Informs 2015

326 / 552

INFORMS Philadelphia – 2015
324
2 - s-plex and s-defective Numbers of a Graph
Vladimir Stozhkov, University of Florida, 2330 SW Williston Rd
Apt. 2826, Gainesville, FL, 32608, United States of America,
, Eduardo Pasiliao, Vladimir Boginski
The presentation is dedicated to two clique relaxation models: s-plex and s-
defective clique. Theoretical properties of the specified objects are investigated.
Analytical and computational bounds for the related optimization problems are
provided. The extensions of the Motzkin-Straus formulation for s-plex and s-
defective clique are derived. The outline of the general procedure for solving the
corresponding maximization problems is given.
3 - Minimum Edge Blocker Dominating Set Problem
Foad Mahdavi Pajouh, Assistant Professor, University of
Massachusetts Boston, 100 Morrissey Boulevard, Boston, MA,
02125, United States of America, ,
Eduardo Pasiliao, Jose Walteros, Vladimir Boginski
Dominating sets are widely used in social and communication networks analysis.
Given a weighted graph and r>0, we consider the problem of removing a
minimum number of edges so that the weight of any dominating set in the
remaining graph is at least r. Complexity results, polyhedral results, a linear 0-1
programming formulation, and an exact algorithm for solving this problem will be
presented.
4 - Minimum Risk Network Covering Location Problem
Konstantin Pavlikov, University of Florida, 1350 N. Poquito Road,
Shalimar, FL, 32579, United States of America,
Alexander Veremyev, Vladimir Boginski,
Eduardo Pasiliao
The network coverage problem under uncertainty is considered. In this problem,
components of the covering set and links connecting them to remaining nodes of
the network are subject to random failures. The emphasis is put on minimizing
the risk of losing coverage in presence of such failures. We formalize the model
and discuss its connection to the maximum expected covering location model.
TC19
19-Franklin 9, Marriott
Modeling and Optimization for Sustainable
Cloud Computing
Sponsor: Computing Society
Sponsored Session
Chair: Yunpeng Pan, South Dakota State University, Mathematics &
Statistics, Box 2220, Brookings, SD, 57007, United States of America,
1 - Remote Sensing Data Mining for Extracting Data Center
Site Characteristics
Yunpeng Pan, South Dakota State University, Mathematics &
Statistics, Box 2220, Brookings, SD, 57007, United States of
America, , Adam Buskirk
Data centers are powerhouses of cloud. Companies rush to build out their cloud
infrastructure to meet fast growing demand. The environmental impact such as
carbon footprint falls into the category of public good, and therefore, calls for
appropriate public policy decisions, which in turn require good information. Our
current work intends to achieve this by mining the Landsat remote sensing data
to extract characteristics of data centers in operation and under construction at a
global scale.
2 - A Dynamic Workflow Framework for Server Provisioning
Wei Lin, Software Engineering Researcher, IBM,
8 Dongbeiwang Western Road, Haidian Dist, Beijing, China,
Brian Peterson, Qinhua Wang,
Zongying Zhang, Christopher Young, Sai Zeng
Cloud service providers support server provisioning to large number of enterprise
customers, who have different functional, security and compliance requirements.
We propose a framework which composes dynamic workflow at runtime to cater
individualized provisioning procedures. In this framework, an onboarding module
configures process steps and dependencies for each customer, and a composition
module dynamically composes execution workflow based on dependency
validation and sequence calculating.
3 - Minimizing Costs in Distributed Cloud Resource Provisioning
Julio Goez, Postdoctoral Fellow, Ecole Polytechnique Montreal
and GERAD, 2900 Boulevard Edouard-Montpetit, Montréal, QC,
H3T 1J4, Canada, Juan F. Pérez
We consider the problem of minimizing the cost of provisioning resources at
different cloud locations, constrained to satisfying a required service-level
objective. We present a mixed integer non-linear optimization model for this
problem and show an equivalent mixed integer second order cone formulation.
