Error with networks consisting of multiple components

[mvszuri]

Hello Circuitscape team,

I am running Circuitscape 5 on a network in advanced mode. I have noticed that Circuitscape throws an error if the input network consists of multiple components. Circuitscape runs fine on the following example input files:

mgNetworkVerify3_graph_resistances.txt
mgNetworkVerify3_grounds_resistances.txt
mgNetworkVerify3_sources.txt
With the following settings:
github_test.txt

However, when an edge is added to the ....graph_resistances.txt file between two nodes that are only connected to each other and not to any other node, e.g.:
mgNetworkVerify3_graph_resistances.txt
The following error is returned: ERROR: DimensionMismatch("tried to assign 7-element array to 5×1 destination")

I'm not sure whether this is an unintended error or whether Circuitscape is not meant to work on networks of multiple components. For raster calculations, the Circuitscape documentation does mention that nodes can be completely isolated, which would suggest networks with multiple components.

Thank you very much for your help in advance!

Best regards,
Maarten

[mvszuri]

Just to clarrify; I have posted this as a new issue as I'm not sure whether it's related to the over issue that I posted: #282 (comment)

[vlandau]

Hi @mvszuri sorry for the delay in getting to you here. Have you been able to resolve your issue? Is this problem still occurring in the latest version of Circuitscape? (version 5.8.3)

[mvszuri]

Hi @vlandau, thanks for coming back to this. I haven't tested it on the latest version of Circuitscape. I "solved" the problem by running Circuitscape on a network consisting of only one component. However, the results were not so useful for me, so I stopped using Circuitscape for my current analysis.

[vlandau]

Sorry to hear that! From a look at the source code, it appears that network mode should indeed support multiple connected components for network mode for "advanced" and pairwise problems. @ranjanan is this correct? I'll look into reproducing this either this or next week and see if I can figure out what's going on.

[ViralBShah]

Yes we generally detect the connected components and loop over them. So perhaps this is not well tested.

[vlandau]

I let this slip off of my radar accidentally. I will work on reproducing/debugging this week.

[ranjanan]

There is a work in progress PR for this #320

[ranjanan]

Fixed by #320