya-kubokawa/
Mean-Field-Type
Games for Engineers
Mean-Field-Type
Games for Engineers
Julian Barreiro-Gomez
Hamidou Tembine
MATLAB® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks
does not warrant the accuracy of the text or exercises in this book. This book’s use or discussion of
MATLAB® software or related products does not constitute endorsement or sponsorship by The
MathWorks of a particular pedagogical approach or particular use of the MATLAB® software.
Reasonable efforts have been made to publish reliable data and information, but the author and pub-
lisher cannot assume responsibility for the validity of all materials or the consequences of their use.
The authors and publishers have attempted to trace the copyright holders of all material reproduced
in this publication and apologize to copyright holders if permission to publish in this form has not
been obtained. If any copyright material has not been acknowledged please write and let us know so
we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.copyright.
com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA
01923, 978-750-8400. For works that are not available on CCC please contact mpkbookspermis-
sions@tandf.co.uk
Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.
DOI: 10.1201/9781003098607
Foreword xxv
Preface xxvii
Acknowledgments xxix
Symbols xxxiii
I Preliminaries 1
1 Introduction 3
1.1 Linear-Quadratic Games . . . . . . . . . . . . . . . . . . . . 4
1.1.1 Structure of the Optimal Strategies and Optimal Costs 4
1.1.2 Solvability of the Linear-Quadratic Gaussian Games . 5
1.1.3 Beyond Brownian Motion . . . . . . . . . . . . . . . . 5
1.2 Linear-Quadratic Gaussian Mean-Field-Type Game . . . . . 6
1.2.1 Variance-Awareness and Higher Order Mean-Field
Terms . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 The Role of the Risk in Engineering Applications . . . 8
1.2.3 Uncertainties in Engineering Applications . . . . . . . 10
1.2.4 Network of Networks/System of Systems . . . . . . . . 14
1.2.5 Optimality Systems . . . . . . . . . . . . . . . . . . . 16
1.3 Game Theoretical Solution Concepts . . . . . . . . . . . . . 19
1.3.1 Non-cooperative Game Problem . . . . . . . . . . . . 20
1.3.2 Fully-Cooperative Game Problem . . . . . . . . . . . . 21
1.3.3 Adversarial Game Problem . . . . . . . . . . . . . . . 21
1.3.4 Berge Game Problem . . . . . . . . . . . . . . . . . . 22
1.3.5 Stackelberg Game Problem . . . . . . . . . . . . . . . 23
1.3.6 Co-opetitive Game Problem . . . . . . . . . . . . . . . 24
1.3.7 Partial-Altruism and Self-Abnegation Game Problem . 25
vii
viii Contents
3 Mean-Field Games 73
3.1 A Continuous-Time Deterministic Mean-Field Game . . . . . 75
3.2 A Continuous-Time Stochastic Mean-Field Game . . . . . . 78
3.3 A Discrete-Time Deterministic Mean-Field Game . . . . . . 86
3.4 A Discrete-Time Stochastic Mean-Field Game . . . . . . . . 91
3.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
16 Applications 405
16.1 Water Distribution Systems . . . . . . . . . . . . . . . . . . 405
16.1.1 Five-Tank Water System . . . . . . . . . . . . . . . . 405
16.1.2 Barcelona Drinking Water Distribution Network . . . 410
16.2 Micro-grid Energy Storage . . . . . . . . . . . . . . . . . . . 415
16.2.1 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
16.2.2 Numerical Results . . . . . . . . . . . . . . . . . . . . 419
16.3 Continuous Stirred Tank Reactor . . . . . . . . . . . . . . . 420
16.3.1 Linearization-Based Scheduling and Risk-Aware Con-
trol Problem . . . . . . . . . . . . . . . . . . . . . . . 423
16.3.2 Gain-Scheduled Mean-Field-Type Control . . . . . . . 425
16.3.2.1 Design . . . . . . . . . . . . . . . . . . . . . 425
16.3.2.2 Local Stability of the Operating Points . . . 427
16.3.3 Risk-Aware Numerical Illustrative Example . . . . . . 428
16.4 Mechanism Design in Evolutionary Games . . . . . . . . . . 433
16.4.1 A Risk-Aware Approach to the Equilibrium Selection 437
16.4.1.1 Known Desired Nash Equilibrium . . . . . . 437
16.4.1.2 Unknown Desired Nash Equilibrium . . . . . 440
16.4.2 Risk-Aware Control Design . . . . . . . . . . . . . . . 441
16.4.3 Illustrative Example . . . . . . . . . . . . . . . . . . . 443
16.5 Multi-level Building Evacuation with Smoke . . . . . . . . . 445
16.5.1 Markov-Chain-Based Motion Model for Evacuation over
Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . 446
16.5.1.1 Reaching Evacuation Areas . . . . . . . . . . 449
