25.11.2015 Views

The MOSEK Python optimizer API manual Version 7.0 (Revision 141)

Optimizer API for Python - Documentation - Mosek

Optimizer API for Python - Documentation - Mosek

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

616 APPENDIX F. <strong>MOSEK</strong> FILE FORMATS<br />

* 1 2 3 4 5 6<br />

*23456789012345678901234567890123456789012345678901234567890<br />

NAME<br />

[name]<br />

?? [vname1] [value1]<br />

ENDATA<br />

Figure F.1: <strong>The</strong> standard ORD format.<br />

<strong>The</strong> format is based on the TAR (USTar) file format. This means that the individual pieces of data<br />

in a .task file can be examined by unpacking it as a TAR file. Please note that the inverse may not<br />

work: Creating a file using TAR will most probably not create a valid <strong>MOSEK</strong> Task file since the<br />

order of the entries is important.<br />

F.5 <strong>The</strong> XML (OSiL) format<br />

<strong>MOSEK</strong> can write data in the standard OSiL xml format. For a definition of the OSiL format please<br />

see http://www.optimizationservices.org/. Only linear constraints (possibly with integer variables) are<br />

supported. By default output files with the extension .xml are written in the OSiL format.<br />

<strong>The</strong> parameter iparam.write xml mode controls if the linear coefficients in the A matrix are written<br />

in row or column order.<br />

F.6 <strong>The</strong> ORD file format<br />

An ORD formatted file specifies in which order the mixed integer <strong>optimizer</strong> branches on variables. <strong>The</strong><br />

format of an ORD file is shown in Figure F.1. In the figure names in capitals are keywords of the ORD<br />

format, whereas names in brackets are custom names or values. <strong>The</strong> ?? is an optional key specifying<br />

the preferred branching direction. <strong>The</strong> possible keys are DN and UP which indicate that down or up<br />

is the preferred branching direction respectively. <strong>The</strong> branching direction key is optional and is left<br />

blank the mixed integer <strong>optimizer</strong> will decide whether to branch up or down.<br />

F.6.1<br />

An example<br />

A concrete example of a ORD file is presented below:<br />

NAME<br />

EXAMPLE<br />

DN x1 2<br />

UP x2 1<br />

x3 10<br />

ENDATA<br />

This implies that the priorities 2, 1, and 10 are assigned to variable x1, x2, and x3 respectively. <strong>The</strong><br />

higher the priority value assigned to a variable the earlier the mixed integer <strong>optimizer</strong> will branch on<br />

that variable. <strong>The</strong> key DN implies that the mixed integer <strong>optimizer</strong> first will branch down on variable<br />

whereas the key UP implies that the mixed integer <strong>optimizer</strong> will first branch up on a variable.<br />

If no branch direction is specified for a variable then the mixed integer <strong>optimizer</strong> will automatically

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!