2016 INFORMS Annual Meeting Program

478 / 561

INFORMS Nashville – 2016
478
2 - Better Bounds On Convergence Rates For First-order Methods In
Sequential Games
Christian Kroer, Carnegie Mellon University, Computer Science
Department, 5000 Forbes Avenue, Pittsburgh, PA, 15213, United
States, , Fatma Kılınç-Karzan, Kevin Waugh,
Tuomas W Sandholm
First-order methods, are known to be very effective in solving large-scale two-
player zero-sum extensive-form games. The convergence rates of these methods
depend heavily on an associated distance-generating function. By introducing a
new weighting scheme for the dilated entropy function, we develop the first
distance-generating function for the strategy spaces of sequential games that has
no dependence on the branching factor of the player. This significantly improves
the convergence rate of several first-order methods. Numerical experiments
further support the effectiveness of our new weighting scheme by demonstrating
better practical convergence rates.
3 - Formulation Of A Continuous Equilibrium Mapping And
A Simplicial Path-following Method For Approximating
Proper Equilibria
Yabin Sun, City University of Hong Kong, Nam Shan Office-705,
Hong Kong, Yin Chen,
Chuangyin Dang
This paper formulates a continuous equilibrium mapping for a specifically defined
perturbed game that deforms with an extra variable. As a result of this
formulation, we develop a simplicial path-following method for approximating a
proper equilibrium. The method follows a simplicial path that starts from any
given totally mixed strategy profile and leads to a proper equilibrium. Numerical
results further confirms the effectiveness of the method.
4 - Price Of Anarchy In Bayesian Congestion Games
Manxi Wu, Massachusetts Institute of Technology, 235 Albany
Street, 3112B, Cambridge, MA, 02139, United States,
, Saurabh Amin
We study a new model of congestion games to understand the effects of
asymmetric information on the equilibrium costs of individual players and the
social cost. We present a full characterization of the Bayesian Wardrop
Equilibrium (BWE) of the game by exploiting the properties of a weighted
potential function. We compare the equilibrium cost in any BWE to the socially
optimal costs under incomplete and perfect information structures. Our main
result estimates the contribution to the price of anarchy (PoA) of the game due to
two factors: selfish routing and information heterogeneity.
WD72
Bass- Omni
Supply Chain, Decision I
Contributed Session
Chair: Richard Titus, PhD Candidate at Penn State, Adjunct Lehigh
Univ, The Pennsylvania State University, 310 Leonhard Building,
University Park, PA, 16802, United States,
1 - Can A Buyer Benefit From Bounded Rationality?
Sha Luo, North Carolina State University, 4335 Avent Ferry Road,
Apt 3, Raleigh, NC, 27606, United States, ,
Russell Edward King, Shu-Cherng Fang
Abstract: Bounded rationality takes into account that human decision makers are
prone to biases and mistakes. General wisdom suggests that people fail to reap the
highest level of benefits when they are not perfectly rational. In this study, we
consider a supplier pricing game when the retailer is not a perfect optimizer. The
decision model of bounded rationality is derived from the classical logit model.
We apply the concept of Nash equilibrium to predict the rational outcome of the
pricing game when two suppliers submit their wholesale prices simultaneously. It
is found that bounded rationality can be advantageous to the buyer when her
bounded rationality level is moderate.
2 - Using Point-of-sale Data To Improve Shelf Replenishment
Performance
Suzanne de Treville, University of Lausanne, Anthropôle 3073,
Lausanne-Dorigny, CH1015, Switzerland,
, Valérie Chavez-Demoulin,
Joerg S Hofstetter
Shelf replenishment for a retail chain is done once daily. Point-of-sale data lets us
estimate the value of a second replenishment opportunity. We begin from a
standard newsvendor approach with flexible second-replenishment capacity made
available, and show that flexible capacity will often outperform capacity allocated
at the level of single stock-keeping units. We define a “Demand at Risk” measure
similar to Value at Risk in quantitative finance that captures the highest demand
that might be reasonably encountered, and use machine-learning techniques to
identify data features of the data that may signal increases in the Demand at Risk,
warranting priority replenishment.
3 - A Structurally Enhanced Fuzzy Inference System Partner
Selection Technique In Forming Virtual Enterprise
Shahrzad Nikghadam, PhD Student, Middle East Technical
University, 425, graduate studies dorm,METU, Cankaya, Ankara,
06800, Turkey, Hakki Ozgur
Unver, Ahmet Murat Ozbayoglu, Sadik Engin Kilic
In this study we considered a partner selection problem of Virtual Enterprise in
forming consortiums based on customer preferences. Previously, we have
presented a Fuzzy Inference System (FIS) based methodology for the problem.
Though, in this study, an enhanced form of the previous model is presented by
editing the fuzzy rules and their reasoning structure. New rules are established by
asking fewer questions from customer, yet the computational results show that it
still gives the reasonable decisions by having less detailed information about the
customer.
4 - Supplier Selection Models With Product Life
Cycle Considerations
Richard Titus, PhD Candidate at Penn State, Adjunct Lehigh Univ,
The Pennsylvania State University, 310 Leonhard Building,
University Park, PA, 16802, United States,
This research examines the global supplier selection process and includes supplier
attributes, determined by an empirical case study and product life cycle
considerations. The empirical study examines the relationship of key supplier
attributes to quality and delivery performance. Goal programming methodologies,
including Preemptive, Non-preemptive, Chebycheff’s Min/Max and Fuzzy GP, are
used to solve the multi-objective supplier selection problem for the four product
life cycle phases. An illustrative example and an industrial case study are included
in the testing of the supplier selection models.
WD73
Legends A- Omni
Operations Management VIII
Contributed Session
Chair: Donald Paul Warsing, North Carolina State University, College of
Management, Campus Box 7229, Raleigh, NC, 27695-7229,
United States,
1 - Leveraging Suppliers To Calibrate Product Specification
Sha Liao, University of the Fraser Valley, Vancouver, BC, Canada,
Hao Zhang, Yimin Wang
We examine a two-period game between an Original Equipment Manufacturer
(OEM) and his supplier during new product development. The supplier may
detect specification problems during production. We first prove that it is strictly
better for the OEM to design the contract so that the supplier will voluntarily
point out specification problems. We then characterize the optimal contract for
the OEM.
2 - Private Label Encroachment And Supply Chain
Information Sharing
Xue Li, Tsinghua University, Beijing, China, ,
Yanchong Zheng, Jian Chen
Our paper investigates the retailer’s private label encroachment strategy and its
implications for the national brand manufacturer and the retailer in the
environment where the national brand manufacturer may have private product
intrinsic information. We consider information sharing via cheap talk in this
setting, investigate the effects of private label encroachment on credible
information sharing and show that the private label strategically introduced by
the retailer may also benefit the national brand manufacturer.
3 - Supply Chain To Product Matching And Firm Performance
Donald Paul Warsing, North Carolina State University, College of
Management, Campus Box 7229, Raleigh, NC, 27695-7229,
United States, , Mohamed Desoky,
Russell Edward King, Mark Walker
We present an empirical analysis of Fisher’s (1997) conceptual model of supply
chain-to-product matching. We use longitudinal COMPUSTAT data across several
industries and develop proxies to measure Fisher’s characterizations of products
(functional vs. innovative) and supply chains (efficient vs. responsive), and we
find a strong, positive relationship between the strength of a firm’s market
performance and the strength of the match between its product and supply chain
characteristics.
WD72