16.5.1.2 Evacuation of the Whole Area . . . . . . . . 453
16.5.1.3 Jump Intensities for Evacuation . . . . . . . 455
16.5.2 Markov-Chain-Based Modeling for Smoke Motion . . . 456
16.5.3 Mean-Field-Type Control for the Evacuation . . . . . 457
16.5.4 Single-Level Numerical Results . . . . . . . . . . . . . 461
16.6 Coronavirus Propagation Control . . . . . . . . . . . . . . . 466
16.6.1 Single-Player Problem . . . . . . . . . . . . . . . . . . 467
16.6.1.1 Control Problem of Mean-Field Type . . . . 469
xii Contents
Bibliography 475
Index 489
List of Figures
xiii
xiv List of Figures
7.5 Evolution of the optimal control inputs u∗i and u∗j for the
scalar-value Stackelberg scenario with Brownian motion and
Poisson jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.6 Evolution of the differential equations αi and αj for the scalar-
value Stackelberg scenario with Brownian motion and Poisson
jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.7 Evolution of the differential equations βi and βj for the scalar-
value Stackelberg scenario with Brownian motion and Poisson
jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.8 Evolution of the differential equations γi and γj for the scalar-
value Stackelberg scenario with Brownian motion and Poisson
jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
7.9 Evolution of the differential equations δi and δj for the scalar-
value Stackelberg scenario with Brownian motion and Poisson
jumps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
12.1 Evolution of the (a) system state and (b) its expectation for
the matrix-value discrete-time non-cooperative scenario. . . 345
12.2 Evolution of the (a) first player strategies and (b) their ex-
pectation for the matrix-value discrete-time non-cooperative
scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
12.3 Evolution of the (a) second player strategies and (b) their ex-
pectation for the matrix-value discrete-time non-cooperative
scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
12.4 Evolution of the equation P1 for the matrix-value discrete-time
non-cooperative scenario. . . . . . . . . . . . . . . . . . . . . 349
12.5 Evolution of the equation P2 for the matrix-value discrete-time
non-cooperative scenario. . . . . . . . . . . . . . . . . . . . . 349
12.6 Evolution of the equation P̄1 for the matrix-value discrete-time
non-cooperative scenario. . . . . . . . . . . . . . . . . . . . . 350
12.7 Evolution of the equation P̄2 for the matrix-value discrete-time
non-cooperative scenario. . . . . . . . . . . . . . . . . . . . . 350
12.8 Evolution of the Riccati equations δ1 , and δ2 for the matrix-
value discrete-time non-cooperative scenario. . . . . . . . . . 351
12.9 Optimal cost function for each player for the matrix-value
discrete-time non-cooperative scenario. . . . . . . . . . . . . 351
12.10 Evolution of the (a) system state and (b) its expectation for
the matrix-value discrete-time fully-cooperative scenario. . . 352
List of Figures xix
12.11 Evolution of the (a) first player strategies and (b) their ex-
pectation for the matrix-value discrete-time fully-cooperative
scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
12.12 Evolution of the (a) second player strategies and (b) their ex-
pectation for the matrix-value discrete-time fully-cooperative
scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
12.13 Evolution of the equation P0 for the matrix-value discrete-time
fully-cooperative scenario. . . . . . . . . . . . . . . . . . . . 355
12.14 Evolution of the equation P̄0 for the matrix-value discrete-time
fully-cooperative scenario. . . . . . . . . . . . . . . . . . . . 355
12.15 Evolution of the Riccati equation δ0 for the matrix-value
discrete-time fully-cooperative scenario. . . . . . . . . . . . . 356
12.16 Optimal cost function for each player for the matrix-value
discrete-time fully-cooperative scenario. . . . . . . . . . . . . 356
13.1 Gap between var xVI = xbest−GNE and var xGlobal as dif-
ferent parameters εij = ε, for all i, j ∈ N , and number of
players n change for a fixed variance σ = 1. . . . . . . . . . . 369
13.2 Different topologies for the comparison among δ−parameters
in (13.11a) with εij ≥ 0, for all i, j ∈ N . . . . . . . . . . . . 371
13.3 Evolution of the variables w and x under the learning algo-
rithm in (13.13) for the constrained MFTG presented in (13.4)
