Number of profiles at N degrees from a profile P0

Question

Quick question to our software dream team. 

Would it be possible to compute the size of a "circle" (or, more exactly "sphere"), meaning the number of profiles at distance N (measured in degrees) of a given "center" profile P0?

I'm sure it's possible, at least in batch, if not on demand, since we have already those distances computed every night. But is it easy and not too costly in resources?

I have an ongoing work for which the distribution of such numbers for a given "center" would be of heavy interest. I will present this work in G2G when it's a bit more advanced.

Thanks for your attention.

[edited : added link to Jean-Joseph Vatant profile, of which "circles" are under investigation at http://goaf.fr/jjv/cercles.html (in French so far)]

Comments

Interesting question. I took the liberty of adding tags.

It would be fun to see how some "circles" expand quickly and perhaps others, less quickly (I happened to add a couple of people in the Montgolfier family yesterday and they are so endogamous - or were, up to the end of the 20th century).

---

Thanks Isabelle for the head up and adding tags. Since you read French, have a look at what I am about and have started to investigate at http://goaf.fr/jjv/cercles.html. The Jean-Joseph "circles" are under systematic exploration, more to come on this page about distribution models I'm scribbling those days over my morning coffee. Beware I've put back on my maths teacher hat, heady stuff coming.

---

I assume that each person-node is only counted once? So that, for example, a parent is not also counted as a spouse of the other parent and a child is not also counted as a sibling of another child.

An extremely simplified case where everybody marries, once and only once and every marriage results in three children should be easy to calculate (and perhaps to visualize). An actual case is of course vastly more complicated.

---

Indeed Eva, no double count. And my use case is quite tricky indeed. Jean-Joseph first circles are very endogamic, with e.g. widows marrying brothers/sisters-in-law, double marriages (siblings marrying the sibling next door, like my grandparents), child of first marriage married with brothers/sisters-in-law, and a lot of marriage between 3rd-4th-5th cousins. 

So I maintain a spreadsheet outside WikiTree, with every profile in a single circle (the smaller one). One quick check-up is given by the connection finder. If distance is 3, the profile belongs to 3rd circle. 

That, with an exhaustive (as far as possible) exploration starting from circle 1 allows to avoid duplicates. Of course this will not scale beyond the few first circles, unless a large task force is gathering around this project, which I doubt. But I have already a cousin helping, hopefully others will join in the course of time. We hope to have mostly completed the third circle by the end of 2020 (about 300+ people, over 200 of them already in WT). The fourth one should take less than a year ...

---

That's a worthy enterprise! And a good use of the connection finder.

With the fourth circle one end is going to dip into the realm of the living, isn't it?

---

Yes indeed. I for one sit in the fourth circle, as well as quite a few living siblings, cousins etc. There's even, according to one of my cousins, a last living member of the third circle, born around 1930. And there are, certainly, members of the sixth circle yet to be born.

There will be for a certain amount of time forbidden expansion paths due to privacy concerns. But that's the rule of the game. Those expansions will happen later on, hopefully, thanks to following generations. The distribution model has to take this into account.

---

@Eva again : looking at your great-great-grandfather profile Petter Andreasson, contemporary of Jean-Joseph, I think we could do the same kind of counting starting from him, and usefully compare the results. 

---

Haha! I was thinking of Olof Andersson. Slightly earlier and in a place where records go far back.

We'll see.

---

I did start on Olof Andersson. It turned out that I had only profiled one of his ten siblings, so that is what I filled my evening with yesterday. His first shell of surrounding profiles contains two parents, ten siblings, one spouse and eight children, that is, all in all 21 nodes.

Beyond the fourth shell the sphere will surely flatten out, because the descendant part will dip into the future and the ancestor part will dip into the unrecorded past - what remains will be the sideways extensions. These are a respectable part of what constitutes the global tree, but perhaps do not say so much about the network of the core person.

---

Eva, good to see someone else taking this crazy path. It would be interesting to ask Aleš to run the same query he did for Jean-Joseph (see below), so that we can compare the distances distribution.

I would gladly make this comparison available at http://goaf.fr/jjv/cercles.html. I just added at the bottom of this page a couple of graphics based on figures provided yesterday by Aleš, showing the local and global distributions.

