Pinchpoint Failure on step 8.

211 views
Skip to first unread message

Lindsay Hunt

unread,
Jan 23, 2017, 10:32:36 AM1/23/17
to Linkage Mapper
Good Morning Everyone, 

I have been working on a large connectivity project that covers the entire coastline of the great lakes with many cores (~374) pretty much evenly distributed across the coastline. I was able to successfully run the linkage mapper program, I then tried to run the Pinchpoint Mapper twice and both times it has failed at the same place. Here is the fail message.....any ideas of what might be wrong or how I might change this?

---------------------------------
****Failed in step 8. Details follow.****
---------------------------------
Python error on **line 243** of s8_pinchpoints.py in Linkage Mapper Version 1.0.9:
MemoryError
---------------------------------
Python error on **line 158** of circuitscape_master.py in Linkage Mapper Version 1.0.9:
SystemExit: 1
Completed script PinchMapper...
Failed to execute (PinchMapper).
Failed at Sat Jan 21 17:32:34 2017 (Elapsed Time: 29 hours 59 minutes 20 seconds)

Thanks for your help!!

Lindsay

Brad McRae

unread,
Jan 23, 2017, 10:44:21 AM1/23/17
to linkage...@googlegroups.com
Hi Lindsay,

Yes, this means that your computer doesn't have enough RAM to run Circuitscape with this large of a study area. This is less likely to happen if you are using the "Calculate adjacent pair pinch points using Circuitscape" option than the "Calculate raster centrality using Circuitscape"- the latter is especially memory intensive.

Feel free to email me directly with follow-up questions.

Brad

--
You received this message because you are subscribed to the Google Groups "Linkage Mapper" group.
To unsubscribe from this group and stop receiving emails from it, send an email to linkage-mapper+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Lindsay Hunt

unread,
Feb 10, 2017, 5:36:29 PM2/10/17
to Linkage Mapper
Good Afternoon, 

I have been running Pinchpoint on another set of files and it got through all the calculations and I received this error

--------------------------------
****Failed in step 8. Details follow.****
---------------------------------
Python error on **line 458** of s8_pinchpoints.py in Linkage Mapper Version 1.0.9:
RuntimeError: ERROR 999998: Unexpected Error.
---------------------------------
Python error on **line 158** of circuitscape_master.py in Linkage Mapper Version 1.0.9:
SystemExit: 1
Completed script PinchMapper...
Failed to execute (PinchMapper).
Failed at Fri Feb 10 12:46:28 2017 (Elapsed Time: 166 hours 39 minutes 59 seconds)


Is this another memory error?

Lindsay

Brad McRae

unread,
Feb 10, 2017, 5:44:03 PM2/10/17
to linkage...@googlegroups.com
Hi Lindsay,

Probably not a memory error- it is failing to export a raster to a
numpy array. If this persists after restarting Arc, let me know
(email me directly at mc...@circuitscape.org) and I can send you some
code that will give a more specific error message.

-Brad
> --
> You received this message because you are subscribed to the Google Groups
> "Linkage Mapper" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to linkage-mappe...@googlegroups.com.

John Gallo

unread,
Feb 13, 2017, 12:58:04 PM2/13/17
to Linkage Mapper
HI Lindsay

Did you get a chance to try out the windows compression we discussed a few months ago?  If so, how did it work?  I'm thinking of bringing that up at a GIS meeting tomorrow in our org tomorrow.  

Also, if you are finding yourself needing to re-run LM in the same directory and it not allowing it because of a schema lock unless you restart Arc, you can let Brad or I know as there is a small fix for this that will go out in the next release. 

Cheers,

John

Lindsay Hunt

unread,
Feb 13, 2017, 3:31:54 PM2/13/17
to linkage...@googlegroups.com
yes it worked really well. My resistance raster file has 31,420,841 cells at a 30 X 30 m resolution, they span the entire coast line of the great lakes. I had a core nodes file with 1110 cores that needed to be connected. This attempt based on previous smaller ( number of cores) attempts should have taken multiple TB of space for memory, a previous one maxed out a 4TB external hard drive. For the run I described above I created the save directory on an external 6TB drive and the screen shot below shows the total space used on the hard drive:
Inline image 1
The directory itself looked like this after completing linkage mapper and pinchpoint analyses.
Inline image 2

I am very pleased with the space saving and the output of this run.

Lindsay

--
You received this message because you are subscribed to a topic in the Google Groups "Linkage Mapper" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/linkage-mapper/NMRddl8TVBU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to linkage-mapper+unsubscribe@googlegroups.com.

John Gallo

unread,
Feb 14, 2017, 3:41:46 PM2/14/17
to Linkage Mapper
That is great!  And 31 million cells is impressive indeed.  Glad to hear and thanks for the report.  And I know it would be hard to tell, but any anecdotal evidence about if it slowed things down, or was the compression/uncompression time compensated by time savings elsewhere as Randal suggested to me a while back?  I know that would be hard to test, so no problem if no data on that.  Good luck with all!  John


On Monday, January 23, 2017 at 7:32:36 AM UTC-8, Lindsay Hunt wrote:

Lindsay Hunt

unread,
Feb 22, 2017, 2:33:16 PM2/22/17
to Linkage Mapper
I have two quick questions, 

1. I have run the linkage mapper on a smaller study area along with the large study area I have been working on. When I open up the un-truncated corridor files from the linkage mapper I find the scale goes from -171 for the smaller study area and -211 for the larger study area, I have seen other questions about negative numbers in the outputs but I am unclear how big of a problem this is. 

2. Could you direct me to a source to better understand the Barrier Mapper tool and who to interpret the output? 

Thanks, 
Lindsay

Brad McRae

unread,
Feb 22, 2017, 4:24:42 PM2/22/17
to linkage...@googlegroups.com
Hi Lindsay,

For 1) did you get a warning message? If so there may be problems with bounding circles being too small. If not it is probably just rounding error.

For 2, have you looked at McRae et al. 2012, Torrubia et al. 2014, and Dutta et al. 2015? The first describes the method, the second two apply it.

To unsubscribe from this group and stop receiving emails from it, send an email to linkage-mapper+unsubscribe@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages