[#2293] Equality constraints

Date:
2012-10-02 21:33 Priority:
2

State:
Open

Submitted by:
Arne Henningsen (arne) Assigned to:
Nobody (None)

Hardware:
All Product:
None

Operating System:
All Component:
None

Version:
None Severity:
minor

Resolution:
Accepted As Bug

URL:

Summary:
Equality constraints

Detailed description
In some (many?) circumstances, solveLP() does not correctly work if the model includes equality constraints. This is demonstrated with the following simple example: min x1 + x2, s.t. x1 + 0.5 * x2 = 2 > cvec <- c( 1, 1 ) > Amat <- matrix( c( 1, 0.5 ), nrow = 1 ) > bvec <- 2 > aa=solveLP( cvec, bvec, Amat, const.dir = "=" ) > aa$solution 1 2 0 0 Function lp(), which is available in the R package "lpSolve" and can be called through solveLP(), can cope with equality constraints: > aa=solveLP( cvec, bvec, Amat, const.dir = "=", lpSolve = TRUE ) > aa$solution 1 2 2 0 I am grateful to Chun He for making me aware of this problem.

Comments:

Message
Date: 2012-10-16 19:22 Sender: Arne Henningsen Another example: cvec <- c(27,9) bvec <- c(8,74) Amat <- matrix(c(1,1,-1,1),nrow=2) solveLP(cvec, bvec, Amat, maximum=TRUE, const.dir=c("==","<=")) solveLP() returns the solution 74 and 0 but the correct solution is 41 and 33. The correct solution can be obtained by function lp(), which is available in the R package "lpSolve" and can be called through solveLP() by setting argument "lpSolve" to "TRUE". solveLP(cvec, bvec, Amat, maximum=TRUE, const.dir=c("==","<="), lpSolve=TRUE) I am grateful to Ulrike Grömping for reporting this bug and for providing this example.

No Changes Have Been Made to This Item