 |
Leo's Page
Thank you for taking the time to check out my corner of the
World Wide Web. My name is Leonardo Lopes, but everyone knows me as
Leo. I am an Assistant Professor in the Systems and Industrial
Engineering Department at the University of Arizona. I am currently
working on modeling and computational aspects of Stochastic
Optimization. On these pages you will find more detailed information
about my
activities. If you find something of interest please contact me.
For appointment
availability, check my public calendar
|
Ph.D., Industrial Engineering and Management Sciences - Northwestern University, 2003
Dissertation Topic:Modeling and Communication of Stochastic Programs
Mature Research Initiatives
Papers related to this research have already been submitted or published. In some cases, software has also been released. Students interested in extending these topics would have well defined paths to graduation.
XML
and Communication of Optimization Problems: Modern interfaces
for
communicating instances of optimization problems.
Modeling of Stochastic
Programming: An Integrated Development Environment for
modeling, solving, and analysing Stochastic Programming Problems.
Evolving Research Initiatives
This research may be at several levels of development. In some cases, all proofs have been written, but no computational experiments have been performed. In others, grants have been written but most of the research has not been done. In others, the problem has been defined, but the exact scope of each project is not yet determined. Students interested in extending these topics will have extraordinary freedom to determine their own research directions.
Large-scale
distributed Solvers
Most practical optimization algorithms developed over the last
50
years use duality theory extensively. However, obtaining dual
information requires the management of large quantities of
dynamic
data. Communicating these data is a major challenge in computing
architectures comprised of collections of workstations. The challenge
addressed by this research is to find a balance between the usage of
duality theory and brute force that is more adequate for the
distributed workstation architecture than the current state of the art.
The goal of this research is to support applications of wide
distributed decision-making, such as traffic management.
Orthopaedic
applications of SP
Current MP-based models for biomechanical parameters of the human body
are not always accurate. The challenge addressed by thisresearch is to
investigate recourse-based formulations for biomechanical models of
human joints and produce an MP-based modeling framework that is more
accurate than the current state of the art under a wider
variety of conditions.
Modeling
Innovation Risk
Many risks are associated with innovation. These risks are difficult to
Quantify. The challenge of this research is to apply engineering
principles to allow businesses and insurance companies to better
strategize and manage innovation.
Scenario
Generation
Stochastic programs have two important components: a structural, static
component; and a dynamic component. The static component describes what
decisions are available at any point in time. Typically, a collection
of Mathematical Programs serves this function. The dynamic component
describes the uncertainty involved in the decision-making. Random
Variables are used for the purpose. For this research, we will focus on
the dynamic components. The challenge addressed by this research is to
help analysts determine which parameters to treat as uncertain, which
distributions to use, how to discretize the distributions and how to
transport the data generated into solvers.
Stochastic
Optimization IT infrastructure
Modern mathematical programming environments use diverse tools and
computing resources. The challenge addressed by this research is to
invent a computational framework, including exchange mechanisms,
libraries, and protocols to enable the development of ever more
powerful, easier to use, and cheaper to deploy mathematicalprogramming
tools.
HUGS and the
Job Shop Problem: Using Human Guided Search to develop
solutions to the Job Shop Problem.
| Instructor
and New Course Development, SIE 375, Software
Integration Tools for Decision Modeling |
| Instructor
and Course
Preparation, SIE 473/573, Concepts
in Information and Communication Systems |
| Instructor
and Course
Preparation, SIE 340, Deterministic
Operations Research |
| Instructor
and Course
Preparation, SIE 547, Computational
Issues in Optimization |
| Instructor, IEMS 201 Introduction to
Statistics, Northwestern University. |
| Graduate Fellow, Ayers Residential College of
Commerce and Industry, Northwestern University. |
Power
of Two Policies A Java illustration of Power of Two policies for
the
Single Item Multi-Retailer problem.
Project
Mapper - A tool for automatic generation of Linear Programming
Problems modeling network flow problems using graphs (undergraduate
sponsored research project).
Professional Experience
Since August of 2004, I am an Assistant Professor at the Department of Systems and Industrial Engineering at the University of Arizona in Tucson. I am teaching, learning about writing grants from some of the best in the business, and getting involved with some very interesting research problems, in addition to working on my own research initiatives. It has been a very busy but very rewarding experience.
From August of 2003 to August of 2004 I was a Visiting Assistant Professor at the Department of Systems and Industrial Engineering at the University of Arizona.From 1997 to 2003, I was a research assistant at Northwestern University under different grants. I worked briefly on an integrated Location, Transportation and Inventory management system. From 1998 to 2003 I worked on optimization modeling systems.
In the summer of 1997, I worked for the NEOS project. I helped build a driver for the solver KNITRO, using Automatic Differentiation to produce derivatives. I also created a script that allowed modelers to use derivatives obtained from Maple in their NEOS submissions.
In the summer of 2000, I had an intership at Cambridge Research Laboratories, formerly MERL, in which we developed a Job-shop problem application based on HUGS, an optimization approach for highly combinatorial problems.
I was a System Administrator at the Departament of Mathematics at UFC. We managed the connection of the department to the internet and installed and maintain UNIX, Windows and Windows95 workstations. Our UNIX flavors were Linux, AIX and Solaris. Our team took care in three shifts of four labs with over 150 users, also offering trainging for the users and the less experienced administrators.
I also designed and installed the Network infrastructure including internet connection for PCC Informática, as described in the article I published in the February '98 issue of Linux Gazette.
From April to September 1997 I was a developer at Townsend Analytics, a real-time trading software company. While there, I wrote software that dealt with clearing, and I automated the company-wide build process.