with n = 2. Figures correspond to: (a) evolution of w1 and w2 ,
and (b)-(c) evolution of x1 and x2 . . . . . . . . . . . . . . . 372
16.6 Control inputs variance comparison for the two different MFT-
MPC controllers. . . . . . . . . . . . . . . . . . . . . . . . . 413
16.7 Results corresponding to the proposed stochastic MFT-MPC
controllers for Scenarios 1 and 2; and behavior of the deter-
ministic MPC controller. . . . . . . . . . . . . . . . . . . . . 414
16.8 General scheme of the micro-grid involving energy storage.
(Adapted from [1].) . . . . . . . . . . . . . . . . . . . . . . . 415
16.9 Evolution of the noise applied to the system. . . . . . . . . . 416
16.10 Evolution of the system state. . . . . . . . . . . . . . . . . . 417
16.11 Evolution of the control input for the first player and its ex-
pectation, i.e., u1,k and E[u1,k ]. . . . . . . . . . . . . . . . . 417
16.12 Evolution of the control input for the second player and its
expectation, i.e., u2,k and E[u2,k ]. . . . . . . . . . . . . . . . 418
16.13 Evolution of the control input for the third player and its
expectation, i.e., u3,k and E[u3,k ]. . . . . . . . . . . . . . . . 418
16.14 Continuous stirred tank reactor. . . . . . . . . . . . . . . . . 420
16.15 Gain-scheduled Mean-Field-Type Control Diagram with n op-
eration points and θ ∈ {1, . . . , n}. . . . . . . . . . . . . . . . 425
16.16 Gain-scheduled mean-field-free control diagram with n opera-
tion points and θ ∈ {1, . . . , n}. . . . . . . . . . . . . . . . . . 426
16.17 Noise Brownian motions for both the reactant concentration
and the reactor temperature. . . . . . . . . . . . . . . . . . . 429
16.18 Performance of the GS-MFTC. Evolution of the reactant con-
centration CA and its expectation E[CA ] tracking the reference
CA ref . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
16.19 Performance of the GS-MFTC. Evolution of the reactor tem-
perature TR and its expectation E[TR ] tracking the reference
TRref . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
16.20 Evolution of the optimal control input and its expectation. . 430
16.21 (a)–(b) Evolutionary dynamics with imperfect fitness observa-
tion for a population game, and two-strategy population game.
(c)–(d) Closed-loop of the multi-layer game for the equilibrium
selection and two-strategy population game. . . . . . . . . . 436
16.22 Projection dynamics behavior for both the RSP and Zeeman
game with and without noisy fitness functions. . . . . . . . . 438
16.23 Behavior of the risk-aware controller over the projection dy-
namics for both the RSP and Zeeman game with imperfect
fitness observation. . . . . . . . . . . . . . . . . . . . . . . . 444
16.24 Representation of the space B, its respective discretization
into n regions, and an example graph G, and V = {46},
O = {43, 49}, and F = {50}. . . . . . . . . . . . . . . . . . . 447
16.25 Representation of spacial constraints such as walls and obsta-
cles in the graph G f . . . . . . . . . . . . . . . . . . . . . . . 448
16.26 Relationship between possible jump intensities in the Markov
chain and the links E in a connected graph G. . . . . . . . . 449
List of Figures xxi
9.1 Exchange rates involving six currencies on October 11th, 2018 254
xxiii
Foreword
Mean field was first studied in physics for the behavior of systems with large
numbers of negligible individual particles. Recently mean-field game theory
was introduced in the economics and engineering literature to study the strate-
gic decision-making by small interacting agents of huge populations. Typically
a mean-field game is described by a Fokker-Planck equation, and solved by
a Hamilton-Jacobi-Bellman equation, which requires the number of agents
approaches infinity. This assumption limits the practical usage of mean-field
game theory in engineering fields.
Thanks to my friend, professor Hamidou Tembine, who also mentored my
former Ph.D. student majored in mean-field game theory, his team and collab-
orators including Dr. Julian Barreiro-Gomez have pioneer works to introduce
mean-field-type game theory to engineering scenarios. Mean-field-type games
differ from mean field games since it takes into account higher-order statis-
tics, it can be employed when dynamic programming cannot be applied, the
number of interacting agents is not necessarily large, and it can handle non-
symmetry non-negligible effect of individual decision on the mean field term.
Those significant advantages of mean-field-type game theory open a whole
gate for solving complex engineering problems that cannot be handled by
classic methods.
