Activity for Oliver Kullmann
-
Oliver Kullmann posted a comment on discussion Feedback
Thanks a lot! Somehow I missed the Stellarium User Guide, which seems a good read. Now downloading both DE430/431. It was just to tempting to go to the year -50000 and imagine how that was. And then forward to +50000!
-
Hello, since three years I got into astronomy, with a small telescope and binoculars. A few months ago I found Stellarium, and find it very useful. For me also the educational aspect is very important, and so I use going back and forth in time, which I find fascinating. I watched especially the moon over the time scale as provided by Stellarium (-10^5 to +10^5 years), and found interesting phenomena, which however I couldn't locate in the literature. Just bought a quite detailed book on the (earth-)moon,...
-
Hello, on the file Test.tlp with the content (tlp "2.3" (nb_nodes 2) (nodes 1..2) (edge 2 1 2) ) the Tulip-app version 5.2 fails with the error message Error when parsing '' at line 4 When using (nodes 1..3) instead, then parsing succeeds. This looks like a clear bug to me? Oliver
-
Hello, beginning of semester here, everything takes it s time. Currently it seems actually most useful to us if we compute the colouring of nodes (and edges) ourselves, putting that already into the tree given to the Tulip-application. We can easily compute anything we want and put into into the data given to Tulip. That seems indeed easier for us, and we have full control over that. Leaving then the display and all of that to Tulip. Our own program is C++. Now it would be great if there would be...
-
Ha, that's fun -- Dan is just next door :-) So that should solve indeed the problem! Thanks Benjamin Concerning that hacking-business: As I said, I worked with properties, in a more generic (powerful) setting, in the Boost (C++) Graph library. From a complexity point of view it's a hack -- but that doesn't mean that it isn't useful! The whole Linux is kind of a hack, and it is useful. And perhaps that hack here is very hard to make better, in general. You might then not call it a "hack", but I didn't...
-
Hi Guy, today I am fully busy with teaching and admin, so can continue properly only tomorrow. But just some quick remarks: 1. We definitely need a lot of experimentation, and thus need much (all) of the functionality available in the Tulip 5.2 application. When you write "script", is this some form of plug-in to that application? 2. What seems fully sufficient would be to add the field "selection" (as you have it for "To labels") to the algorithm "Color Mapping". Then we could compute the colouring...
-
You write: "There are several options: associate the whole tuple at once or spread them over different properties (it makes sense since the tuples all have the same size and k may be fixed)." Yes, but that's just a technicality. Having it as single properties is perhaps better, since then you have proper names for it. You write: "I believe your problem is to propermy define a colormap so you don't need to inspect individual values but rather get an overview of how values distribute over nodes. Am...
-
You write: "There are several options: associate the whole tuple at once or spread them over different properties (it makes sense since the tuples all have the same size and k may be fixed)." Yes, but that's just a technicality. Having it a single properties is perhaps better, since then you have proper names for it. You write: "I believe your problem is to propermy define a colormap so you don't need to inspect individual values but rather get an overview of how values distribute over nodes. Am...
-
Hi Melancon, I think that property-business is more of a hack, and you can't get that implemented really efficiently. But that's not the problem, the Boost graph library takes a similar (but more generic) approach. One just needed to take, in that application, the properties serious! That is, allow for example assignments for them. Then we could just assign the local colouring to the global colouring, and problem solved! The scale of colouring-algorithms available in the application looks fine, and...
-
Dear Melancon, you wrote "Haven't you seen my answer to your first post?" I indeed updated the Forum-page permanently, but your answers did show up only some time later ... If I can, I definitely want to avoid any scripting (never Python), and any programming (all of that Qt ... ). It's not just me, it's also the students --- they can waste endless time with that. And if that problem with colouring only a subclass of nodes is solved (which should be possible), then we are there! The selection of...
-
Hello, thanks for your efforts! You write "You may not want that to happen, and a possibility is that you want the property to be imported at the (super) graph level. This is what the "Always store results ..." does." That would be good, but it doesn't work: One runs the colouring-algorithm on the subgraph, indeed it re-computes the colours only on the vertices of the subgraphs -- but the colours of the whole graphs are not changed, unfortunately. If this would work, then problem solved! Wouldn't...
-
Under algorithms there exists an option "Always store results in an existing property of the graphs hierarchy", which sounds as if one could get lift the colouring-results for the subgraph to the whole graph, but it doesn't work. What's the point of these "local colours"??
-
I hoped that with "clusters" one could colour the inner nodes differently from the leaves. But unfortunately the colouring-algorithm has no effect, when a cluster is selected! The variable "input property" is reset when clicking on a different cluster/sub-graph, but that's it, no other effect (except of changing the "local variable" viewColor, which seems not usable). The whole usage seems rather unintuitive to me. It seems that at least the application is driven by intuitions coming from very special...
-
Through experimentation meanwhile I found out the following (seem to be speaking to myself, but who knows): 1. There is the rendering-algorithm "Hierarchical Tree (R-T Extended)", which should be the Rheinhold-Tilford algorithm. Seems basically to work, except that the handling of the node-sizes, which is important here, can not be directly influenced. 2. For the colour-scheme there is "Color Mapping", which computes the attribute viewColor. Concerning "alpha", there is "Alpha Mapping", by which...
-
That "AppImage" I just downloaded (I have an old Suse-Linux installation), and then clicked on it (or run it from the command-line), and it worked.
-
Searching the literature for visualisation of graphs, I can't find anything in the direction we needed (somewhat strange). Actually, the binary-tree-structure is rather central, so instead of a graph we should focus on binary trees here (rooted trees, where every node has at most two children, and the children are either "left" or "right" children). We also want to look at small trees, with just hundreds of nodes, but really interesting case should at least include millions of nodes, hopefully up...
-
Searching the literature for visualisation of graphs, I can't find anything in the direction we needed (somewhat strange). Actually, the binary-tree-structure is rather central, so instead of a graph we should focus on binary trees here (rooted trees, where every node has at most two children, and the children are either "left" or "right" children). We also want to look at small trees, with just hundreds of nodes, but really interesting case should at least include millions of nodes, hopefully up...
-
Searching the literature for visualisation of graphs, I can't find anything in the direction we needed (somewhat strange). Actually, the binary-tree-structure is rather central, so instead of a graph we should focus on binary trees here (rooted trees, where every node has at most two children, and the children are either "left" or "right" children). We also want to look at small trees, with just hundreds of nodes, but really interesting case should at least include millions of nodes, hopefully up...
-
Hello, I have the following generic problem: Given is a graph (actually it is typically a binary tree), and a natural number k, such that every node has a k-tuple of real numbers associated with it. Visualise this labelled graph. Via property-type double I can create such graphs, but how to visualise them? The easiest possibility which comes into my mind is a grey scale, for just one component of the k-tuple. In the application one can download, I couldn't find such an option? Of course, I would...
1