summaryrefslogtreecommitdiffstats
path: root/dis/original/inputs/WSSS
diff options
context:
space:
mode:
authorJoshua Bakita <jbakita@cs.unc.edu>2020-10-17 00:52:09 -0400
committerJoshua Bakita <jbakita@cs.unc.edu>2020-10-17 00:52:09 -0400
commit917499f6257ac51c05e8302af877d56a22f28cb5 (patch)
tree880ad3fed41cfa521deb96b81a5fb89227a31e81 /dis/original/inputs/WSSS
parent6ea9939e0610a809f6f47d13ec68df00d1ca0afc (diff)
Cleanup in extra.h for non-paired benchmarks
- No parameter or output format changes - Better error messages - Auto-detect core - Fix off-by-one error in job counting - Explicitly invoke the scheduler with sched_yield() between jobs so that there's less work to do when (if) the scheduling timer interrupt comes around. (This has a surprisingly significant impact on timings.)
Diffstat (limited to 'dis/original/inputs/WSSS')
0 files changed, 0 insertions, 0 deletions
ef='#n86'>86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108











































































































                                                                               
% function [v,s,na,d] = ncutd(a,nv,beta,sigma,offset) 
% Input:
%    a = (attraction + i*repulsion), both nonnegative but could be asymmetrical
%    nv = number of eigenvectors v
%    beta = weighting between undirected graph and directed graph
%        beta = 1, undirected; beta = 0, directed.  Default = 0.5
%    sigma = refer to EIGS.M, 0 for smallest, default = 'LR'
%    offset = vector, each for an affinity matrix, 
%       per_edge offset for nondirectional attraction and repulsion
% Output:
%    v = generalized eigenvectors of A and D
%    s = eigenvalues
%    na = normalized affinity matrix