With such a demand from engineering audiences, this book is very timely
and provides a thorough study of mean-field-type game theory. The strenu-
ous protagonist of this book is to bridge between the theoretical findings and
engineering solutions. The book introduces the basics first, and then math-
ematical frameworks are elaborately explained. The engineering application
examples are shown in detail, and the popular learning approaches are also
investigated. Those advantageous characteristics will make this book a com-
prehensive handbook of many engineering fields for many years, and I will
buy one when it gets published.
xxv
Preface
If you have picked this book, you are probably already aware about how
powerful and suitable the mean-field-type control and game theory is in order
to solve risk-aware problems in the engineering framework, and that a large
variety of control and dynamic game problems can be set as particular cases
of the mean-field-type games.
Our main goal in this textbook is to provide a quite comprehensive and
simple treatment of the mean-field-type control and game theory, which can
also be interpreted as risk-aware optimal interactive decision-making tech-
niques. To this end, we exclusively focus on the so-called direct method either
in continuous or discrete time. Our experience indicates that other existing
methods reported in the literature to solve the class of stochastic problems
we address in this book, such as partial-differential-equation-based methods,
chaos expansion, dynamic programming, or the stochastic maximum princi-
ple, are not appropriate to start teaching beginner students in the field neither
early-career researchers. We recommend to focus on understanding this book
prior to moving on the study of other research manuscripts using other theo-
retical directions. In this regard, the contents of this book comprises an appro-
priate background to start working and doing research in this game-theoretical
field.
To make the exposition and explanation even easier, we first study the de-
terministic optimal control and differential linear-quadratic games. Then, we
progressively add complexity step-by-step and little-by-little to the problem
settings until we finally study and analyze mean-field-type control and game
problems incorporating several stochastic processes, e.g., Brownian motions,
Poisson jumps, and random coefficients.
This smooth trip, starting with a scalar-valued state optimal control prob-
lem in continuous and discrete time, passes through the scalar-valued deter-
ministic differential games and mean field games, the stochastic state-and-
control-input independent diffusion differential games and mean field games,
until we finally address the mean-field-type games with state-and-control-
input dependent diffusion terms and incorporating Poisson jumps and random
coefficients by means of switching regimes. On the other hand, we go beyond
the Nash equilibrium, which provides a solution for non-cooperative games,
by analyzing other game-theoretical concepts such as the Berge, Stackelberg,
adversarial and co-opetitive equilibria. For the mean-field-type game analysis,
we provide several numerical examples, which are obtained from a MatLab-
xxvii
xxviii Preface
based user-friendly toolbox that is available for the free use of the readers of
this book.
We devote a whole part of the book to discuss about some learning ap-
proaches that guarantee converge to mean-field-type solutions. In particular,
we present the constrained and static mean-field-type games where optimiza-
tion algorithms may be applied such as distributed evolutionary dynamics, the
receding horizon mean-field-type control also know as risk-aware model pre-
dictive technique, and the data-driven mean-field-type games motivating the
use of artificial intelligence tools such as machine learning with either neural
networks or simple linear regression. Finally, we present several engineering
applications in both continuous and discrete time. Among these applications
we find the following: water distribution systems, micro-grid energy storage,
stirred tank reactor, mechanism design for evolutionary dynamics, multi-level
building evacuation problem, and the COVID-19 propagation control.
Julian Barreiro-Gomez
Hamidou Tembine
Acknowledgments
We gratefully acknowledge support from the US Air Force, and the New York
University in the US campus (NYU) and the UAE campus (NYUAD), for
the research conducted at the Learning & Game Theory Laboratory (L&G
Lab) and at the Center on Stability, Instability and Turbulence (SITE). This
material is based upon work supported by Tamkeen under the NYU Abu
Dhabi Research Institute grant CG002.
We also acknowledge our friends, faculty members, and researchers with
whom we have had several scientific discussions about mean-field-type control
and game theory, and also regarding its potential for engineering applications.
We specially thank Prof. Tyrone E. Duncan and Prof. Bozenna Pasik-Duncan
from the mathematics department at Kansas University in the US, and Prof.
Boualem Djehiche from the mathematics department at Royal KTH in Swe-
den. We finally acknowledge all our co-authors with whom we have published
several articles in the mean-field-type field.
xxix
Author Biographies
Julian Barreiro-Gomez received his B.S. degree (cum laude) in Electronics En-
gineering from Universidad Santo Tomás (USTA), Bogota, Colombia, in 2011.
