June 2017 - graph-tool - archives.skewed.de

do_djk_search without distance initialization
by François Kawala 03 Jul '17

03 Jul '17

Hello, I try to dig up an old idea that we've discussed previously (the thread is here <http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com…> ). summary: when one runs shortest distance on a graph with several million vertices the distance initialization cost is high with respect to the actual Dijsktra cost. We want to circumvent the initialization [see L326 of src/graph/topology/graph_distance.cc <https://git.skewed.de/count0/graph-tool/blob/master/src/graph/topology/grap…> ] The proposed approach (as per this boost thread <https://groups.google.com/forum/#!topic/boost-list/7nCzvkkEXCk>) is to use the `dijkstra_shortest_paths_no_color_map_no_init`. The distance initialization is done "on the fly" by the `examine_edge` function. More precisely, the Dijkstra visitor maintains a set of discovered vertices. This set is read in the `examine_edge` function to set the distance to infinity when a the target vertex does not belong to the set of discovered vertices. The `examine_edge` function is also responsible to update the set of discovered vertices. As proposed in the initial thread, we would like to add a init-or-not switch to the shortest_distance parameters (see L1568 of /src/graph_tool/topology/__init__.py <https://git.skewed.de/count0/graph-tool/blob/master/src/graph_tool/topology…> ). I plan to : - define a new Dijkstra visitor named `djk_max_visitor_no_init` (and maybe a second one to handle multiple targets scenario) - update the `do_djk_search` function to use `dijkstra_shortest_paths_no_color_map_no_init` - update the python shortest_distance (add the init-or-not switch) How does it sounds ? Am I missing something ? I have two side questions : - what is the "standard" development environment ? I tried to use CLION <https://www.jetbrains.com/clion/> but I cannot figure out how to configure it properly so that all the boost dependencies are correctly resolved. - how do you debug during development ? Best regards, François.

2 13

Can I calculate likelihood ratios if I calculated get_edges_prob in different runs?
by P-M 20 Jun '17

20 Jun '17

I am trying to run "get_edges_probs" for a large number of edges on a reasonably sized network. This would take too long to run on my local machine so I am considering sending it off to a cluster. Speaking to the admin I would need to split the job into several smaller jobs, each of which I can then run on a different node. My question is thus: If I split my list of edges into, say, 50 separate lists and then set up 50 jobs, each of which processes one of these lists, am I able to compare the results of these jobs to each other by simply recombining the resulting set of loglikelihoods to then calculate my likelihood ratios as outlined in the cookbook? Or does the stochastic nature of the algorithm mean that I can't necessarily compare results from different runs to each other in this manner? Thank you for any help in advance. Best wishes, Philipp -- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com… Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.

2 2

Divert the output of graph_draw to matplotlib axes
by Snehal Shekatkar 19 Jun '17

19 Jun '17

Dear Tiago, Is it possible to divert an output of graph_draw to matplotlib axes? For example, what can I do if I want to draw several graphs using subplots? Thank you Snehal -- Snehal M. Shekatkar Pune India

2 2

gt.minimize_blockmodel_dl for directed networks
by yerali 15 Jun '17

15 Jun '17

Hello dear members of the mailing list, I am not sure I am doing the gt.minimize_blockmodel_dl properly, over directed networks. In the plot <http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com…> you can see the yellow nodes 27 and 147 are in the center of the star-like structures. However, with opposite arrows sense. I would expect the same kind of connectivity, then characterized by similar incomming and outcomming connectivity. My instruction is: State = gt.minimize_blockmodel_dl (g2, deg_corr = False, overlap = False) Where the directionality has been already established in the graphml input file. Shall I change any option in the gt.minimize_blockmodel_dl instruction? Thank you very much in advance, Yérali. -- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com… Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.

