A decomposition method for quadratic zero-one programming
Article Abstract:
The unconstrained zero-one quadratic programming problem is considered. A decomposition procedure is introduced for calculating a lower bound for this quadratic function. It is demonstrated that any quadratic function can be modelled as a sum of specific functions whose minima can be derived using a simple branch and bound algorithm. It is also shown that the best among all possible decompositions can be formulated using a Lagrangian decomposition approach. The algorithm is compared with Pardalos and Rodgers' algorithm, and is found to be extremely efficient, particularly for hard problem cases. The proposed method for computing a lower bound is proven to work very well for problem sizes up to 100 variables. Gaps are virtually nonexistent for problem sizes with 50 variables, but their incidence increases with the size of problems.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1995
User Contributions:
Comment about this article or add new information about this topic:
Holt-Winters method with missing observations
Article Abstract:
A simple method is introduced for interpolating, smoothing and predicting in time series analysis with missing observations or irregularly observed data. This approach involves extending a classical smoothing approach, the Holt-Winters method, for seasonal data with missing observations. Extending the simple exponential smoothing method and Holt's method with nonseasonal data issued at variable time inverval was first suggested by D. J. Wright in a 1986 paper. The extension of a classical smoothing approach to the Holt-Winters method proves to be an effective procedure for interpolating, smoothing and predicting seasonal time series even when only half of the needed data is observed. Numerical examples are discussed to show the use of the method.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1995
User Contributions:
Comment about this article or add new information about this topic:
Revising Forecats of Accounting Earnings: A Ccomparison with the Box-Jenkins Method
Article Abstract:
Forecasting methodologies are compared. Past annual earnings data were used to forecast future earnings. Theil's correction technique to improve the accuracy of Box-Jenkins was evaluated. Simple extrapolative forecasting models were compared with the Box-Jenkins method. Results are included in a table.
Publication Name: Management Science
Subject: Business, general
ISSN: 0025-1909
Year: 1983
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: CEO compensation and corporate performance. Implementing quality and re-engineering programmes: managing the people issues
- Abstracts: Flow management in flexible manufacturing cells with pipeline operations. Manufacturing Technology in the 1980's: A Survey of Federal Programs and Practices
- Abstracts: Decomposition and nondifferentiable optimization with the projective algorithm. Monitoring An Input-Output Model for Production: I. The Control Charts
- Abstracts: Fleet sizing and empty equipment redistribution for center-terminal transportation networks. Learning Effects in Economic Lot Sizing