International Society on Dynamic Games

News


March 1, 2021

Dynamic Games and Applications Seminar

 

March 4, 2021 11:00 AM – 12:00 PM (Montreal time)

Robust communication on networks

Marie Laclau – HEC Paris, France

Marie Laclau

Webinar link
Webinar link: 962 7774 9870
Passcode: 285404

We consider sender-receiver games, where the sender and the receiver are two distinct nodes in a communication network. Communication between the sender and the receiver is thus indirect. We ask when it is possible to robustly implement the equilibrium outcomes of the direct communication game as equilibrium outcomes of in- direct communication games on the network. Robust implementation requires that: (i) the implementation is independent of the preferences of the intermediaries and (ii) the implementation is guaranteed at all histories consistent with unilateral deviations by the intermediaries. We show that robust implementation of direct communication is possible if and only if either the sender and receiver are directly connected or there exist two dis- joint paths between the sender and the receiver. We also show that having two disjoint paths between the sender and the receiver guarantees the robust implementation of all communication equilibria of the direct game. We use our results to reflect on organizational arrangements.

 

(joint with Ludovic Renou and Xavier Venel)

Published by Sergey Kumkov
February 22, 2021

ISS Informal Systems Seminar

Feb 26, 2021 02:00 PM – 03:00 PM (Montreal time)

A neural network approach for high-dimensional optimal control

Derek Onken – Emory University, United States

 

Webinar link
Webinar ID: 910 7928 6959
Passcode: VISS

 

Optimal control (OC) problems aim to find an optimal policy that control given dynamics over a period of time. For systems with high-dimensional state (for example, systems with many centrally controlled agents), OC problems can be difficult to solve globally. We propose a neural network approach for solving such problems. When trained offline in a semi-global manner, the model is robust to shocks or disturbances that may occur in real-time deployment (e.g., wind interference). Our unsupervised approach is grid-free and scales efficiently to dimensions where grids become impractical or infeasible. We demonstrate the effectiveness of our approach on several multi-agent collision-avoidance problems in up to 150 dimensions

Published by Sergey Kumkov
February 22, 2021

Dynamic Games and Applications Seminar; GERAD, Chair in Game Theory and Management

Feb 25, 2021 11:00 AM – 12:00 PM (Montreal time)

Improving treatment of metastatic cancer through game theory

Katerina Stankova – Department of Data Science and Knowledge Engineering, Maastricht University, Netherlands

 

Webinar link
Webinar ID: 962 7774 9870
Passcode: 285404

 

In this talk, we will investigate cancer treatment as a game-theoretic contest between the physician's therapy and the cancer cells' resistance strategies. This game has two critical asymmetries: (1) Only the physician can play rationally. Cancer cells, like all evolving organisms, can only adapt to current conditions; they can neither anticipate nor evolve adaptations for treatments that the physician has not yet applied. (2) It has a distinctive Stackelberg structure; the "leader" oncologist plays first and the "follower" cancer cells then respond and adapt to therapy. We will learn how the physician can utilize his advantages in this game. 

Published by Sergey Kumkov
February 15, 2021

Dynamic Games and Applications Seminar

Feb 18, 2021 11:00 AM – 12:00 PM

Provable reinforcement learning for multi-agent and robust control systems

 Kaiqing Zhang – University of Illinois at Urbana-Champaign, United States

Webinar link
Webinar ID: 962 7774 9870
Passcode: 285404

 

Recent years have witnessed both tremendous empirical successes and fast-growing theoretical development of reinforcement learning (RL), in solving many sequential decision-making and control tasks. However, many RL algorithms are still several miles away from being applied to practical autonomous systems, which usually involve more complicated scenarios with multiple decision-makers and safety-critical concerns. In this talk, I will introduce our work on the development of RL algorithms with provable guarantees, with focuses on the multi-agent and safety-critical settings. I will first show that policy optimization, one of the main drivers of the empirical successes of RL, enjoys global convergence and sample complexity guarantees for a class of robust control problems. More importantly, we show that certain policy optimization approaches automatically preserve some "robustness" during the iterations, some property we termed as "implicit regularization". Interestingly, such a setting naturally unifies other important benchmark settings in control and game theory: risk-sensitive control design, and linear quadratic zero-sum dynamic games, while the latter is the benchmark multi-agent RL (MARL) setting that mirrors the role played by linear quadratic regulators (LQR) for single-agent RL. Despite the nonconvexity and the fundamental challenges in the optimization landscape, our theory shows that policy optimization enjoys global convergence guarantees in these problems as well. The results have then provided some theoretical justifications for several basic robust RL and MARL settings that are popular in the empirical RL world. In addition, I will introduce several other works along this line of provable MARL and robust RL, including decentralized MARL with networked agents, sample complexity of model-based MARL, etc. Time permitting, I will also share several future directions based on the previous results, towards large-scale and reliable autonomy.