He received the M.Sc. degree in Electrical Engineering and the Ph.D. degree
in Engineering from Universidad de Los Andes (UAndes), Bogota, Colombia,
in 2013 and 2017, respectively. He received the Ph.D. degree (cum laude) in
Automatic, Robotics and Computer Vision from the Technical University of
Catalonia (UPC), Barcelona, Spain, in 2017; the best Ph.D. thesis in control
engineering 2017 award from the Spanish National Committee of Automatic
Control (CEA) and Springer; and the EECI Ph.D. Award from the European
Embedded Control Institute in recognition to the best Ph.D. thesis in Eu-
rope in the field of Control for Complex and Heterogeneous Systems 2017.
He received the ISA Transactions Best Paper Award 2018 in Recognition to
the best paper published in the previous year. Since August 2017, he has
been a Post-Doctoral Associate in the Learning & Game Theory Laboratory
(L&G-Lab) at the New York University in Abu Dhabi (NYUAD), United
Arab Emirates, and since 2019, he has also been with the Research Center
on Stability, Instability and Turbulence (SITE) at the New York University
in Abu Dhabi (NYUAD). His main research interests are: risk-aware control
and games, mean-field-type games, constrained evolutionary game dynamics,
distributed optimization, stochastic optimal control, and distributed predic-
tive control.
Hamidou Tembine received the M.S. degree in applied mathematics from Ecole
Polytechnique, Palaiseau, France, in 2006 and the Ph.D. degree in computer
science from the University of Avignon, Avignon, France, in 2009. He is a
prolific Researcher and holds more than 150 scientific publications including
magazines, letters, journals, and conferences. He is an author of the book on
Distributed Strategic Learning for Engineers (CRC Press, Taylor & Francis
2012), and Coauthor of the book Game Theory and Learning in Wireless Net-
works (Elsevier Academic Press). He has been co-organizer of several scientific
meetings on game theory in networking, wireless communications, smart en-
ergy systems, and smart transportation systems. His current research interests
include evolutionary games, mean-field stochastic games and applications. Dr.
Tembine received the IEEE ComSoc Outstanding Young Researcher Award
for his promising research activities for the benefit of the society in 2014. He
received the best paper awards in the applications of game theory.
xxxi
Symbols
Symbol Description
xxxiii
xxxiv Symbols
Preliminaries
1
Introduction
We truly live in a more and more interconnected and interactive world. In re-
cent years, we have seen emerging technologies such as internet of everything,
collective intelligence including Artificial Intelligence (AI), blockchains, next-
generation wireless networks, among many others. The quantities-of-interest
in these systems involve both volatilities and risks.
A typical example of risk concerns in the current online market is the
evolution of prices for the digital and cryptocurrencies (e.g., bitcoin, litecoin,
ethereum, dash, and other altcoins (alternatives to bitcoin, etc.). The variance
plays a base model for many risk measures. From random-variable perspective
(probability theory) the volatility can be captured by means of the variance,
which is a mean-field term comprising the second moment and the square of
the mean. Another example concerns the variations of wireless channels in
multiple-input-multiple-output systems. Non-Gaussianity of wireless channels
has been observed experimentally and empirically, and its variability affects
the quality of the communication.
The term mean-field has been referred to as a physics concept that at-
tempts to describe the effect of an infinite number of particles on the motion
of a single particle. Researchers began to apply the concept to social sciences
in the early 1960s to study how an infinite number of factors affect individual
decisions. However, the key ingredient in a game-theoretic context is the in-
fluence of the distribution of states and/or control actions onto the payoffs of
the decision-makers. Notice that there is no need to have a large population
of decision-makers. A mean-field-type game is a game in which the payoffs
and/or the state dynamics coefficient functions involve not only the state
and actions profiles but also the distributions of state-action process (or its
marginal distributions).
Games with distribution-dependent quantity-of-interest such as state
and/or payoffs are particularly attractive because they capture not only the
mean, but also the variance and higher order terms. Such incorporation of
these mean and variance terms is directly associated with the paradigm intro-
duced by H. Markowitz, 1990 Nobel Laureate in Economics. The Markowitz
paradigm, also termed as the mean-variance paradigm, is often characterized
as dealing with portfolio risk and (expected) returns [2–4].
In this book, we address variance reduction problems when several
decision-making entities take place. When the decisions made by the
agents/players/decision-makers influence each other, the decision-making is
DOI: 10.1201/9781003098607-1 3
4 Mean-Field-Type Games for Engineers
Robert Aumman
[5, page 47]