---

It's not so crazy - I'm adding profiles every day, anyhow. This is not a bad organizing principle. I think it's too early for a bigger comparison, though. I have explored my ancestors as fat as I can safely go (don't we all) and I have explored my own second, third and fourth cousins - and then I  have been working on other things. So for this purpose there is still a lot to fill in.

---

I'm not sure it's too early. I'm also just beginning, so to speak, the systematic population of first circles, and it's interesting to see the state of affairs at the beginning. I've seen that Olof seems to stand at distances from the bulk of WikiTree similar to the ones of Jean-Joseph. My hunch is that the global distribution would be quite similar for our two ancestors beyond circle 20 or so. Only the local distributions would reflect the state of our respective work on our "local families", and I would bet they are not very different.

---

It sounds very likely that Olof an Jean-Joseph are at similar distances from the mainstream, yes.

If this is your interest it's OK with me if you ask Ales for the full set (as below) for Andersson-5056 even if t's a bit beyond my current focus - I think what I can get on Tuesday mornings (or so) from WikiTree+ will suffice for quite some time.

---

Hello, Bernard.  I know you have said you will present your work in G2G at a later time.  But, because you have already discussed it here, I wonder if you would be interested in setting up a free-space page about the project.  The purpose I'm envisioning would be to create a more permanent WikiTree location for discussion of the project, rather than a G2G discussion which will quickly fade from view.  It could be linked to this discussion, etc.  I'm thinking it could include an "FAQ" (frequently asked questions), so that if people have questions about your work, you only have to answer them once.  I'd be happy to help set up and maintain such a page, as long as you will give me a little help, such as explaining what "fly paws" (pattes de mouche) have to do with anything.  

---

Julie, I've been reluctant so far to create any free-space page, but it would be a good idea indeed insofar as some people here seem interested. This page should be in English, and linked to my personal page which will stay in French.

I indeed somehow remember to have written "pattes de mouche" somewhere in this forum, but where and when?

https://en.wiktionary.org/wiki/pattes_de_mouche 

---

It wasn't in the forum, it was in your Circles page.  Paragraph 9, about the birth of Jean-Joseph.  I could not find a good translation, but did wonder if it was similar to the English expression "chicken scratch," meaning hard-to-read handwriting.  You have confirmed that.  

Edited to correct typo.

---

Oh well. No bad memories. Supposed to be a precursor of some old age disease, but can't remember which one at the moment.

---

Imagine how I feel!    Having to try and understand French, mathematics, and WikiTree apps all at the same time!

---

A freespace page sounds like a good idea.

And in Swedish it's "kråkfötter" - crow's feet.

---

Wow, that "Cercles" page is impressive. Among many things, the description of WikiTree is excellent (for the French readers, who most likely never heard of it) and your writing style very entertaining. And I enjoyed the maths talk.

---

Thx Isabelle. So I have at least one French reader . 

---

Something new closely related to this:

https://www.wikitree.com/g2g/1414499/reporting-your-number-of-connected-profiles-to-what-degree

Answer 1

You can see all the profiles including sizes of each generation on WikiTree+ Here are the first 20 ones.

https://wikitree.sdms.si/default.htm?report=srch2&WikiTreeID=Vatant-5&Generations=20

Is that what you were asking for?

Comments

To be honest, it takes 7 seconds to compute and store all distances for each profile to the disk. Just checking 208 million relations takes 3 seconds.

I have to brag a little, since I wrote such a fast program.

---

 I must admit it's impressive. But I'm not surprised you could do it. And I will not hesitate to ask for 7 seconds once in a while. Let's say, every time a new million milestone is passed by the Single Tree. 21 million soon, BTW. This is every four months or so. Are you OK with that deal?

---

sure. Just send me the an email with link.

---

Just let me confirm my understanding: the WikiTree connection finder counts siblings as one step apart, whereas in terms of WikiTree+ generations siblings are in the second generation (two steps from the target person = going through a parent). Do I have it right?

---

That's the way I get it also, Eva. Of course it would be better if the two apps were consistent.

---

Yes. And the other difference is that CF is using also private profiles/connections.

---

Ales, since you are here, did you notice our other exchange with Eva above? She would be (and I would also) interested to compare the distribution of Jean-Joseph's circles with the ones of her ancestor [[Andersson-5056|Olof  Andersson (1793-1860)]]. Could you run your magic query on this profile too?

