Maxent batch issue

[Ajay Ranipeta]

Hi all,

I'm currently running Maxent on an EC2 (Fedora 8) server as a batch
mode, but I'm getting the following error:

==================================
Gtk-WARNING **: cannot open display:
==================================

I think the error is telling me that maxent is trying open a window,
but can't seem to. I've set the -a and -z option so the display
doesn't come up, but I still seem to get this error. I've successfully
run the same command on a Windows server, Windows XP and Mac OSX as a
batch and they all works fine.

The process does generates 4 files, rather than all the necessary
files in the output directory:
1) maxent.log
2) maxentResults.csv (but nothing in it)
3) species.html
4) species.lambdas

The command im running is:
--------------------------------------------------------------------------------
java -mx900m -jar /var/maxent/maxent.jar -e /var/maxent/data/modelling/
WorldClimCurrent/10minutes/ -s /env/tomcat/temp/points_15731.csv -z -a
warnings=false tooltips=false -X 0 -N world -N world_10_bio01 -N
world_10_bio02 -N world_10_bio03 -N world_10_bio04 -N world_10_bio05 -
N world_10_bio06 -o /var/maxent/output/
--------------------------------------------------------------------------------

Any help would be greatly appreciated.

Cheers,
Ajay

[Steven Phillips]

Hi all,

We've installed a new Maxent release on the download website, http://www.cs.princeton.edu/~schapire/maxent. 

Main changes in Version 3.3.2:

• "Novel" tool to identify areas of novel climate conditions when projecting
• "Explain" tool to aid understanding of Maxent predictions
• Faster reading of ascii files
• Bug fixes
• Updated tutorial, including Novel and Explain tools
• A more detailed API for calling Maxent from other Java code

Please try it out and let me know if you see any problems.

I'd also like to respond to some recent postings in this group:

Ajay Ranipeta asked about Maxent trying to open a window when running in batch mode on a EC2 server.  Unfortunately, I don't know how to avoid this problem - it happens when Maxent makes its ROC plot and other plots, relying on some java graphics methods that seem to try to access a display even when making a graphics object that will never be displayed.

Ajay also asked about the API for calling Maxent from other java code.  I've added some new functionality to the API, but am still working on the documentation.  I'll send it to you directly, Ajay.  If anyone else would like it, please contact me.

Moris Abolafya and a couple of others asked about errors starting up Maxent.  I don't know what the issue is here, but the "error" that Moris quoted, "if errorlevel 1 pause", is not the error message.  That cryptic message is just something Microsoft Windows requires in order for the DOS window to remain visible after an error when you click on the maxent.bat file.  The actual error is described in the text above the "if errorlevel 1 pause" line in the DOS window.

Anne Overgaard asked how Maxent computes the estimate of maximum achievable AUC.  I think you're talking about the number given in the html output files, just before the ROC curve.  This estimate is not the same as the (1-prevalence/2) estimate from my 2006 Ecological Modelling paper.  The html output says this estimate is calculated by assuming that "test data is drawn from the Maxent distribution itself".  In more detail: imagine drawing a collection of test data from Maxent's raw output distribution.  Some of the test points would be in places with very suitable conditions for the species, at least according to the Maxent output.  Others would have only moderately suitable conditions, or even fairly marginal conditions, again according to Maxent's estimate.  The former points have hight Maxent output values, so they contribute towards a higher AUC.  The latter points have moderate or fairly low Maxent output values, which lowers the AUC.

"Crawfish" asked about older versions of Maxent.  We've now archived all previous Maxent releases on the Maxent download site.

Matteo asked about the maximum value of logistic output.  The output always lies between 0 and 1, but can technically never exactly equal 0 or 1.  With floating point rounding you'll often get numbers that appear to be 0 or 1 though.  The maximum value for any particular prediction does not need to be close to 1 -- for example, for a null distribution that predicts the species is equally likely everywhere, Maxent gives logistic values of 0.5 everywhere.


