[igraph] Negative eigenvalue centrality

[Michael Ewens]

I have encountered this bug:

http://lists.gnu.org/archive/html/igraph-help/2009-11/msg00045.html

where the output of evcent(g)$vector is in [-1,1].  When I run a similar 
centrality measure like authority.score(g)$vector, the results are all 
non-negative.  I tried the suggestion in the above thread to no avail: I 
still get values in -1 to 1.  Any other suggestions?

Mike