To:ljporter
cc:
Subject:
--------

Hi Lisa,

I fixed the problem of many zero's. As you pointed out, it's because
the function jacobi requires two N's. One is the usage size, the other
is the memeory allocation.

I put everything in /tmp/Dyn/ on mightylp. The main program is
"dyn.f". It can read "config" from df.f. The only difference here with
my previous program is that I changed NM, the maximum number of
particles, to 250, in order to save memory. So the NM in corresponding
df.f or mn.f, if you want to use their "config"s, should also be 250.

The input file "con" for dyn.f is

0/1    (0: read from config.  1: construct perfect crystal)
4.29   (size of unit cell)
216    (# of particles)
0/1    (0: generate DOS for disordered cell.  1: DOS for perfect crystal) 
200    (How many bins you want to devide up DOS)
20000  (How many k points you want to sample using Monte Carlo. Useless for the disordered 
       case)

After running "dyn < con". You can directly get density of states plot
by running the Matlab program "DOS.m" on Matlab, both for the
disordered and crystalline case.

In order to get the dispersion curve, use the following "con"

0 4.29 216 1 30 100

It will very soon finish up the 100 Monte Carlo k-points. It then will
always give a standardized output file "dm.out", which is used by the
Matlab program "dispersion.m", and "phonon.m". So just "dyn < con",
and type ">dispersion" on Matlab, you'll get the dispersion curve
plot.  It will have some problems printing, though. To get the
frequencies of a speceific k-point, type

>load dm.out
>phonon ([0.1 0.2 0.3])



Li Ju.