29 lines
1.1 KiB
Matlab
29 lines
1.1 KiB
Matlab
|
% Sort nodes by degree, and when there's equality,
|
||
|
|
% by sum of neighbor degrees and then neighbors' neighbors degree and so on
|
||
|
|
% Ideas from s-max algorithm by Li et al 2005 "Towards a theory of scale-free graphs"
|
||
|
|
% and Guo, Chen, Zhou, "Fingerprint for Network Topologies"
|
||
|
|
%
|
||
|
|
% INPUTS: adjacency matrix, 0s and 1s, nxn
|
||
|
|
% OUTPUTS: sorted (decreasing) sequence (nx1), where n is the number of
|
||
|
|
% rows/cols of the adjacency
|
||
|
|
%
|
||
|
|
% Other routines used: degrees.m, kneighbors.m
|
||
|
|
% GB: last update, Oct 4, 2012
|
||
|
|
|
||
|
|
|
||
|
|
function I=sortNodesBySumNeighborDegrees(adj)
|
||
|
|
|
||
|
|
[deg,~,~]=degrees(adj); % compute all degrees, use "deg" assuming symmetry
|
||
|
|
|
||
|
|
degmat=zeros(size(adj,1),size(adj,1)); % nxn matrix [index i,sum(deg of neighbors of index i, <=j links away)]
|
||
|
|
|
||
|
|
for x=1:size(adj,1) % across all nodes
|
||
|
|
degmat(x,1)=deg(x);
|
||
|
|
|
||
|
|
% across all threshold distances, 2 to n-1 [should really be 2 to diameter, if the diameter is known]
|
||
|
|
for kk=2:size(adj,1); degmat(x,kk)=sum( deg(kneighbors(adj,x,kk-1)) ); end
|
||
|
|
|
||
|
|
end
|
||
|
|
|
||
|
|
[sortmat,I]=sortrows(degmat);
|
||
|
|
I=I(end:-1:1);
|