October 2015 - graph-tool - archives.skewed.de

Have the nodes spread apart in graph_draw output
by Yannis Haralambous 14 Oct '15

14 Oct '15

dear all, I have an attributed graph and want to display it using graph_draw. You can see the result I obtain in the attached PDF file. It is hardly readable, and still there is a lot of empty space in the PDF page. The command I used is the following: graph_draw(g, vertex_text=labelv, output_size=(2000,2000), vertex_fill_color=typev, edge_text=labele, output="test.pdf") How can I get the vertices spread apart in order to increase readability? Additional question: there must be some random part in graph_draw because every time I run it I get a completely different visual representation. Is there some way to insure that I get the same (or very close, if the data have changed slightly) output between two runs? Thanks in advance Yannis (not Panagis but Haralambous :-) -- ------------------------------------------------------- Yannis Haralambous Professor Institut Mines-Télécom, Télécom Bretagne Computer Science Department UMR CNRS 6285 Lab-STICC Technopôle Brest Iroise CS 83818, 29238 Brest Cedex 3, France Email: yannis.haralambous(a)telecom-bretagne.eu Internet: http://perso.telecom-bretagne.eu/yannisharalambous/ ICBM address: 48°21'31.57"N 4°34'16.76"W Twitter: y_haralambous ------------------------------------------------------- ...the ball I threw while playing in the park has not yet reached the ground (Dylan Thomas) Es gab eine Zeit, wo ich nur ungern über Schubert sprechen, nur Nächtens den Bäumen und Sternen von ihm vorerzählen mögen. (Robert Schumann)

2 1

Nested loop over graph objects.
by Christopher Morris 14 Oct '15

14 Oct '15

Hi, given a set graph_db of graphs, I want to compare the vertices of each graph with the vertices of each other graph from graph_db. Naturally, this results in a fourfold nested loop. Unfortunately , this is really slow in python. Is there a way to way to speed up this process. I tried iterating over indices of graph objects instead of graph objects itself, but this doesn't improve performance. Best regards Christopher

2 1

install graph tool in Centos 7
by 俊赵 11 Oct '15

11 Oct '15

hello, every one I want to install graph tool in CentOs 7, but graph tool has too many dependencies. Such as boost, the Centos 7 default repo only provide version 1.53, but graph tool requires 1.54 And centos 7 does not provide CGAL at all. So I want to ask you, does anyboy have installed graph tool in centos7, which repo do you use? I really do not want to compile all the dependencies through source code, it is a tough work! Simply using yum command is perfect! Many thanks. Zhao Jun

3 5

Controlling vertex labels in GraphWidget
by yannis panagis 09 Oct '15

09 Oct '15

Hi, I apologise for the long post. I am trying to draw an evolving graph and create some snapshots every year. I want to show the growth of the graph by increasing the size of the nodes and the size of the node labels. In the beginning because my node labels were long, drawing the graph caused the nodes to fully enclose the label (example attached), despite any effort to control their size programmatically, making the graph unreadable. Then I discovered the (undocumented?) parameter "vertex_text_position" and started setting it to e.g. 4. Now whenever the nodes grow in size, as the time goes by, the labels are being displaced far away from the corresponding nodes. Three questions: 1. What are the possible values for "vertex_text_position" and what is their meaning 2. Why are the labels being displaced when node grows in size instead of staying next to the node and 3. Is it possible to align the label at center of the node and at the same time ensure the node size is not adjusted to cover the entire label? My code looks approximately like that: pos = g.new_vertex_property("vector<double>") col = g.new_vertex_property("vector<double>") vf_size = g.new_vertex_property("float") for v in g.vertices(): pos[g.vertex(v)] = [float(g.vp['x'][v]), float(g.vp['y'][v])] # load layout from the existing file col[g.vertex(v)] = [int(g.vp['r'][v])/255.0, int(g.vp['g'][v])/255.0, int(g.vp['b'][v])/255.0, 0.8] if not offscreen: win = GraphWindow(g, pos=pos, geometry=(1024, 768), edge_color=[0.6, 0.6, 0.6, 1], vertex_fill_color=col, ) else: win.graph = GraphWidget(g, pos=pos, edge_color=[0.8, 0.8, 0.8, 1], vertex_fill_color=col, vertex_size=g.vp.size, vertex_text=g.vp.label, vertex_text_position=4, vertex_font_size=vf_size, vertex_font_family="sans-serif" ) win.add(win.graph) def update_state(): # Some logic to filter the graph, keep the nodes # that are active at specific milestones # and call pixbuf.savev global count v_list = nodes_in_years[count] # maintains the nodes of that year for v in v_list: active[v] = True for e in list(v.out_edges()): e_active[e] = True g.set_vertex_filter(active) g.set_edge_filter(e_active) for v in g.vertices(): g.vp.size[v] = v.in_degree()*0.9 + 1 # This has a dynamic effect of the node sizes vf_size[v] = v.in_degree()/2.0 # changes the font size win.graph.regenerate_surface() win.graph.queue_draw() # if doing an offscreen animation, dump frame to disk if offscreen: pixbuf = win.get_pixbuf() pixbuf.savev(r'./frames/graph%04d.png' % count, 'png', [], []) if count > years[-1]: sys.exit(0) count += 1 # We need to return True so that the main loop will call this function more # than once. return True Thank you Yannis