-- Steven

[Alexandre Silva de Paula]

I am doing distribution analysis using the Maxent and I would like to
know what is the minimum number of points (locations) to be used in
the program. Does anyone have bibliography about this question?

Best wishes,

Alexandre Silva de Paula.

[Ajay Ranipeta]

Hi Steven,

Thanks for that. I'm happy to test it out and get back to you with
feedback. I'm currently already using version 3.3.2 of maxent.

Also, do you know/think that the API usage will generate the graphic
issue with plot generation?

Cheers,
[a]

On Mar 3, 2:46 am, Steven Phillips <phill...@research.att.com> wrote:
> Hi all,
>

> We've installed a new Maxent release on the download website,http://www.cs.princeton.edu/~schapire/maxent.

>
> Main changes in Version 3.3.2:
>

>     * "Novel" tool to identify areas of novel climate conditions when
>       projecting
>     * "Explain" tool to aid understanding of Maxent predictions
>     * Faster reading of ascii files
>     * Bug fixes
>     * Updated tutorial, including Novel and Explain tools
>     * A more detailed API for calling Maxent from other Java code

[Ayse]

Hi all,

 

Many thanks to Steven Philips for the new version. The new tools are very usefull.

 

However, I’m having problem using the “Explain” tool. Everythig is okay if I use only 3  environmental layers, but if I increase this number, the memory is not enough, although I increased java heap size to 1300Mb. Is there any way I can work around this problem. (Each layer is around 80Mb)

 

Thanks in advance

Ayşe

 

 

 

--------------------------------

Ayşe S. Turak, Ph.D.
Conservation GIS and Modelling Expert

Nature Conservation Centre www.dkm.org.tr

320. Sok. No:4, Ortadoğu Sitesi, Yüzüncüyıl, Ankara, Turkey

tel:+90 312 287 8144 fax:+90 312 287 4067

ayse....@dkm.org.tr

 

 

---

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To post to this group, send email to max...@googlegroups.com.
To unsubscribe from this group, send email to maxent+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/maxent?hl=en.

[Steven Phillips]

Ajay,

Yes, the graphic issue seems to be inherent in java graphics, so it
isn't affected by API usage.

-- Steven

[Steven Phillips]

Ayşe,

The Explain tool needs all the grids to be in memory, so it will have trouble with very big grids.  It sounds like you should be able to use more than 3 layers though - I'll have to look into it to see if I can find more ways to be memory efficient. 

In the meanwhile, the way to get around the problem is to subsample your grids.  You can do that in a GIS, or there's a quick command-line method to do it using the maxent jar file - for example,

   java -mx1000m -cp maxent.jar density.tools.SubSample layers newlayers 3 asc

will subsample each layer in "layers" by keeping only every third row and every third column, writing the results in the directory "newlayers" (which you need to make first).  The resulting grids will be 9 times smaller than the originals.  You may need to put the full path to maxent.jar in there, and the paths to the "layers" directory.

-- Steven

[Ayse]

Thanks Steven,

 

 I would heve liked to keep the high resolution, but meanwhile I will take up your sudgesiton.

 

Ayşe

[RAKOTOMANJAKA Andry Jean Marc]

Hi Steven,

 

I've just downaloaded the latest version, where I can see these new Tools ?

 

Thanks

 

Andry

__________ Information from ESET NOD32 Antivirus, version of virus signature database 4891 (20100223) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com

--
You received this message because you are subscribed to the Google Groups "Maxent" group.
To post to this group, send email to max...@googlegroups.com.
To unsubscribe from this group, send email to maxent+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/maxent?hl=en.

__________ Information from ESET NOD32 Antivirus, version of virus signature database 4891 (20100223) __________

The message was checked by ESET NOD32 Antivirus.

http://www.eset.com

[direwolf...@gmail.com]

Hi All,

How i can get The Explain Tool

Best regards,

Amri

[Jamie M. Kass]

The Explain tool is in the Maxent download folder. It’s called explain.bat.

Jamie