Piecewise linearisation of the first order loss function for families of arbitrarily distributed random variables

Alejandro G.  Alcoba; Eligius M.T. Hendrix; Roberto Rossi

Piecewise linearisation of the first order loss function for families of arbitrarily distributed random variables

Alejandro G. Alcoba, Eligius M.T. Hendrix, Roberto Rossi

Research output: Contribution to conference › Abstract › peer-review

Abstract

We discuss the problem of computing optimal linearisation parameters for the first order loss function of a family of arbitrarily distributed random variable. We demonstrate that, in contrast to the problem in which parameters must be determined for the loss function of a single random variable, this problem is nonlinear and features several local optima and plateaus. We introduce a simple and yet effective heuristic for determining these parameters and we demonstrate its effectiveness
via a numerical analysis carried out on a well known stochastic lot sizing problem.

Original language	English
Publication status	Published - 2015
Event	27th European Conference on Operational Research (EURO 2015) - Glasgow, United Kingdom Duration: 12 Jul 2015 → 15 Jul 2015

Conference

Conference	27th European Conference on Operational Research (EURO 2015)
Country/Territory	United Kingdom
City	Glasgow
Period	12/07/15 → 15/07/15

27th European Conference on Operational Research (EURO 2015)
Roberto Rossi (Chair and Invited Speaker)
12 Jul 2015 → 15 Jul 2015
Activity: Participating in or organising an event types › Participation in conference

Cite this

@conference{0ab48c0b050c4c168e98254826fab7ba,

title = "Piecewise linearisation of the first order loss function for families of arbitrarily distributed random variables",

abstract = "We discuss the problem of computing optimal linearisation parameters for the first order loss function of a family of arbitrarily distributed random variable. We demonstrate that, in contrast to the problem in which parameters must be determined for the loss function of a single random variable, this problem is nonlinear and features several local optima and plateaus. We introduce a simple and yet effective heuristic for determining these parameters and we demonstrate its effectivenessvia a numerical analysis carried out on a well known stochastic lot sizing problem.",

author = "Alcoba, {Alejandro G.} and Hendrix, {Eligius M.T.} and Roberto Rossi",

year = "2015",

language = "English",

note = "27th European Conference on Operational Research (EURO 2015) ; Conference date: 12-07-2015 Through 15-07-2015",

}

TY - CONF

T1 - Piecewise linearisation of the first order loss function for families of arbitrarily distributed random variables

AU - Alcoba, Alejandro G.

AU - Hendrix, Eligius M.T.

AU - Rossi, Roberto

PY - 2015

Y1 - 2015

N2 - We discuss the problem of computing optimal linearisation parameters for the first order loss function of a family of arbitrarily distributed random variable. We demonstrate that, in contrast to the problem in which parameters must be determined for the loss function of a single random variable, this problem is nonlinear and features several local optima and plateaus. We introduce a simple and yet effective heuristic for determining these parameters and we demonstrate its effectivenessvia a numerical analysis carried out on a well known stochastic lot sizing problem.

AB - We discuss the problem of computing optimal linearisation parameters for the first order loss function of a family of arbitrarily distributed random variable. We demonstrate that, in contrast to the problem in which parameters must be determined for the loss function of a single random variable, this problem is nonlinear and features several local optima and plateaus. We introduce a simple and yet effective heuristic for determining these parameters and we demonstrate its effectivenessvia a numerical analysis carried out on a well known stochastic lot sizing problem.

M3 - Abstract

T2 - 27th European Conference on Operational Research (EURO 2015)

Y2 - 12 July 2015 through 15 July 2015

ER -

Piecewise linearisation of the first order loss function for families of arbitrarily distributed random variables

Abstract

Conference

Fingerprint

Activities

27th European Conference on Operational Research (EURO 2015)

Cite this