Room for question 139896: recreating-a-hierarchical-graph-from-a-compact-representation

MB1965

18:48

@ralph please post the edge list for spooled graph you're working from

Then provide me the image you think it should produce

Take a screen shot and upload it via this: imgur.com/upload

Then send me the link to that

I don't want to trawl through your google drive

Okay, what is z in this case?

I'm not pulling some notebook off your google drive to work with.

@ralph you now have > 20 rep. Now we can get to work.

@ralph you should know that I'm no longer going to respond to comments on that post. From now forward if you want my help you'll have to post here. And so I can see that you've posted you'll have to send a message tagged with @MB1965.

@ralph try reloading the page. Adding any more to that page will only serve to confuse other people who come to see the question. The comments are not intended for extended discussion.

You have 24 reputation and so chat should become accessible

ralph

19:59

Now it is working

MB1965

Great

What's z in this case?

ralph

I still testing algorithm

MB1965

Okay

I think it's pretty solid at this point.

How are you generating these graphs?

Do you start with linearized graphs and convert them to that form?

Or do they start in that form?

Because as I pointed out one in my update one of your graphs can't actually be reversed.

That is, once you linearize it you can't convert it back.

ralph

20:23

These graphs I wrote a hand for the needs of the algorithm :)
Up until now, I have transformed some physical issues into graphs (graph2).
Then I took the graphs analogous to graph1. I studied their fractal properties, distribution of nodes degree, ...
Now the question arises how the graph1 transforms to the other side ie to the primary form (graph 2).

MB1965

Cool

Well then after you down convert to what you're calling graph2 you can't always go back.

Just something you need to be aware of

ralph

20:35

z1= {1000 <-> 1, 1000 <-> 2, 1000 <-> 3, 1000 <-> 4, 1000 <-> 5,
1000 <-> 6, 1000 <-> 7, 1000 <-> 8, 1000 <-> 9, 1000 <-> 10,
1000 <-> 11, 1000 <-> 12, 1000 <-> 13, 1000 <-> 14, 1000 <-> 15,
1000 <-> 16, 1000 <-> 17, 1000 <-> 18, 1 <-> 19, 1 <-> 20, 1 <-> 21,
1 <-> 22, 1 <-> 23, 19 <-> 24, 19 <-> 25, 19 <-> 26, 2 <-> 27,
2 <-> 28, 3 <-> 29, 4 <-> 30}

Interesting. For such a simple graph it gets a strange result ie for unSpoolGraph[] we receives a disjoint graph (two graphs)

MB1965

Where are you seeing this?

Did you pull unSpoolGraph from the code after I updated it ~1 hr ago?

Because I don't get a disjoint graph.

If you didn't, you're getting that because I used MapIndexed

Then I updated it for the case that the nodes don't simply go in order.

ralph

I am sorry, my mistake. I loaded the wrong data. But I get (in my opinion) wrong result:

{1 <-> 2, 2 <-> 3, 3 <-> 4, 4 <-> 5, 5 <-> 6, 6 <-> 7, 7 <-> 8,
8 <-> 9, 9 <-> 10, 10 <-> 11, 11 <-> 12, 12 <-> 13, 13 <-> 14,
14 <-> 15, 15 <-> 16, 16 <-> 17, 17 <-> 18, 19 <-> 20, 20 <-> 21,
21 <-> 22, 22 <-> 23, 23 <-> 24, 46 <-> 47, 47 <-> 48, 48 <-> 49,
25 <-> 26, 26 <-> 27, 30 <-> 31, 28 <-> 29, 52 <-> 23, 51 <-> 22,
50 <-> 21, 45 <-> 1, 44 <-> 2, 43 <-> 3, 42 <-> 4, 41 <-> 5,
40 <-> 6, 39 <-> 7, 38 <-> 8, 37 <-> 9, 36 <-> 10, 35 <-> 11,
34 <-> 12, 33 <-> 13, 32 <-> 14, 1 <-> 20, 46 <-> 19, 25 <-> 15,

MB1965

Ahh. Here you're gonna have a problem because you have that -1 and -2.

Unless this is the output?

Or is this the input?

ralph

20:50

node -1 has 4 degree

MB1965

You didn't answer my question...

Is that the output of unSpoolGraph

ralph

This is output

MB1965

Or the input to it?

It's degree 1...

VertexDegree[<that_data>,-1]

is 1

How'd you even generate that graph? Did you update unSpoolGraph?

Because unSpoolGraph naturally puts the nodes in the center.

Like the long chains.

I realized I can refine my claim on when this transformation is reversible and so can improve unSpoolGraph

ralph

ok, VertexDegree[<that_data>,1] is 4

MB1965

But that isn't the output from the version that's up right now

Yeah, I saw that

When is the last time you pulled the code?

ralph

20:57

Yes, I update unSpoolGraph

MB1965

And also unSpoolingChoices?

Send me the input you ran unSpoolGraph on

ralph

input: {1000 <-> 1,
1000 <-> 2, 1000 <-> 3, 1000 <-> 4, 1000 <-> 5, 1000 <-> 6,
1000 <-> 7, 1000 <-> 8, 1000 <-> 9, 1000 <-> 10, 1000 <-> 11,
1000 <-> 12, 1000 <-> 13, 1000 <-> 14, 1000 <-> 15, 1000 <-> 16,
1000 <-> 17, 1000 <-> 18, 1 <-> 19, 1 <-> 20, 1 <-> 21, 1 <-> 22,
1 <-> 23, 19 <-> 24, 19 <-> 25, 19 <-> 26, 2 <-> 27, 2 <-> 28,
3 <-> 29, 4 <-> 30}

MB1965

Interesting. That long chain gets stuck on an end.

Well I'm writing an update for that anyway

ralph

21:27

Sorry but, I have to leave the forum for some time. Also I will be looking for an error in the code

MB1965

No need

I think I've found it.

I'll just edit the answer

The issue was that the central nodewas 1000

Not 1

And at one point I used the 1

I mean this is really going beyond what I should be providing you, but maybe future users will be able to find something of use in the morass

MB1965

22:01

@ralph I updated the stuff. It works on all your test cases now. So that's cool. I often feel like you don't actually know what the graph transformation is supposed to be--like I'm doing the math for you--so if that's the case please read my answer closely so you can understand that too. I can't promise I'll be willing to answer more of your questions, but if you have them ask me here, not in the comments.

Transcript for

Room for question 139896: recreating-…