John D'Errico

This file has been on the FEX for a while without much attention, so let me take a look.

It is a pretty specialized area, where I am not an expert, so I would not rate this tool on its intended usage. (Some of the pieces are pretty generic, so the eigen.m code is something I could talk about. It was well written.)

However, I can talk about the overall tool itself, in terms of the code. Every part I looked at had excellent documentation. There were internal comments, H1 lines, all things I like to see. Where appropriate, I found matrix operations instead of loops, etc.

Note that:

- Fully half the code is commented out.
- It starts with a clear all, that is commented out, but is unnecessary in a function.
- There is NO documentation, except for the trivial line: "How does it work", which does not actually tell how it works!!!!!
- The line "for i = 1" effectively executes "i = 1", but takes more programming effort, and confuses things. Why add an end statement when none is needed?

%clear all;
function widthA(d)

%how does it work?
%widthA(d)
%d=[2 2 4 3 6 4 3 2 7 9 5 7 18];
% where d is your data
N=length(d);

%i=1;
%SD1 = d(:,i:N-1)
%SD2 = d(:,i+1:N)
%d1=SD1-SD2

for i = 1
SD1 = d(:,i:N-1);
SD2 = d(:,i+1:N);
d1=(abs(SD1))-(abs(SD2));
%corr(SD1(:,i:N-1),SD2(:,i:N-1))
%plot(ds)
end
df = max(d1)
%ad = min(d1)

[r,c,v] = find(d1==df);
%d1n = find(min(d1))
fl1=d(1,c)
fl2=d(1,c+1)

This is exactly what I needed. Unfortunately, it doesn't seem to work on eigenvalue problems where the crossings happen over a wide range, or where they cross and then cross back. For my system, this really only worked for small crossings.