Whilst I’m in the area of ‘influences’ I thought it would be interesting to examine a few subnetworks within the large network of everyone. This time I set my scopes on *mathematicians*. There is no primary reason why other than – *I can*. I’ve long been interested in the history of mathematics and so I wondered what a network of great mathematicians actually looked like? Could there be underlying structures between mathematicians who have influenced each other over history? This time I used Freebase which has an excellent query system which enables you to pull out information on pretty much… anything! I focused on the Influence Node for this task. I set a few filters such as; everyone in the network has to have ‘Profession’ = ‘Mathematician’. This removes a lot of fluff and creates a nice csv file which is useable within Gephi. As usual I had to do a bit of cleaning up in Microsoft Excel but here you have it (click images to enlarge or click here for dynamic zoom of the entire network):

The Graph Of Mathematicians: careful how you interpret this one!

Interestingly, after applying the Modularity, a few little subnetworks within the mathematician network appear. Many of the ‘traditional’ mathematicians cluster together; i.e. Gauss, Jacobi, Riemann, Dedekind etc.

The “No Surprises” Section

Arab, Persian and Muslim mathematicians cluster together.

Applied mathematicians (physicists/astronomers) are also clustered together.

The geometers.

The logicians.

There are other networks – see if you can spot why certain names are clustered together (e.g. linguistics, algebra, philosophy etc).

Again before you leave, you must be made aware of the caveats!

1. There are only nodes for which there was available data. There are obviously a great number of influential mathematicians missing from the network.

2. I had to restrict to Profession = Mathematician and this might be too restricting in that if a mathematician in the database doesn’t have their profession set, then they would have been excluded.

3. There may be some contamination as I found Ralph Emerson, Mohammed and Buddha in amongst the mathematicians which I hate to say it – just aren’t. I tried to clear up these contaminants but there still might be residuals. Feel free to point these out.

4. It is a graph of influence *between people*, **not** a graph of how influential they were on mathematics!

Onwards and upwards.

### Like this:

Like Loading...

*Related*

Tags: gephi, mathematicians, mathematics, networks

Where is Erdos?

Exactly my thought! Again, this is an incomplete data set and only designed to show what is possible with more complete data in the future. There are a great number of people missing from the graph.

Bah, Erdos – where the eff is Grothendieck?

Yes I know! Remember though, it is an influenced map, not an influential map! Still, data set is incomplete.

Great graph. Regarding Erdos, there is a concept called the Erdos number (anyone who co-authored a paper with him has a 1, if you co-authored a paper who co-authored with him you get a 2, etc). See http://www.oakland.edu/enp/ for complete datasets if you are interested. The data quality should be pretty good.

Thanks for your comment Jim. I am aware of the Erdos number and that website you provided – I had forgotten how good the data was! I’ll be sure to check it out – thanks for the reminder. I’ve got a few ideas already. :)

Pretty cool! Still some non-mathematicians in there. I love Curie, but mathematician she was not. Also, despite his association with famous equations, Einstein was not a mathematician either.

What is your working definition of a mathematician? just curious!

maybe it should be ovious, but … what do the colours mean?

cheers

They are assigned by the

Modularity module of Gephi. It is just to highlight more closely connected people. It isn’t a clear cut boundary. The colour choice was random.

Why don’t you use the mathematician’s genealogy project http://www.genealogy.ams.org/ ?

The mathematics genealogy is a family tree rather than a network so other than what they have already done in terms of diagrams I’m not sure what else I can add. The best type of data to visualise is when every component relates to every other component in a more networked style.