Published by Sergey Kumkov
February 8, 2021

Dynamic Games and Applications Seminar

Chair in Game Theory and Management , GERAD

 Feb 11, 2021 11:00 AM – 12:00 PM

Advertising, goodwill, and the Veblen effect

Régis Chenavaz KEDGE Business School, France

Régis Chenavaz

Webinar link
Webinar ID: 962 7774 9870
Passcode: 285404

 

The increase of demand in price, an exception to the law of demand, is known as the Veblen effect. We show that advertising and goodwill play an important role in the Veblen effect. By employing optimal control theory, we capture the evolution of the variables over time which may exhibit the Veblen effect where price and demand move in the same direction. Incorporating this dynamics into firms’ decisions has a promising impact on long-term profit. Consequently, it may even trigger a slew of studies on product line extension, competition and pricing by allowing firms to control their status.

Published by Sergey Kumkov
February 3, 2021

Dear Colleagues,

 The 19th International Symposium on Dynamic Games and Applications will be held in Porto, on August 23-27, 2021. Please save the date!

 Given the still ongoing uncertainties about the feasibility of in-person conferences next summer, a final decision will be communicated by end of April. The decision will be either to have the Symposium in person in Porto at the above date, or to postpone it to 2022.

 We hope to be able to meet in Porto in August.

 

Best wishes,   Mark Broom, Alberto Pinto and Georges Zaccour

 

Published by Sergey Kumkov
February 2, 2021

ISS Informal Systems Seminar

Feb 5, 2021 10:00 AM – 11:00 AM (Montreal time)

Dynamic games among teams with asymmetric information

Dengwang Tang – University of Michigan, United States

Dengwang Tang

Webinar link
Webinar ID: 910 7928 6959
Passcode: VISS

 

Dynamic games with asymmetric information appear in many social-economic contexts. In these games, multiple agents/decision makers, interact repeatedly in a changing environment. Agents have different information and seek to optimize their respective long-term payoffs. Examples include market competition, cyber-security, and transportation networks. In some settings, agents can form groups, or teams. The agents in the same group share a common goal but may have different information available to them. In this talk, I focus on a class of stochastic dynamic games among teams with asymmetric information, where members of a team share their observations internally with a delay of d>0. I will describe a general approach to characterize a subset of Nash Equilibria where the agents can use a compressed version of their information, instead of the full information, to choose their actions. Our results highlight the tension among compression of information, existence of (compression based) equilibria, and backward inductive sequential computation of such equilibria in stochastic dynamic games with asymmetric information. This is joint work with Hamidreza Tavafoghi, Vijay Subramanian, Ashutosh Nayyar, and Demosthenis Teneketzis.

Published by Sergey Kumkov
February 2, 2021

Dynamic Games and Applications Seminar

Feb 4, 2021 11:00 AM – 12:00 PM (Montreal time)

Incentive mechanism design using linear matrix inequality approach

Pegah Rokh Foroz – School of Electrical and Computer Engineering, University of Tehran, Iran, and ETH Zurich, Switzerland

Webinar link
Webinar ID: 962 7774 9870
Passcode: 285404

We consider a centralized multi-agent optimization problem with coupling constraint among agents, where the information is distributed between agents who are strategic, selfish and have the private value functions. To achieve a global optimal solution, we propose an incentive mechanism based on a message space and payment function, which makes a non-cooperative game among agents where an individual utility function of each agent aligns with the centralized optimization problem. We construct a family of payment function, which leads to a potential game among agents and ensures the strongly Nash implementation, budget balance and individual rationality of the induced mechanism using the linear matrix inequality approach. 

Published by Sergey Kumkov
January 25, 2021

ISS Informal Systems Seminar

Jan 29, 2021 02:00 PM – 03:00 PM (Montreal time)

LQG mean field games with a major agent: Nash certainty equivalence versus probabilistic approach

Dena Firoozi – Department of Decision Sciences, HEC Montréal, Canada

 

 

Webinar link

Webinar ID: 910 7928 6959

Passcode: VISS

Mean field game systems consisting of a major agent and a large population of minor agents were introduced in (Huang, 2010) in an LQG setup. In the past years several approaches towards major-minor mean field games have been developed, principally (i) the Nash certainty equivalence (Huang, 2010), (ii) master equations, (iii) asymptotic solvability, and (iv) the probabilistic approach. In a recent work (Huang, 2020), for the LQG case the equivalence of the solutions obtained via approaches (i)-(iii) was established. In this talk we first review approaches (i) and (iv). We then demonstrate that the closed-loop Nash equilibrium derived in the infinite-population limit through (i) and (iv) are identical.

Published by Sergey Kumkov
January 25, 2021

Dynamic Games and Applications Seminar

Chair in Game Theory and Management

GERAD

Jan 28, 2021 11:00 AM – 12:00 PM (Montreal time)

Fuzzy fractional-order model of the novel coronavirus: The impact of delay strategies on the pandemic dynamics model with nonlinear incidence rate

Massimiliano Ferrara – Mediterranea University of Reggio Calabria, Italy

 

Webinar link

Meeting ID: 962 7774 9870

Pass code: 285404

 

In this paper, a novel coronavirus infection system with a fuzzy fractional differential equation defined in Caputo’s sense is developed. By using the fuzzy Laplace method coupled with Adomian decomposition transform, numerical results are obtained for better understanding of the dynamical structures of the physical behavior of COVID-19. Such behavior on the general properties of RNA in COVID-19 is also investigated for the governing model. Due to non-availability of the vaccination, delay strategies such as social distancing, travel restrictions, extension in holidays, use of facemask, and self- quarantine are the effective treatment to control the pandemic of coronavirus. So, we proposed the delayed susceptible-exposed- infected-recovered model with nonlinear incidence rate to study the effective role of delay strategies. For this analysis, we discussed three types of equilibria of the model such as trivial, coronavirus free and coronavirus existence with delay term. The local and global stabilities are investigated by using well-posed notation, Routh Hurwitz criterion, Lyapunov function, and Lasalle invariance principle. 

