Linear-fractional programming: Difference between revisions

Revision as of 17:21, 1 March 2010

Latest revision as of 11:52, 20 August 2025

(134 intermediate revisions by 49 users not shown)

Line 1:

{{Short description|Concept in mathematical optimization}}

In [[applied mathematics]], '''linear-fractional programming (LFP)''' formally is almost the same as [[linear programming]] (LP) but instead of linear [[objective function]] one has a ratio of two linear functions, subject to linear constraints. Informally, if linear programming determines the way to achieve the best outcome (such as maximum profit or lowest cost) in a given mathematical model and given some list of requirements represented as linear equations, in linear-fractional programming model we can achieve the best (highest) ratio of outcome to cost, i.e. highest efficiency.

In [[mathematical optimization]], '''linear-fractional programming''' ('''LFP''') is a generalization of [[linear programming]] (LP). Whereas the objective function in a linear program is a [[linear functional|linear function]], the objective function in a linear-fractional program is a ratio of two linear functions. A linear program can be regarded as a special case of a linear-fractional program in which the denominator is the constant function 1.

Formally, a linear-fractional program is defined as the problem of maximizing (or minimizing) a ratio of [[affine function]]s over a [[polyhedron]],

For example, if in the frame of LP we maximize '''profit = income − cost''' and obtain maximal profit of 100 units (= $1100 of income − 1000$ of cost), then using LFP we can obtain only $10 of profit which requires only $50 of investment. Thus, in LP we have efficiency $100/$1000 = 0.1, at the same time LFP provides efficiency equal to $10/$50 = 0.2.

:<math>

\begin{align}

\text{maximize} \quad & \frac{\mathbf{c}^T \mathbf{x} + \alpha}{\mathbf{d}^T \mathbf{x} + \beta} \\

\text{subject to} \quad & A\mathbf{x} \leq \mathbf{b},

\end{align}

</math>

where <math>\mathbf{x} \in \mathbb{R}^n</math> represents the vector of variables to be determined, <math>\mathbf{c}, \mathbf{d} \in \mathbb{R}^n</math> and <math>\mathbf{b} \in \mathbb{R}^m</math> are vectors of (known) coefficients, <math>A \in \mathbb{R}^{m \times n}</math> is a (known) matrix of coefficients and <math>\alpha, \beta \in \mathbb{R}</math> are constants. The constraints have to restrict the [[feasible region]] to <math>\{\mathbf{x} | \mathbf{d}^T\mathbf{x} + \beta > 0\}</math>, i.e. the region on which the denominator is positive.<ref name="CC">{{cite journal |last1=Charnes |first1=A. |last2=Cooper |first2=W. W. |author2-link=William W. Cooper |year=1962 |title=Programming with Linear Fractional Functionals |journal=Naval Research Logistics Quarterly |volume=9 |issue=3–4 |pages=181–186 |doi=10.1002/nav.3800090303 |mr=152370}}</ref><ref name="BV">{{cite book |last1=Boyd |first1=Stephen P. |url=https://web.stanford.edu/~boyd/cvxbook/bv_cvxbook.pdf |title=Convex Optimization |last2=Vandenberghe |first2=Lieven |publisher=Cambridge University Press |year=2004 |isbn=978-0-521-83378-3 |page=151 |access-date=October 15, 2011}}</ref> Alternatively, the denominator of the objective function has to be strictly negative in the entire feasible region.

==Motivation by comparison to linear programming==

Linear-fractional programming can be used in the same real-world applications as LP, in various fields of study. Most extensively it is used in business and economic situations, especially in the situations of deficit of financial resources. Also LFP can be utilized for wide range of engineering problems. Some industries that use linear programming models including transportation, energy, telecommunications, and manufacturing, all of which may use LPF as well as LP.

Both linear programming and linear-fractional programming represent optimization problems using linear equations and linear inequalities, which for each problem-instance define a [[feasible set]]. Fractional linear programs have a richer set of objective functions. Informally, linear programming computes a policy delivering the best outcome, such as maximum profit or lowest cost. In contrast, a linear-fractional programming is used to achieve the highest ''ratio'' of outcome to cost, the ratio representing the highest efficiency. For example, in the context of LP we maximize the objective function '''profit&nbsp;=&nbsp;income&nbsp;&minus;&nbsp;cost''' and might obtain maximum profit of $100 (=&nbsp;$1100&nbsp;of&nbsp;income&nbsp;&minus;&nbsp;$1000 of cost). Thus, in LP we have an efficiency of $100/$1000&nbsp;=&nbsp;0.1. Using LFP we might obtain an efficiency of $10/$50&nbsp;=&nbsp;0.2 with a profit of only $10, but only requiring $50 of investment.

