# dsdp: Solve semidefinite programm with DSDP In Rdsdp: R Interface to DSDP Semidefinite Programming Library

## Description

Interface to DSDP semidefinite programming library.

## Usage

 1 dsdp(A,b,C,K,OPTIONS=NULL) 

## Arguments

 A An object of class "matrix" with m rows defining the block diagonal constraint matrices A_i. Each constraint matrix A_i is specified by a row of A as explained in the Details section. b A numeric vector of length m containg the right hand side of the constraints. C An object of class "matrix" with one row or a valid class from the class hierarchy in the "Matrix" package. It defines the objective coefficient matrix C with the same structure of A as explained above. K Describes the sizes of each block of the sdp problem. It is a list with the following elements: "s":A vector of integers listing the dimension of positive semidefinite cone blocks. "l":A scaler integer indicating the dimension of the linear nonnegative cone block. OPTIONS A list of OPTIONS parameters passed to dsdp. It may contain any of the following fields:
print:

= k to display output at each k iteration, else = 0 [default 10].

logsummary:

= 1 print timing information if set to 1.

save:

to set the filename to save solution file in SDPA format.

outputstats:

= 1 to output full information about the solution statistics in STATS.

gaptol:

tolerance for duality gap as a fraction of the value of the objective functions [default 1e-6].

maxit:

maximum number of iterations allowed [default 1000].

## Details

All problem matrices are assumed to be of block diagonal structure, the input matrix A must be specified as follows:

## Examples

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19  K=NULL K$s=c(2,3) K$l=2 C=matrix(c(0,0,2,1,1,2,c(3,0,1, 0,2,0, 1,0,3)),1,15,byrow=TRUE) A=matrix(c(0,1,0,0,0,0,c(3,0,1, 0,4,0, 1,0,5), 1,0,3,1,1,3,rep(0,9)), 2,15,byrow=TRUE) b <- c(1,2) OPTIONS=NULL OPTIONS$gaptol=0.000001 OPTIONS$logsummary=0 OPTIONS\$outputstats=1 result = dsdp(A,b,C,K,OPTIONS) 

Rdsdp documentation built on May 30, 2017, 7:21 a.m.