We also show that a simple round-up provides an initial feasible solution for the
problem. We use this property to design a heuristic procedure to improve the
quality of the initial solution.
4 - Renewable Energy Prediction and Prescription in the
Internet-of-things (IoT)
Hans Schlenker, IBM, Hollerithstr 1, Munich, 81829, Germany,
Yianni Gamvros
The IoT connects all sorts of devices — from sensors to embedded devices to
smartphones to laptops to servers. IBM connected 1600 solar fields to its
Renewable Energy IoT. Sensor data is collected, combined in the cloud, and
further analyzed by analytics services to generate accurate local energy
production forecast. These predictions are then used by (prescriptive)
mathematical optimization in a network distribution model to balance under-runs
and over-production in all connected areas.
TC20
20-Franklin 10, Marriott
Financial Engineering and Optimization
Contributed Session
Chair: Zhen Liu, Options Clearing Corp (OCC), One North Wacker
Drive, Suite 500, Chicago, IL, 60606, United States of America,
1 - An Optimization Procedure for a Delta Neutral Constrained Theta
with Maximum Gamma Portfolio
Arik Sadeh, Dean, HIT Holon Institute of Technology,
52 Golomb St. 5810201, Holon, Israel,
A large gamma portfolio is attractive for investors in order to get benefits from
large increase or decrease in the value of the underlying asset. In large gamma
portfolio the theta is negative. In this study, a delta neutral portfolio with
maximum gamma and constrained theta, was developed. An optimization model
was designed and solved for small time steps within a planning horizon. The
model was run for many simulation scenarios as well as real world data, followed
by statistical tests.
2 - Optimal Portfolio Liquidation and Dynamic
Mean-variance Criterion
Jiawen Gu, Postdoc, University of Copenhagen, Department of
Mathematical Science, University of Copenhagen, Copenhagen,
2100, Denmark, , Mogens Steffensen
We consider the portfolio liquidation problem under the dynamic mean-variance
criterion and derive time-consistent solutions in three important models. We get
explicit trading strategies in the basic model and when random pricing signals are
incorporated. When consider stochastic liquidity and volatility, we construct an
exact HJB equations under general assumptions for the parameters.
3 - Genetic Programming Optimization for a Sentiment Feedback
Strength Based Trading Strategy
Steve Yang, Assistant Professor, Stevens Institute of Technology,
1 Castle Point on Hudson, Hoboken, NJ, 07030,
United States of America,
Based on the evidence that tweets are faster than news in revealing new market
information, whereas news is regarded broadly a more reliable source of
information than tweets, we develop a trading strategy based on the sentiment
feedback strength between the news and tweets using generic programming
optimization method. Result shows that this strategy generates over 14.7%
Sterling ratio compared with 10.4% and 13.6% from the technical indicator-
based and the buy-and-hold strategy respectively.
4 - Algorithmic Options Trading by Integer Programming
Vadim Timkovski, Keiser University, Port St. Lucie, FL,
United States of America,
Algorithmic options trading has only begun its evolution. This work presents an
integer programming system that simulates the activities of an experienced option
trader on the construction and adjustment of option portfolios. The system adopts
algorithms based on a recent discovery of an algebraic classification of option
trading strategies, without which this kind of automation would not be possible
and which has not been considered before as attainable.
5 - Linear Programming Approach to American Option Pricing
Zhen Liu, Options Clearing Corp (OCC), One North Wacker
Drive, Suite 500, Chicago, IL, 60606, United States of America,
We solve the variational inequality (VI) from American option pricing problem by
linear programming (LP) approach. We approximate its solution by a combination
of Chebyshev basis functions. The objective is to minimize the absolute error of
the solution and the max operator in VI is converted into linear constraints of LP.
We discuss its convergence, and compare our results with Longstaff-Schwartz
least-square approach and numerical partial differential equation (PDE) approach.
TC19