I studied strongly vs. weakly connected components little bit. So... 1. As long as there are positive costs in both arc_column and arc_backward_column between nodes, they are in the same strongly connected components. 2. If any of arc_column and arc_backward_column is positive between nodes, they are in the same weakly connected components. 3. If both columns are negative between two nodes, they are disconnected even if there is a line between them. 4. Because of point 3 above, using and not using both columns may result in different weakly connected components. No connected lines vs. no traversal using a negative cost with connected lines.
Hope my understanding is correct. Thank you. Huidae On Sun, Oct 29, 2017 at 12:07 PM, Markus Metz <[email protected] > wrote: > > > On Sat, Oct 28, 2017 at 3:18 PM, Huidae Cho <[email protected]> wrote: > > > > BTW, let me understand weakly vs strongly connected components generated > by v.net.components. > > > > Strongly connected: All paths between nodes u and v should have the same > direction. > > From the manual: > "there are directed paths from u to v and from v to u" > > You can create directed graphs by assigning costs according to line > direction with the arc_column and arc_backward_column options. The forward > or backward direction of a line can be closed by setting costs to -1. > > By default, (no cost columns given), line lengths are used as costs, > ignoring line direction because 1) line direction can be arbitrary, 2) > typically lines can be traversed in both directions. > > Therefore, method=strong only makes sense if both arc_column and > arc_backward_column are given and if some lines are closed in one > direction. Otherwise, strong and weak will give the same results. > > > > Weakly connected: Only requires "connections" between nodes u and v, and > their directions don't matter. > > To be precise, u must be reachable from v or v must be reachable from u. > > > > If my understanding is correct, a stream vector map generated by > thinning and converting the stream raster map (flow directions not > guaranteed) to vector would only be weakly connected. > > > > The stream vector output from r.streams.extract would be both strongly > and weakly connected because it honors flow directions. > > Different stream networks are always different components. If you use > arc_backward_column and close backward traversal, each stream network is > again broken up into various strongly connected components. > > If you use r.thin + r.to.vect + arc_backward_column, a single stream > network might even be broken up into several weakly connected components. > > HTH, > > Markus M > > > > Please let me know if I'm mistaken. > > > > Thanks. > > Huidae > > > > > > > > On Sat, Oct 28, 2017 at 9:06 AM, Huidae Cho <[email protected]> wrote: > >> > >> It's working great! Thanks for the quick implementation. It will help a > lot! > >> > >> Huidae > >> > >> On Fri, Oct 27, 2017 at 3:58 PM, Markus Metz < > [email protected]> wrote: > >>> > >>> > >>> > >>> On Fri, Oct 27, 2017 at 2:38 PM, Huidae Cho <[email protected]> wrote: > >>> > > >>> > [...] I tried your r.stream.extract to extract stream lines and its > output makes much more sense. It would be great if this module also outputs > unique stream "network" numbers. > >>> > >>> Unique stream network numbers are added as new attribute "network" in > trunk r71599. > >>> > >>> Markus M > >>> > > >>> > Thanks. > >>> > Huidae > >>> > > >>> > > >>> > On Fri, Oct 27, 2017 at 8:14 AM, Markus Metz < > [email protected]> wrote: > >>> >> > >>> >> > >>> >> > >>> >> On Fri, Oct 27, 2017 at 12:52 PM, Huidae Cho <[email protected]> > wrote: > >>> >> > > >>> >> > Actually, I'm getting the same error as Markus in the NC dataset. > I recently discleaned and rebuilt the latest SVN. > >>> >> > >>> >> In trunk, this bug appeared with r71267 and was fixed with r71477. > >>> >> In relbr72, this bug appeared with r71268 and was fixed with r71478. > >>> >> > >>> >> Please check your revision. > >>> >> > >>> >> Markus M > >>> >> > >>> >> > > >>> >> > Huidae > >>> >> > > >>> >> > On Thu, Oct 26, 2017 at 4:14 PM, Markus Metz < > [email protected]> wrote: > >>> >> >> > >>> >> >> > >>> >> >> > >>> >> >> On Thu, Oct 26, 2017 at 5:02 PM, Markus Neteler < > [email protected]> wrote: > >>> >> >> > > >>> >> >> > On Thu, Oct 26, 2017 at 9:58 AM, Stefan Blumentrath > >>> >> >> > <[email protected]> wrote: > >>> >> >> > > Hi, > >>> >> >> > > > >>> >> >> > > Did you try: > >>> >> >> > > > >>> >> >> > > https://grass.osgeo.org/grass72/manuals/v.net. > components.html > >>> >> >> > > > >>> >> >> > > That might give you a common attribute for all connected > lines in your > >>> >> >> > > network… > >>> >> >> > > >>> >> >> > I just tried that with NC but: > >>> >> >> > > >>> >> >> > GRASS 7.3.svn (nc_spm_08_grass7):~ > v.net.components > input=roadsmajor > >>> >> >> > output=roadsmajor_components method=strong > >>> >> >> > Building graph... > >>> >> >> > Registering arcs... > >>> >> >> > 100% > >>> >> >> > Flattening the graph... > >>> >> >> > ERROR: GngFlatten error > >>> >> >> > > >>> >> >> > > >>> >> >> > GRASS 7.3.svn (nc_spm_08_grass7):~ > v.net.components > input=railroads > >>> >> >> > output=railroads_components method=strong > >>> >> >> > Building graph... > >>> >> >> > Registering arcs... > >>> >> >> > 100% > >>> >> >> > Flattening the graph... > >>> >> >> > ERROR: GngFlatten error > >>> >> >> > > >>> >> >> > Does the network need any special treatment? > >>> >> >> > >>> >> >> I can not reproduce these errors. Please update your local copy: > >>> >> >> make distclean > >>> >> >> svn --no-ignore status > >>> >> >> svn up > >>> >> >> configure > >>> >> >> make > >>> >> >> > >>> >> >> Markus M > >>> >> >> > > >>> >> >> > Markus > >>> >> >> > _______________________________________________ > >>> >> >> > grass-user mailing list > >>> >> >> > [email protected] > >>> >> >> > https://lists.osgeo.org/mailman/listinfo/grass-user > >>> >> >> > >>> >> >> > >>> >> >> _______________________________________________ > >>> >> >> grass-user mailing list > >>> >> >> [email protected] > >>> >> >> https://lists.osgeo.org/mailman/listinfo/grass-user > >>> >> > > >>> >> > > >>> > > >>> > > >> > >> > > >
_______________________________________________ grass-user mailing list [email protected] https://lists.osgeo.org/mailman/listinfo/grass-user
