Fwd: Workshop in Operations Management

[Margaret Jaffey]

>Subject: Workshop in Operations Management
>MIME-Version: 1.0
>
>
>     Speaker: Dan Adelman
>       Title: Price-Directed Replenishment of Subsets
>Organization: University of Chicago
>        Date: 5 October 2000
>        Time: 11:00 AM  (Note slightly different time)
>       Place: Stuart 216
>
>Abstract:
>The idea of price-directed control is to use an operating policy that
>exploits optimal dual prices from a mathematical programming relaxation of
>the underlying control problem.  We apply it to the problem of
>replenishing inventory to subsets of items (or customers), such as in the
>distribution of industrial gases, so as to minimize the long-run time
>average replenishment cost.
>
>Using marginal values set by management for each item, the dispatcher
>compares the total value of each possible replenishment with its cost, and
>chooses ones that maximize the surplus. We derive this operating policy
>using a linear functional appoximation to the optimal value function of a
>semi-markov decision process.  We also derive a math program whose optimal
>dual prices yield values and whose objective function gives a lower bound
>on system performance.  We use duality theory to show that optimal prices
>satisfy several intuitive properties.  The resulting price-directed policy
>is scalable and achieves optimal or near-optimal performance on many
>problem instances.
>
>Paper available in Stuart 202C
>
>
>Future talks:
>
>     Speaker: Willard Zangwill
>       Title: A Mathematical Model of an Electronic Commerce Firm
>Organization: University of Chicago
>        Date: 12 October
>        Time: 11:00 AM
>       Place: Stuart 216
>
>Abstract
>(joint work with Deb Sarkar)
>
>We present a mathematical optimization model of a business-to-consumer type
>of electronic commerce (EC) firm.  Firms like this have some special
>characteristics.  They tend to grow rapidly and spend so much money to
>achieve that growth that for some period of time they often generate
>financial losses.  Our model analyzes these issues and determines the
>optimal growth rate.
>
>We consider not only the deterministic case but also the stochastic
>situation where the number of customers follows a random walk diffusion
>process.  Interestingly, the results in both situations are the same.  The
>model thus seems stable and robust against this common type of uncertainty.
>
>We also discovered an  "explosion point."  This point signals that the
>profit situation has become highly favorable, and that the firm should then
>grow as rapidly as possible.  The explosion point might help understand why
>some EC firms try to expand so quickly.
>
>The model should help management evaluate major capital investments
>including some acquisitions.  Further, the model should help evaluate the
>progress of the firm to see if the firm is on track towards its goals or not.
>
>In addition to its management implications, the model also provides insight
>into the financial value of an EC firm.
>
>Mathematically, the model is particularly clean and concise.  Despite the
>many factors and complexities, the optimal solution is provided by a simple
>quadratic equation.  This simplicity of solution and cleanliness of
>formulation makes the model easy to apply and extend.
>
>We believe that this is the first mathematical model to analyze the special
>characteristics of e-commerce firms.  The model's major results are
>obtained using a special four step methodology.  This methodology is
>broadly applicable and promises to provide insights not just into EC firms,
>but also into the firms of almost any industry.
>
>
>
>     Speaker: Jeff Camm
>       Title: Conjoint Optimization
>Organization: University of Cincinnati
>        Date: 26 October
>        Time: 11:00 AM
>       Place: Stuart 216
>
>
> Abstract
>
> Products and Services can be viewed as bundles of attributes that are
> made up of many levels.  Customers make trade-offs between various
> attributes and levels before making a purchase decision for a given
> product.  A popular marketing research technique called conjoint
> analysis is used to study these trade-offs and determine a judicious
> combination of attribute levels that is likely to perform well in a
> market containing competitors' products.
>
> The data on customer preferences is decomposed using conjoint analysis
> into utility values that the customer attaches to each level of each
> attribute.  In order to develop a successful product, researchers have
> proposed the share-of-choices problem.  The objective is to maximize the
> number of respondents (customers) in a study who prefer a potential new
> product to the status-quo product that is already in the market.
>
> The share-of-choices problem is NP-hard and general LP-based Branch and
> Bound algorithms are not effective in solving realistic sized problems in
> reasonable amounts of time.  A few heuristic procedures appear in the
> literature.  We develop an optimal backtracking algorithm strengthened
> by lagrangian and logic-based bounds to solve the share-of-choices
> problem.
>
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Margaret P. Jaffey			margaret at cs.uchicago.edu
Department of Computer Science
Student Support Rep (Ry 161A)		(773) 702-6011
The University of Chicago		http://www.cs.uchicago.edu
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=