==Transformation to a linear program==

== References ==

Any linear-fractional program can be transformed into a linear program, assuming that the feasible region is non-empty and bounded, using the '''Charnes–Cooper transformation'''.<ref name="CC" /> The main idea is to introduce a new non-negative variable <math>t </math> to the program which will be used to rescale the constants involved in the program (<math>\alpha, \beta, \mathbf{b}</math>). This allows us to require that the denominator of the objective function (<math>\mathbf{d}^T \mathbf{x} + \beta</math>) equals 1. (To understand the transformation, it is instructive to consider the simpler special case with <math>\alpha = \beta = 0</math>.)

* E.Bajalinov, ''Linear-Fractional Programming: Theory, Methods, Applications and Software''. «Kluwer Academic Publishers», 2003.

Formally, the linear program obtained via the Charnes–Cooper transformation uses the transformed variables <math>\mathbf{y} \in \mathbb{R}^n</math> and <math>t \ge 0 </math>:

== Software ==

* [http://zeus.nyf.hu/~bajalinov/WinGulf/wingulf.html WinGULF] &ndash; educational interactive linear and linear-fractional programming solver with a lot of special options (pivoting, pricing, branching variables etc.).

:<math>

[[Category:Mathematical optimization]]

\begin{align}

[[Category:Operations research]]

\text{maximize} \quad & \mathbf{c}^T \mathbf{y} + \alpha t \\

\text{subject to} \quad & A\mathbf{y} \leq \mathbf{b} t \\

& \mathbf{d}^T \mathbf{y} + \beta t = 1 \\

& t \geq 0.

\end{align}

</math>

A solution <math>\mathbf{x}</math> to the original linear-fractional program can be translated to a solution of the transformed linear program via the equalities

[[ru:Дробно-линейное программирование]]

:<math>\mathbf{y} = \frac{1}{\mathbf{d}^T \mathbf{x} + \beta} \cdot \mathbf{x}\quad \text{and} \quad t = \frac{1}{\mathbf{d}^T \mathbf{x} + \beta}.</math>

[[uk:Задача дробово-лінійного програмування]]

Conversely, a solution for <math>\mathbf{y}</math> and <math>t </math> of the transformed linear program can be translated to a solution of the original linear-fractional program via

:<math>\mathbf{x}=\frac{1}{t}\mathbf{y}.</math>

==Duality==

Let the [[duality (optimization)|dual variables]] associated with the constraints <math>A\mathbf{y} - \mathbf{b} t \leq \mathbf{0}</math> and <math>\mathbf{d}^T \mathbf{y} + \beta t - 1 = 0</math> be denoted by <math>\mathbf{u}</math> and <math>\lambda</math>, respectively. Then the dual of the LFP above is <ref>{{cite journal|last1=Schaible |first1=Siegfried |title=Parameter-free Convex Equivalent and Dual Programs|journal=Zeitschrift für Operations Research |volume=18 |year=1974 |issue=5 |pages=187–196|doi=10.1007/BF02026600|mr=351464|s2cid=28885670 }}</ref><ref>{{cite journal|title=Fractional programming&nbsp;I: Duality |last1=Schaible |first1=Siegfried | journal=Management Science |volume=22 |issue=8 |pages=858–867 |year=1976|jstor=2630017|mr=421679|doi=10.1287/mnsc.22.8.858}}</ref>

:<math>

\begin{align}

\text{minimize} \quad & \lambda \\

\text{subject to} \quad & A^T\mathbf{u} + \lambda \mathbf{d} = \mathbf{c} \\

& -\mathbf{b}^T \mathbf{u} + \lambda \beta \geq \alpha \\

& \mathbf{u} \in \mathbb{R}_+^m, \lambda \in \mathbb{R},

\end{align}

</math>

which is an LP and which coincides with the dual of the equivalent linear program resulting from the Charnes–Cooper transformation.

==Properties and algorithms==

The objective function in a linear-fractional problem is both quasiconcave and [[quasiconvex function|quasiconvex]] (hence quasilinear) with a [[monotonicity|monotone]] property, [[pseudoconvex function|pseudoconvexity]], which is a stronger property than [[quasiconvex function|quasiconvexity]]. A linear-fractional objective function is both pseudoconvex and pseudoconcave, hence [[pseudolinear function|pseudolinear]]. Since an LFP can be transformed to an LP, it can be solved using any LP solution method, such as the [[simplex algorithm]] (of [[George B. Dantzig]]),<ref>

Chapter five: {{cite book| last=Craven|first=B. D.|title=Fractional programming|series=Sigma Series in Applied Mathematics|volume=4|publisher=Heldermann Verlag|___location=Berlin|year=1988|pages=145|isbn=978-3-88538-404-5 |mr=949209}}</ref><ref>{{cite journal| last1=Kruk | first1=Serge|last2=Wolkowicz|first2=Henry|title=Pseudolinear programming |journal=[[SIAM Review]]|volume=41 |year=1999 |issue=4 |pages=795–805 |mr=1723002|jstor=2653207|doi=10.1137/S0036144598335259| bibcode=1999SIAMR..41..795K|citeseerx=10.1.1.53.7355}}

</ref><ref>{{cite journal | last1=Mathis|first1=Frank H.|last2=Mathis|first2=Lenora Jane|title=A nonlinear programming algorithm for hospital management |journal=[[SIAM Review]]|volume=37 |year=1995 |issue=2 |pages=230–234|mr=1343214|jstor=2132826|doi=10.1137/1037046|s2cid=120626738 }}

</ref><ref>{{harvtxt|Murty|1983|loc=Chapter&nbsp;3.20 (pp.&nbsp;160–164) and pp.&nbsp;168 and&nbsp;179}}</ref> the [[criss-cross algorithm]],<ref>{{cite journal|title=The finite criss-cross method for hyperbolic programming|journal=European Journal of Operational Research|volume=114|issue=1|

pages=198–214|year=1999 <!-- issn=0377-2217 -->|doi=10.1016/S0377-2217(98)00049-6|first1=Tibor|last1=Illés|first2=Ákos|last2=Szirmai|first3=Tamás|last3=Terlaky|zbl=0953.90055|id=[http://www.cas.mcmaster.ca/~terlaky/files/dut-twi-96-103.ps.gz Postscript preprint]|citeseerx=10.1.1.36.7090}}</ref> or [[interior-point method]]s.

==Notes==

<references />

==Sources==

*{{cite book|last=Murty|first=Katta&nbsp;G.|author-link=Katta G. Murty|chapter=3.10 Fractional programming (pp. 160–164)|title=Linear programming|publisher=John Wiley & Sons, Inc.|___location=New York|year=1983|pages=xix+482|isbn=978-0-471-09725-9|mr=720547}}

==Further reading==

*{{cite book|first=E.Bajalinov, ''B.|last=Bajalinov|title=Linear-Fractional Programming: Theory, Methods, Applications and Software''.| «publisher=Kluwer Academic Publishers», |___location=Boston|year=2003.}}

*{{cite book|last=Barros|first=Ana Isabel|title=Discrete and fractional programming techniques for ___location models|series=Combinatorial Optimization|volume=3|publisher=Kluwer Academic Publishers|___location=Dordrecht|year=1998|pages=xviii+178|isbn=978-0-7923-5002-6|mr=1626973}}

*{{cite book|last=Martos|first=Béla|title=Nonlinear programming: Theory and methods|publisher=North-Holland Publishing Co.|___location=Amsterdam-Oxford|year=1975|pages=279|isbn=978-0-7204-2817-9|mr=496692}}

*{{cite book|last=Schaible|first=S.|chapter=Fractional programming|pages=495–608|mr=1377091|title=Handbook of global optimization|editor=Reiner Horst and Panos M. Pardalos|

series=Nonconvex optimization and its applications|volume=2|publisher=Kluwer Academic Publishers|___location=Dordrecht|year=1995|isbn=978-0-7923-3120-9}}

*{{cite book | last=Stancu-Minasian | first=I. M.| title=Fractional programming: Theory, methods and applications | others=Translated by Victor Giurgiutiu from the 1992 Romanian | series=Mathematics and its applications|volume=409|publisher=Kluwer Academic Publishers Group | ___location=Dordrecht | year=1997 | pages=viii+418 | isbn=978-0-7923-4580-0 | mr=1472981 }}

{{DEFAULTSORT:Linear-Fractional Programming}}

[[Category:Optimization algorithms and methods]]

[[Category:Linear programming]]

[[Category:OperationsGeneralized researchconvexity]]