---

I suspect my Olof Andersson would give a very similar result to Jean-Joseph Vatant on the "big Aleš" test, since they are both from the outskirts of the Global Tree.

I wonder what it would look like for someone more central.

---

Results for /tree.json?userid1=12412011&debug=1

"0:1:1:0",

"1:21:22:12",

"2:44:66:449",

"3:86:152:1114",

"4:147:299:2198",

"5:258:557:3771",

"6:447:1004:6714",

"7:585:1589:12555",

"8:847:2436:18686",

"9:1302:3738:28176",

"10:1568:5306:42735",

"11:2174:7480:59588",

"12:2338:9818:81320",

"13:2988:12806:101929",

"14:3465:16271:129269",

"15:3500:19771:161448",

"16:3996:23767:195599",

"17:5303:29070:234105",

"18:7082:36152:289305",

"19:10924:47076:365136",

"20:18794:65870:491006",

"21:36923:102793:724149",

"22:78016:180809:1213771",

"23:171587:352396:2302928",

"24:348179:700575:4787465",

"25:595126:1295701:9766527",

"26:905981:2201682:17855272",

"27:1287646:3489328:29529626",

"28:1659003:5148331:45271595",

"29:1917002:7065333:64320039",

"30:2003687:9069020:84929564",

"31:1913671:10982691:104997592",

"32:1703069:12685760:122906902",

"33:1467013:14152773:138021281",

"34:1240659:15393432:150831833",

"35:1040351:16433783:161720025",

"36:863119:17296902:170926902",

"37:707666:18004568:178600550",

"38:577623:18582191:184864623",

"39:460841:19043032:189924047",

"40:363905:19406937:193923291",

"41:282549:19689486:197030059",

"42:219287:19908773:199392132",

"43:171888:20080661:201202169",

"44:134965:20215626:202616536",

"45:106291:20321917:203720804",

"46:84412:20406329:204590467",

"47:69814:20476143:205277457",

"48:57759:20533902:205848188",

"49:47889:20581791:206317882",

"50:39974:20621765:206699761",

"51:31867:20653632:207023010",

"52:26552:20680184:207267926",

"53:21071:20701255:207474759",

"54:17474:20718729:207634699",

"55:14554:20733283:207767634",

"56:12039:20745322:207879086",

"57:10153:20755475:207970323",

"58:9064:20764539:208048307",

"59:7758:20772297:208116738",

"60:7613:20779910:208176231",

"61:6930:20786840:208242313",

"62:5085:20791925:208299922",

"63:4004:20795929:208337194",

"64:3029:20798958:208364157",

"65:2197:20801155:208385734",

"66:1947:20803102:208401378",

"67:1777:20804879:208418120",

"68:1913:20806792:208433853",

"69:1593:20808385:208449452",

"70:1391:20809776:208460120",

"71:1465:20811241:208470873",

"72:1193:20812434:208482140",

"73:1098:20813532:208490606",

"74:862:20814394:208499175",

"75:769:20815163:208504087",

"76:664:20815827:208508773",

"77:447:20816274:208511953",

"78:344:20816618:208513847",

"79:297:20816915:208515193",

"80:304:20817219:208516504",

"81:156:20817375:208517779",

"82:118:20817493:208518266",

"83:97:20817590:208518785",

"84:64:20817654:208519132",

"85:55:20817709:208519370",

"86:31:20817740:208519529",

"87:37:20817777:208519698",

"88:57:20817834:208519958",

"89:41:20817875:208520237",

"90:52:20817927:208520438",

"91:63:20817990:208520674",

"92:76:20818066:208521091",

"93:109:20818175:208521560",

"94:101:20818276:208522300",

"95:59:20818335:208522620",

"96:46:20818381:208522819",

"97:38:20818419:208523085",

"98:17:20818436:208523319",

"99:20:20818456:208523431",

"100:21:20818477:208523525",

"101:12:20818489:208523673",

"102:5:20818494:208523710",

"103:3:20818497:208523718",

"104:1:20818498:208523722",

"105:1:20818499:208523724",

"106:1:20818500:208523726",

"107:1:20818501:208523728",

"108:1:20818502:208523730",

"109:1:20818503:208523732",

---

 more numbers to crunch ...

