DECOMP: an Implementation of Dantzig-Wolfe Decomposition for by Professor James K. Ho, Professor Rangaraja P. Sundarraj

By Professor James K. Ho, Professor Rangaraja P. Sundarraj (auth.)

For linear optimization types that may be formulated as linear courses with the block-angular constitution, i.e. self sufficient subproblems with coupling constraints, the Dantzig-Wolfe decomposition precept offers a sublime framework of resolution algorithms in addition to fiscal interpretation. This monograph is the whole documentation of DECOMP: a strong implementation of the Dantzig-Wolfe decomposition process in FORTRAN. The code can function a truly handy place to begin for additional research, either computational and monetary, of parallelism in large-scale platforms. it may well even be used as supplemental fabric in a moment direction in linear programming, computational mathematical programming, or large-scale systems.

Show description

Read Online or Download DECOMP: an Implementation of Dantzig-Wolfe Decomposition for Linear Programming PDF

Best linear books

Mathematik für Ingenieure: Eine anschauliche Einführung für das praxisorientierte Studium (Springer-Lehrbuch) (German Edition)

"Mathematik in entspannter Atmosphäre" ist das Leitbild dieses leicht verständlichen Lehrbuchs. Im Erzählstil und mit vielen Beispielen beleuchtet der Autor nicht nur die Höhere Mathematik, sondern er stellt auch den Lehrstoff in Bezug zu den Anwendungen. Die gesamte für den Ingenieurstudenten wichtige Mathematik wird in einem Band behandelt.

Applications of Lie Algebras to Hyperbolic and Stochastic Differential Equations (Mathematics and Its Applications)

The most a part of the publication is predicated on a one semester graduate path for college students in arithmetic. i've got tried to improve the idea of hyperbolic platforms of differen­ tial equations in a scientific approach, making as a lot use as attainable ofgradient platforms and their algebraic illustration. even if, regardless of the powerful sim­ ilarities among the advance of rules right here and that present in a Lie alge­ bras path this isn't a ebook on Lie algebras.

Linear Operators and Matrices: The Peter Lancaster Anniversary Volume

In September 1998, through the 'International Workshop on research and Vibrat­ ing structures' held in Canmore, Alberta, Canada, it used to be made up our minds by means of a gaggle of individuals to honour Peter Lancaster at the get together of his seventieth birthday with a quantity within the sequence 'Operator idea: Advances and Applications'.

Harmonic Analysis on Exponential Solvable Lie Groups (Springer Monographs in Mathematics)

This ebook is the 1st one who brings jointly contemporary effects at the harmonic research of exponential solvable Lie teams. There nonetheless are many attention-grabbing open difficulties, and the ebook contributes to the longer term development of this learn box. to boot, numerous comparable subject matters are provided to inspire younger researchers.

Additional info for DECOMP: an Implementation of Dantzig-Wolfe Decomposition for Linear Programming

Sample text

2. In the absence of basic equality logicals, to remove as many infeasibilities as possible if there are any remaining infeasibilities. 54 Since the rhs is not necessarily positive, some changes as illustrated in the example below will have to be made to the normal RSM ratio test. 22) -9x1 Here a1 and a2 are equality logicals, x2 to x5 are variables that are not equality logicals and xl is the entering variable. The following observations can be made with this example. 1. 22) cannot be ignored.

C '1' indicates step 2 of D-W Phase 1 or D-W Phase 2. '3' indicates D-W C Phase 3. C MSTAT C DX C C Indicates status of the problem. Subproblem objective value minus the dual variable of the corresponding convexity constraint. DX is the reduced cost of the proposal column. KRIT(*): Indicates the type of the strategy chosen by the user. '2' if intermediate C proposals are allowed. '1' if proposals only at optimality of unboundedness C of the subproblem. C LSUB Indicates problem index. '0' for master problem.

Lines 56 - 59] - Set KFASE to 1 and call FORMC. If the solution is infeasible then C since KFASE is 1, FORMC would already have set up the cost vector in C the buffer YA(*, 5). Otherwise set KF ASE to 2 and call FORMC again C to set up the cost vector in YA(*, 6). [lines 60 - 69] C C - Write the proposal into Z(*, *) and the reduced cost of the proposal column in the (NROWO+ 1)st row of Z(*, *). [lines 70 - 72] 46 C C C C C C column of Z(*, *). [lines 53 - 55] - If KSTR is not 5 and there is space in Z(*, *) then write the current proposal to the ftrst available colurnn of Z(*, *).

Download PDF sample

Rated 4.74 of 5 – based on 28 votes