ANN: Circuitscape 5.0.0!

[Ranjan Anantharaman]

Hi, 

I am very happy to announce the release of Circuitscape 5.0.0, the next iteration of Circuitscape, which is written to be modern, fast and scalable. This project is open source and hosted on GitHub here: https://github.com/Circuitscape/Circuitscape.jl

It runs up to 4x faster on our benchmark tests (all raster pairwise mode), and includes new features: 

- Parallel runs on Windows! No more slow Circuitscape simulations in serial on Windows anymore. Take advantage of your entire system's compute power, no matter which platform (Windows, Mac or Linux)! 

- New solver mode (cholmod) based on cholesky factorisation. For certain problem sizes runs up to 8x faster than the previous Circuitscape version. Use it by changing "solver = cholmod" in your INI file.

- Single precision support for even more speed (experimental feature). Single precision sacrifices precision for performance. If you're interested in sacrificing some accuracy for performance, you can try this mode! Use it by adding "precision = single" to your INI file.

Installation: 

Install Circuitscape in two easy steps: 

1. To install the new Circuitscape, you need to install the Julia language on your system. Go to this link, and download the executable specific to your system:

a. If you're on Windows, double click on the installer and follow the instructions. Double click on the shortcut to launch the Julia prompt. 

b. If you're on a Mac, download the .dmg file and copy the application to your /Applications/ folder. Double click on the .app to launch the julia prompt. 

c. If you're on Linux, download the tar ball, and untar it. The julia binary is at UNTAR_FOLDER/bin/julia. Run it to launch the julia prompt. 

2. Once you launch the Julia prompt, type `Pkg.add("Circuitscape")` and hit enter. This adds the Circuitscape package to your Julia environment. 

That's it! Circuitscape is now installed on your system. 

Usage: 

At the julia prompt, you must load the Circuitscape package into your workspace by: 

julia> using Circuitscape

Once you do this, you can use it by typing: 

`compute("path/to/ini/file")` and hitting enter

The new Circuitscape is designed to work with all your old INI files, so feel free to use them! 

Bug Reporting and Contributing: 

Circuitscape is an open source project so contributions in terms of bug reports and code submissions are most welcome! 

Bug reports can be submitted by opening a GitHub issue. Go to this link, and click on "New Issue" and proceed to lay out the problem. I'll do my best to fix it as soon as I can! You should also feel free to open Issues for feature requests and new ideas as well!  

Future Work: 

I am currently hard at work putting together a UI with an installer together that would make the new Circuitscape even easier to use. I will announce here whenever that happens too. 

Disclaimer: This project is still at an alpha stage, and there still may be bugs I may have missed. I can promise it will only get better with time, and bug reports and new tests are more than welcome. 

Acknowledgements: Special thanks to the late Dr. Brad McRae, the original author of Circuitscape, for his initial help and whose legacy I hope to keep alive through my work, Dr. Viral Shah, Brad's collaborator on the project, and Dr. Kimberly Hall of the Nature Conservancy.  

I'd love feedback on the project. Please feel free to post on this Google Group or on the GitHub issues. 

I hope this software helps you and your work. 

Thank you, 

Ranjan Anantharaman

[Kim Hall]

Hooray Ranjan!  Hey everyone, please check this new version out and give us your feedback!

[Srinivas V]

Hi Rajan,

I tried installing version 5, but I I get the following error. Not sure if I need to install any other dependencies after installing julia. I'm using Ubuntu 18.04.01 LTS

Thanks and Regards

Srinivas Vaidyanathan

--
Visit Circuitscape at http://www.circuitscape.org/
To post to this group, send email to circui...@googlegroups.com
To unsubscribe from this group, send email to circuitscape...@googlegroups.com
For other options, visit this group at http://groups.google.com/group/circuitscape?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "Circuitscape" group.
To unsubscribe from this group and stop receiving emails from it, send an email to circuitscape...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Viral Shah]

A fix is on the way. We are still updating circuitscape to Julia 1.0.

For now you still have to use Julia 0.6. sorry about the confusion error msg. It will all be fixed in a few days as we upgrade.

-viral

[Srinivas V]

Thanks Viral,

Will downgrade the version of Julia and keep you posted.

Cheers

Srinivas Vaidyanathan 

[Johannes Sörensen]

For those who cannot downgrade or do not want to. I managed to install it in julia 1.0 with dev "Circuitscape" instead of add (in package mode). This will however install the latest trunk version from github instead of a release version so depending what changes were made recently this might not work for you.

[Srinivas V]

I have managed to install version 5 on julia 0.6 and run the analysis. It took about 4 days to analyse 28632880 nodes,160 focal points, 26 connected components and a total number of 12403 pairs to solve.

For now, I'm interested only in the 3 column/matrix of resistances. However, I have noticed that the cumulative current map has only NA values. I'm unsure if this is due to the CHOLMOD solver or due to the pairwise scenario. I have analysed the same dataset earlier using version 4, cg+amg solver and one-to-all scenario and the output maps were generated.

Any solutions/ suggestions to fix the error with the output maps will be very helpful before I setup the final analysis. I also hope that the resistance matrices are correct and that the error with the current map has no bearing on the values in the matrices.   I have attached the ini, log to help troubleshoot this problem.

I also notice that the resistance.out file from the one-to-all analysis shows negative values for some focal points, which I assume are disconnected. Whereas there are no negative values in the pairwise analysis suggesting that no two pairs are disconnected. Is my interpretation correct? And what would lead to this differences?

Thanks and regards

Srinivas Vaidyanathan

[Benjamin Siggery]

Hi Ranjan,

Thank you for all your hard work on this. I have just discovered this and have been trying to use Circuitscape 4 but encountered errors in Core Mapper that I can't quite solve.. I am not familiar with Julia at all - is there a simple guide anywhere for how to step-by-step use my Circuitscape 4 files with the new version in Julia? Or is it best to wait for the release of the GUI? I need to be able to analyse this data before Jan/Feb time.

Cheers

Ben Siggery

[Srinivas V]

Dear Viral and Rajan,

I had earlier posted about errors in the cumulative current map, it only NA values. I'm unsure if this is due to the CHOLMOD solver or due to the pairwise scenario. I have analysed the same dataset earlier using version 4, cg+amg solver and one-to-all scenario and the output maps were generated.

A solution to fix the output maps will be very helpful before I set up the final analysis. I also hope that the resistance matrices are correct and that the error with the current map has no bearing on the values in the matrices.

I also notice that the resistance.out file from the one-to-all analysis shows negative values for some focal points, which I assume are disconnected. Whereas there are no negative values in the pairwise analysis suggesting that no two pairs are disconnected. Is my interpretation correct? And what would lead to this differences?

Thanks and best regards

Srinivas Vaidyanathan