[Os-project-managers] Question about OSoL and OSrL

[Kipp Martin]

Hi Jun:

> Let's clarify first: the options in the OSoL file is *really* options to 
> be used in optimization; they are NOT descriptions about the option file.
> So obviously, the three suggested elements, by the fact that they are 
> "borrowed" from OSiL, are descriptions of what the OSoL file is.
> Is it necessary? Or is it that important?
> The reason that they are in OSiL is because we think they are necessary 
> given the instance is usually quite important. But we didn't even put 
> <author> there, because <source> is just general enough and
> we don't want to over engineer it with too heavy a structure. After all 
> they are mainly for archiving and reporting purposes, and not for the 
> optimization purposes.

I think an option file can have an "identity." For example, we might be 
running experiments where we want to test the performance of an 
algorithm given different sets of options and starting values. In other 
words I want to test my algorithm on, for example, 10 different option 
files. In this case I would want to clearly identify each option file 
with a <name> and <description>.  In this case my <description> really 
is about the option file.  For example

<description>test with loose tolerances</description>


<description>test with tight tolerances</description>

Then in the solver options the various tolerances differ. So indeed I 
want to describe my file.

We might also want to link an option file with an osil file. We could do 
this with  <description> or <source>.

Cheers


> 
> Jun
> 
> --------------------------------------------------
> From: "Horand Gassmann" <Horand.Gassmann at dal.ca>
> Sent: Monday, January 03, 2011 2:11 PM
> To: <os-project-managers at list.coin-or.org>
> Subject: Re: [Os-project-managers] Question about OSoL and OSrL
> 
>> Kipp Martin <kmartin at chicagobooth.edu> wrote:
>>
>>> Hi Gus:
>>>>
>>>> I know that I am making more work for myself, but it occurred to me
>>>> that the OSoL (and OSrL) schemas are lacking the file information we
>>>> provide in the OSiL files in form of the <instanceHeader>. Trying to
>>>> debug the OSoL parser and the unit test I am forced to look at old
>>>> osol files that I wrote to test the osol parser that existed at that
>>>> time, and it would now be useful to have had some information as to
>>>> what I was thinking at the time.
>>>>
>>>> I propose to add the following optional information:
>>>>
>>>> <optionHeader>
>>>>      <name>
>>>>      <author>
>>>>      <source>
>>>>      <description>
>>>>
>>>> (and similarly for <resultHeader>). I also propose to add <author> to
>>>> the <instanceHeader> in OSiL.
>>>>
>>>> Since this is optional stuff, I hope we can treat this as a consent
>>>> agenda item.
>>>>
>>>> Any thoughts?
>>>
>>> I consent and think it is a good idea. But where would it go? Would
>>> <optionHeader> be a child of <general> or would it go before <general>?
>>
>> My thinking is that it should go before <general>, just for better
>> visibility. But I do not feel strongly about that.
>>
>> Cheers
>>
>> gus
>>
>> _______________________________________________
>> Os-project-managers mailing list
>> Os-project-managers at list.coin-or.org
>> http://list.coin-or.org/mailman/listinfo/os-project-managers
>>


-- 
Kipp Martin
Professor of Operations Research
and Computing Technology
Booth School of Business
University of Chicago
5807 South Woodlawn Avenue
Chicago, IL 60637
773-702-7456
kmartin at chicagobooth.edu
http://www.chicagobooth.edu/faculty/bio.aspx?person_id=12825325568
http://projects.coin-or.org/OS