First analysis :

- At large scale, distributions are pretty much similar, with a median / peak value at d=30 +/- 1

- The Olof distribution is a bit sharper than Jean-Joseph, passing 10% at d=23 vs d=27 and 90% at d=38 vs d=40

- The first circles have a more regular growth for Olof.

Those are basically details. The global geometry is really similar, without surprise.

Answer 2

This question sank very rapidly in the flow. I think your tags did not draw the attention of the people who could answer. I'm not sure I know better tags - I do believe "aleš" and "jamie" exist as tags. Just not sure if it is too "attention-seeking" to use them here.

Comments

Thanks Eva. Difficult indeed to know who to attract.and how ...

Answer 3

You can get the number of ancestors/ relatives via the API.

https://www.wikitree.com/wiki/Help:API_Documentation#getAncestors

If you know which you want to include in your circle, you could repeat the call for N-1, thus fill your circle.

But I reckon this will be intensive if you want to do this continuously. Then it might be better to use the data dumps and do the calculation on your own dataset.

I would suggest to add the tag 'wt_apps' to your post, or maybe even join the group.

Comments

Thanks Michel. I'm not a techie myself (not a coder, that is), so I would not be very useful in the group I'm afraid beyond asking questions. Since you read French also (I guess) see the link in the above answer to Isabelle to figure what I am about.

---

If you have specific profiles you are interested in, you could download the Gedcom with them as the starting point and import it into Gramps. There you have the option to plot the circle (and more). There it is called a Fan Chart. I promise: low tech

But your question is assuming that you want this on the fly which will require some coding.

---

Thx Michel. I don't want to mingle with GEDCOM, one of the ugliest data format I've ever seen in about 15 years I'd been involved in data migration.  Actually since I'm retired I'd sworn to myself to crunch no data no more. I'll have a look to Gramps that I don't know. 

I don't want to have those figures on the fly. An idea of the distribution at some point in time would be enough. Refresh those figures once a month, or even once a year, would be OK. More to come ...

---

Oh, here we welcome retired people as they have loads of time . I can't agree more on Gedcom... but there is no universal alternative. In my country we have one for documents, but that needs updating as well. API is the way to go really.

I forgot to check the current apps page, but there is a fan chart one: https://www.wikitree.com/wiki/Help:Apps

by Greg. https://apps.wikitree.com/apps/clarke11007/fan.php

Maybe that is, for now, what you would want?

---

Actually the "circles" defined by the distance in degrees include ancestors, descendants and collateral extensions, following the four directions of connection: parent, child, sibling and spouse. I don't think the fan charts expand in those four directions.

---

No, it doesn't. It would make it a bit hard to visualize.. but also interesting. I am wondering if the Topola Viewer could be expanded to do this in a visual way.

Maybe one of the active developers might be interested?

---

Is this what you mean?

https://apps.wikitree.com/apps/wiech13/

---

Try the Family Circles visualization within WikiTree+:

https://wikitree.sdms.si/default.htm?report=disp&WikiTreeID=Crawford-7109&Generations=5&OnlyBlood=0

Not exactly what you're saying, but similar

---

Jonathan, that sounds interesting, but the page does not seem to load in my browser (Chrome)

---

@Michel, Julie and Jonathan : to be clear I don't care about any kind of fancy visualization (none of them scales beyond 100 nodes). Just a flat distribution table (distance, number of profiles) would make me more than happy.

---

Ok, try the Family List page in the Genealogy Research section on the profile (Family Tree & Tools, click on Genealogy Research, scroll down to Family List)

You can select how many generations, include ancestors/descendants/both, include siblings.

Not perfect, but closer.

https://www.wikitree.com/index.php?title=Special:FamilyList&p=8822218

---

Indeed, this is useful, but expansion of circles goes also through spouses and their families. That's how connection distance (degrees) in WT is computed.

For example, the family list for Jean-Joseph on 3 generations including ancestors, descendants and their siblings as of today, counts 192 profiles. In my local spreadsheet adding expansions through spouses, I count 354. And this is only the 3rd circle ...

---

@Jonathan,

that is new to me, and sounds very interesting (but is also not working for me on FF latest)