|
|
|
|
|
|
Lijian Chen (陈力简)
Home of novel methods for stochastic
programming |
|
5241 Craigs Creek Dr.
Louisville, KY 40241
Spam free email: lijian.chen@outlook.com
Office: 502-852-2197
    PHP Web Counter Cell: 502-298-4672
|
|
|
I am interested in developing novel,
largely computational methods, to solve the large scale stochastic
programming. These novel methods become increasingly implementable because of
two facts: first, the exponential growth rates of computational capabilities
in the last three decades described by the Moore’s
law; second, all of my novel methods have polynomial computational
complexities. Our contribution is to incorporate the
recently emerged, efficient optimization techniques to the field of
stochastic programming. The state-of-the-art research of stochastic
programming is largely about the statistical properties of the approximation,
such as the convergence with probability one or the large deviation theory.
Despite a few impressive applications, the numerical performances of
difficult problems, e.g., the chance constrained optimization, the
large-scale two stage recourse problem, and the multi-stage stochastic
programming, are still up in the air. It is fair to say that the currently
solver can only solve low-dimensional, highly simplified problems. On the
other hand, the optimization techniques in general, have been fast improved
in the past two decades, thanks to the development and implementation of the
interior-point methods. The marriage between stochastic programming and
efficient optimization techniques, has not yet, but surely will be
established. I sincerely hope that my novel methods will play an important
role for this exciting liaison. All my novel methods decompose the large
scale stochastic programming into many but a manageable number of auxiliary
convex problems without compromising the solution quality. The implementation
of our methods will endorse a distributed computational infrastructure which
will lead to considerable savings and timely solutions. Our methods have been
used in the following applied research topics toward gaining scientific
knowledge to meet a recognized need:
|
|
|
|
|
|
|
|
|
|