Published by Sergey Kumkov
January 18, 2021

 

ISS Informal Systems Seminar

Jan 22, 10:00 AM – 11:00 AM (Montreal time)

Mean-field games models of price formation

Joao Saude – JP Morgan AI Research, United States

 

Joao Saude

Webinar link
Webinar link: 910 7928 6959
Passcode: VISS

We consider dynamical systems with a large number of agents that can store and trade a commodity such as electricity. We present a price-formation model consisting of constrained mean-field games where the price is a Lagrange multiplier for the supply vs. demand balance condition. We illustrate the model using real data of daily energy consumption in the UK. Then we present a Fourier approximation method for the solutions of first-order nonlocal mean-field games. We approximate the system by a simpler one that is equivalent to a convex optimization problem over a finite-dimensional subspace of continuous curves. Time permitting, we discuss possible applications to price formation problems where prices depend on state and time.

Published by Sergey Kumkov
January 18, 2021

Dynamic Games and Applications Seminar

Jan 21, 11:00 AM – 12:00 PM (Montreal time)

Control of an epidemic with endogenous treatment capability under popular discontent and social fatigue

Fouad El Ouardighi – ESSEC Business School, France

Fouad El Ouardighi

 

Webinar link
Meeting ID: 962 7774 9870
Passcode: 285404

The primary issue in this paper is to determine whether mobility restrictions or securing social interactions is most effective in countering an epidemic disease that spreads also via asymptomatic transmission. We develop an optimal control policy model wherein i) treatment capabilities are endogenous, ii) the social loss due to disease-related deaths is part of the tradeoff in terms of health and social welfare perspectives, iii) the policymaker's inability to counter the disease gives rise to growing popular discontent over time, and iv) nontherapeutic intervention policy engenders social fatigue over time. We also allow for partial immunity upon recovery. In many ways, our model applies to the recent pandemic caused by the SARS-Cov-2 virus. In this setup, we identify which non-therapeutic policy option between mobility restrictions or securing social interactions most effectively minimizes both the impact of policymaker’s inability and the ensuing popular discontent and social fatigue.

Published by Sergey Kumkov
January 11, 2021

ISS Informal Systems Seminar

Jan 15, 02:00 PM – 03:00 PM (Montreal time)

 

Social learning under behavioral assumptions

Rabih Salhab – Institute for Data, Systems, and Society, MIT, United States

Rabih Salhab

Webinar link

Webinar ID: 910 7928 6959

Passcode: VISS

I will present two recent works on social learning under behavioral assumptions. The first is Social Learning with Sparse Belief Samples. In this work, we introduce a non-Bayesian model of learning over a social network where a group of agents with insufficient and heterogeneous sources of information share their experiences to learn an underlying state of the world. Inspired by a recent body of research in cognitive science on human decision making, we presume two behavioral assumptions. Motivated by the coarseness of communication, our first assumption posits that agents only share samples taken from their belief distribution over the set of states, to which we refer as their actions. This situation is to be contrasted with that of sharing the full belief, i.e. probability distribution over the entire set of states. The second assumption is limited cognitive power, based on which individuals incorporate their neighbors’ actions into their beliefs following a simple DeGroot-like social learning rule which suffers from redundancy neglect and imperfect recall of the past history. We show that so long as all the individuals trust their neighbors’ actions more than their private signals, they may end up mislearning the state with positive probability. Learning, on the other hand, requires that the population includes a group of self-confident experts in different states. This means that for each state, there is an agent whose signaling function for her state of expertise is distinguishable from the convex hull of the remaining signaling functions, and that her private signals sufficiently weigh in her social learning rule. This is a joint work with Amir Ajorlou and Ali Jadbabaie.

The second work is Social Learning with Unreliable Agents and Self-reinforcing Stochastic Dynamics. We consider a group of agents that have fixed unobservable binary ``beliefs’’. An individual’s belief models for example their political support (Democrat or Republican). At each time period, agents broadcast binary opinions on a social network. We assume that individuals may lie and declare opinions different from their true beliefs to conform with their neighbors. This raises the natural question as to whether one can estimate the agents’ true beliefs from observations of declared opinions. We analyze this question in the special case of complete graph. We show that, as long as the population does not include large majorities, estimation of aggregate true belief and individual true beliefs is possible. On the other hand, large majorities force minorities to lie as time goes to infinity, which makes asymptotic estimation impossible. This is a joint work with Anuran Makur, Ali Jadbabaie, and Elchanan Mossel.

Published by Sergey Kumkov