Multiple solutions using ResNet?

[John Simaika]

Hello again,

In a paper by Kelley et al. (2002), one of the authors, Sarkar, mentions that he generated 32 distinct (and valid) solutions from a dataset (last paragraph, page 305). Unfortunately the authors do not provide more infromation on how these solutions were obtained, and neither is the work cited in the references. 

My question to the list is, whether it is possible to generate multiple solutions keeping the parameters and dataset the same? In other words, would ResNet create distintly different reserve networks? Obviously this depends on the dataset, but given that it is suffciently large, is it possible that ties could be broken in favour of one cell than another, in different runs?

Reference

Kelley, C., Garson, J., Aggarwal A., and Sarkar S. (2002) Place prioritization for biodiversity reserve network design: a comparison of SITES and ResNet software packages for coverage and efficiency. Diversity and Distributions (8) 297-306.

[Trevon Fuller]

John,

 

We typically generate multiple solutions from a ResNet input file by constructing multiple randomizations of the input file and  then running ResNet on each randomized input file. The randomization of the input file consists of reordering the rows of the file at random without changing the contents of any row. Recall that each row of the ResNet input file represents one cell. If the rows are reordered, then the cells are encountered in a different order. This may result in the selection of different cells when ties with respect to rarity and complementarity are broken based on lexical order. A reference that explains this is

Sarkar, S., Moffett, A., Sierra, R., Fuller, T., Cameron, S., and Garson, J. 2004. "Incorporating Multiple Criteria into the Design of Conservation Area Networks," Endangered Species Update 21: 100 - 107.

This article is available by going to http://uts.cc.utexas.edu/~consbio/Cons/Labframeset.html and clicking the “Publications” link. The discussion of multiple solutions is the first paragraph on p. 107. Thank you for your continued interest in ResNet.

 

Trevon

[John Simaika]

Hi Trevon,

I would like to try the randomizations on my dataset to see where it well get me. Is there a way to reorder the cells in the input file and attributes (without doing so manually!)?

Thanks for the link and information.

John.

[Trevon Fuller]

John,

I uploaded a zip to the “Files” section of the Google Group called
Randomize_ResNet_input. The zip contains a C++ program to randomize
the rows of a ResNet input file as well as documentation for the
program.

Trevon

On Aug 15, 7:42 am, "John Simaika" <simai...@gmail.com> wrote:
> Hi Trevon,
>
> I would like to try the randomizations on my dataset to see where it well
> get me. Is there a way to reorder the cells in the input file and attributes
> (without doing so manually!)?
>
> Thanks for the link and information.
>
> John.
>

> On Thu, Aug 14, 2008 at 11:45 AM, Trevon Fuller <tful...@mail.utexas.edu>wrote:
>
>
>
> >  John,
>
> > We typically generate multiple solutions from a ResNet input file by
> > constructing multiple randomizations of the input file and  then running
> > ResNet on each randomized input file. The randomization of the input file
> > consists of reordering the rows of the file at random without changing the
> > contents of any row. Recall that each row of the ResNet input file
> > represents one cell. If the rows are reordered, then the cells are
> > encountered in a different order. This may result in the selection of
> > different cells when ties with respect to rarity and complementarity are
> > broken based on lexical order. A reference that explains this is
>
> > Sarkar, S., Moffett, A., Sierra, R., Fuller, T., Cameron, S., and Garson,
> > J. 2004. "Incorporating Multiple Criteria into the Design of Conservation

> > Area Networks," *Endangered Species Update* *21*: 100 - 107.

>
> > This article is available by going to

> >http://uts.cc.utexas.edu/~consbio/Cons/Labframeset.html<http://uts.cc.utexas.edu/%7Econsbio/Cons/Labframeset.html>and clicking the "Publications" link. The discussion of multiple solutions

> > is the first paragraph on p. 107. Thank you for your continued interest in
> > ResNet.
>
> > Trevon
>

> > *From:* bbcl_s...@googlegroups.com [mailto:
> > bbcl_s...@googlegroups.com] *On Behalf Of *John Simaika
> > *Sent:* Thursday, August 14, 2008 6:16 AM
> > *To:* bbcl_s...@googlegroups.com
> > *Subject:* BBCL_software Multiple solutions using ResNet?

>
> > Hello again,
>
> > In a paper by Kelley et al. (2002), one of the authors, Sarkar, mentions
> > that he generated 32 distinct (and valid) solutions from a dataset (last
> > paragraph, page 305). Unfortunately the authors do not provide more
> > infromation on how these solutions were obtained, and neither is the work
> > cited in the references.
>
> > My question to the list is, whether it is possible to generate multiple
> > solutions keeping the parameters and dataset the same? In other words, would
> > ResNet create distintly different reserve networks? Obviously this depends
> > on the dataset, but given that it is suffciently large, is it possible that
> > ties could be broken in favour of one cell than another, in different runs?
>
> > Reference
>
> > Kelley, C., Garson, J., Aggarwal A., and Sarkar S. (2002) Place
> > prioritization for biodiversity reserve network design: a comparison of

> > SITES and ResNet software packages for coverage and efficiency. *Diversity
> > and Distributions* (8) 297-306.- Hide quoted text -
>
> - Show quoted text -

[John Simaika]

Hi Trevon,

Thanks for the program! That will make life much easier. However, at the moment I can't get it to work. The program aborts as soon as I double click on it. The directory it is in is as follows:

C:\Randomize_ResNet_input\Randomize_ResNet_input.exe

I have also opened a command prompt and tried to get to the program by typing the root in or by clicking and dragging it into the command prompt:

C:\DOCUME~1\JOHN>C:\Randomize_ResNet_input\Randomize_ResNet_input.exe

What must I do???

Thanks for your help,

John.

[Trevon Fuller]

John,

 

Thanks for your interest in Randomize_ResNet_input.exe. Randomize_ResNet_input.exe reads instructions from a file called logfile.txt that must be placed in the same directory as Randomize_ResNet_input.exe. logfile.txt is a text file that has five rows and indicates the location of the ResNet input file to be randomized, the number of cells in the ResNet input file, etc. This is explained in more detailed in the pdf included in Randomize_ResNet_input.zip. In order to diagnose the problem that you’ve run into, I need to know the contents of logfile.txt.

 

Thanks,

Trevon

From: bbcl_s...@googlegroups.com [mailto:bbcl_s...@googlegroups.com] On Behalf Of John Simaika

Sent: Sunday, August 17, 2008 11:06 AM
To: bbcl_s...@googlegroups.com

[Trevon Fuller]

John,

 

I’m glad that the randomization program works. Unfortunately, ResNet-1.2, Randomize_ResNet_input, and the other software will not work if you drag the logfile to the command prompt. Generally we edit the logfile using Notepad then place the revised logfile in the same directory as the executable. We then run the executable by double-clicking on the icon that represent the executable. It would also work to open the command prompt, cd into the directory that contains the revised logfile and executable, and then type the name of the executable in the command prompt and then hit enter. E.g. >ResNet-1.2.exe or >Randomize_ResNet_input.exe. This presupposes that logfile.txt has already been modified appropriately. Feel free to contact me if you have any other questions.

 

Trevon

 

From: John Simaika [mailto:sima...@gmail.com]
Sent: Tuesday, August 19, 2008 5:22 AM
To: tfu...@mail.utexas.edu
Subject: Re: BBCL_software Re: Multiple solutions using ResNet?

 

Hi Trevon,

That makes sense: I had not modified the logfile that's included in the zipfile. That probably caused the program to crash. (I thought that the logfile etc. had to be dragged into the open command prompt, so it wouldn't matter whether the program is open or not.)

The randomization process works really well now. I'm pretty impressed with the speed too.

Thanks again,

John.

 

[John Simaika]

Hi Trevon,

I'm interested in including attributes other than species in the intput file for ResNet. My approach has been this: At the end of the Input file, I've added columns with the new attribute information, that is in integer format. I have followed the manual for this, which suggests that for attribute information other than species 1s or 0s should not be used, to help ResNet distinguish the attributes from the species presence/absence information in the other columns. The approach has not worked, however. The program simply terminates.

Any suggestions on what can/must be done?

With best wishes,

John.

[Trevon Fuller]

John,

 

We are delighted that you are continuing to use ResNet. Although I can’t be sure what the problem is without looking at your ResNet input file and target file, I have a theory about the source of the problem.

The version of ResNet that can be downloaded by going to http://uts.cc.utexas.edu/~consbio/Cons/Labframeset.html and clicking the “Conservation Software” link assumes that the attribute data are either 0 or 1. A version of ResNet that does not make this assumption is accessible via the ConsNet Portal. I have created an account for you on the ConsNet Portal. To access the account and run ResNet, go to http://consnet.org/ and click “Log  In”. Then enter the username and password in the attachment to this e-mail. Next, click “Run ResNet” in the panel on the left side of your internet browser window. This will take you to a web site that can be used to upload your ResNet input file and target file and run ResNet.

                If the version of ResNet on the ConsNet Portal does not work with your ResNet input file, please send me the input file and target file so that I can diagnose the problem.

 

Thanks again,