2 4

Drawing float values on the vertices
by Snehal Shekatkar 14 Jun '17

14 Jun '17

Dear Tiago, I am trying to draw float values associated with vertices in a graph as vertex texts. However, I get a peculiar output (file attached). My code is given below: import graph_tool.all as gt import numpy as np g = gt.collection.data['karate'] value = g.new_vertex_property('float') for v in g.vertices(): value[v] = np.random.choice([0.5, 0.8]) gt.graph_draw(g, vertex_text = value, output = 'test.pdf') How can I draw the float values as such? Thank you Snehal -- Snehal M. Shekatkar Pune India

2 1

Drawing networks with dissortative communities
by Snehal Shekatkar 12 Jun '17

12 Jun '17

Dear Tiago, I am trying to draw a graph with a dissortative community structure. Is there any way I can force graph-tool to draw the individual nodes into different communities? Unlike assortative communities, all the nodes get mixed up in this case. Thank you Snehal -- Snehal M. Shekatkar Pune India

2 1

weighted undirected graph partitioning
by Luke Taylor 12 Jun '17

12 Jun '17

Is weighted undirected graph partitioning possible? I can't seem to find any pointers in the docs! Thanks in advance, Luke

2 1

vsize_scale in draw_hierarchy not responding.
by Rogelio Basurto 06 Jun '17

06 Jun '17

Hi: I have been working with minimize_nested_blockmodel_dl and while drawing with draw_hierarchy I just would like to improve the default options to make it a bit more appealing. So I increased the label size and would like to do so with the nodes. I try to change the vsize_scale property from its default but nothing happened. The code for the drawing is next: draw_hierarchy(state,bg_color = [1,1,1,1], deg_size = True, vsize_scale = 1000, hvprops = {"color":"black","fill_color":"gray","size":50}, heprops = {"color":"black","pen_width":5}, vprops = {"color":"black"}, eprops = {"pen_width":3}, output_size = (2000,2000), vertex_text = label, vertex_font_size = 60, vertex_text_position = "centered", fit_view = 0.75,#(-500,-500, 1000, 1000),5 output = "figure5a_9.eps"); **Thanks in advance! ** ____ *Rogelio Basurto Flores* Laboratorio de Sistemas Complejos * Unidad Profesional Interdisciplinaria en Ingeniería y * *Tecnologías Avanzadas,* *Instituto Politécnico Nacional,* *Av. IPN # 2580, Col. Laguna Ticoman,* *Mexico D. F., 07340, Mexico* *Mobile Phone: 33 1551 3665* *E-Mail: rogelio.basurto(a)gmail.com <rogelio.basurto(a)gmail.com>*

2 1

Dropping constants while calculating Description Length/Log Likelihood?
by isukritgupta 05 Jun '17

05 Jun '17

Hi Tiago, I wanted to enquire whether the DL that is returned using entropy() function drops any constant factors during calculations. In the calculations to be found in most papers on this topic, there are many factors omitted from the expressions for likelihood, because they are independent of the community assignment. However, in order to calculate the actual value of the likelihood, so as to be able to compare it between degree corrected and non degree corrected models, one must take these factors into consideration. Do you take all the factors into consideration? -- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com… Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.

2 1

y-position in matplotlib subplots outside axis bounds
by william_h277 02 Jun '17

02 Jun '17

I am trying to create a figure with multiple graphs embedded as a pdf using matplotlib subplots. When I try and specify a particular sub figure, the graph gets flipped in the y-direction outside the bounds of the axis. The rectangles and titles are in the correct row. Same axis was passed as mplfig=a to graph_tool.draw(). If i switch axis to bottom row, it flips to the top row. <http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com…> plot command: a1 = pyplot.subplot2grid((2, 18), (0, 6),rowspan=1, colspan=5) graph_tool.graph_draw(graph,inline=False, pos=react.vp.pos, vertex_fill_color=graph.vp.color, edge_pen_width=.5, fit_view=True, vertex_color=[0,0,0,1], vertex_size=10, edge_color=[0.179, 0.203,0.210, 0.5], res=0, output_size=outsize, mplfig=a1) Thanks, William -- View this message in context: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com… Sent from the Main discussion list for the graph-tool project mailing list archive at Nabble.com.

1 0