3 4

A more simple way to filter vertices of degree 1 an the property (weight) of their edge
by jean-patrick 08 Oct '15

08 Oct '15

Hello, Is there an obvious way to collect vertices of degree 1 for reading the property of their edge? The following code succeed in doing the job on a tiny scale, but I am sure this not the way to do: /import graph_tool.all as gt #from graph_tool import topology ## Make a graph T = gt.Graph(directed = False) edge_weights = T.new_edge_property('double') T.properties[("e","weight")] = edge_weights T.add_vertex(n=5) e_1 = T.add_edge(0,1) e_2 = T.add_edge(1,2) e_3 = T.add_edge(0,0) e_4 = T.add_edge(1,3) edge_weights[e_1]= 2 edge_weights[e_2]= 6 edge_weights[e_3]= 1 edge_weights[e_4]= 3 ## Get the vertices of degree 1 in a dictionary (keys= vertices degree, values=vertices) from collections import defaultdict deg_dic = defaultdict(list) for v in T.vertices(): degree = v.out_degree() deg_dic[degree].append(v) ## Read the weight of the bounded edge print 'vertices of degree 1:' for d1 in deg_dic[1]: weight_of_bound_edge = [edge_weights[e] for e in d1.out_edges()] print 'vertex',d1,' weight of out edge:', weight_of_bound_edge/ By the way, the objective is to count the vertices of degree 1 as a function of the pruning steps. Is there sucha a function in graph-tool? Best regards Jean-Patrick -- 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

vertex_size with GraphView
by Richard Ree 08 Oct '15

08 Oct '15

Hi, I think I've found a bug. Basically I'm trying to use vertex_size with graph_draw() using a GraphView, but it only seems to work with the original graph. import graph_tool.all as gt from numpy.random import randint, random from itertools import izip N = 50 g = gt.Graph() g.add_vertex(N) # insert some random links for s,t in izip(randint(0, N, N), randint(0, N, N)): g.add_edge(g.vertex(s), g.vertex(t)) vsize = g.new_vertex_property('float') vsize.a = random(N)**2 vsize = gt.prop_to_size(vsize) vsize.a *= 10 # this works gt.graph_draw(g, vertex_size=vsize) gv = gt.GraphView(g, vfilt=lambda v:v.out_degree()>1) # this doesn't gt.graph_draw(gv, vertex_size=vsize) The traceback is: /usr/lib/python2.7/dist-packages/graph_tool/draw/cairo_draw.pyc in get_bb(g, pos, size, pen_width, size_scale, text, font_family, font_size, cr) 1279 x_range = [pos_x.fa.min(), pos_x.fa.max()] 1280 y_range = [pos_y.fa.min(), pos_y.fa.max()] -> 1281 x_delta = [x_range[0] - (pos_x.fa - delta).min(), 1282 (pos_x.fa + delta).max() - x_range[1]] 1283 y_delta = [y_range[0] - (pos_y.fa - delta).min(), ValueError: operands could not be broadcast together with shapes (13,) (50,) Thanks, -Rick

2 2

Beginner's trouble with simple graph_tool code snippets
by Yogesh Karpate 08 Oct '15

08 Oct '15

Hi All, After compilation of 40 Hours on my laptop, I have just begun working with graph-tool. I have installed graph-tool-2.9 and my interest lies in finding minimum spanning tree of a graph ..so just tried following code import numpy as np from graph_tool.all import * g, pos = triangulation(np.random((500, 2)) * 4, type="delaunay") tree = min_spanning_tree(g) graph_draw(g, pos=pos, edge_color=tree, output="min_tree.pdf") It throws error " g, pos = triangulation(np.random((500, 2)) * 4, type="delaunay") TypeError: 'module' object is not callable " When I do, g, pos = gt.triangulation(np.random((400, 2)) * 10, type="delaunay") It shows the error NameError: name 'gt' is not defined Is gt python alias for graph_tool? Thanks in advance. -- Warm Regards Yogesh Karpate

3 2

Large Node ID
by Christopher Morris 06 Oct '15

06 Oct '15

Hi, I have a huge graph saved in a text file as an edge list. i.e., "node_id_v, node_id_w". As the node IDs are very large I have to use "long()" to convert "node_id_v" to a number. Therefore, I run in trouble when passing those longs to "add_edge()". Is there a workaround? Best, Chris

2 1

graph_tool.topology.shortest_distance with disconnected graphs
by Christopher Morris 02 Oct '15

02 Oct '15

|Hi,| || |how does dist_map of graph_tool.topology.shortest_distance indicate that two vertices are not connected. Is the value simply set to 2147483647?| || |Best regards,| |Christopher| ||| |

2 1