517. Very briefly: Prime95 (GIMPS) on linux

I'm very unhappy about a newly built node which randomly crashes and reboots when running long jobs. More about that later, but here are the specs: FX 8350, 4x8 Gb RAM GSkill Ripjaws, ASRock FX990 Extreme3, Corsair GS700, MSI N210, ASUS NX1101 in an Antec GX700 case, running Wheezy with stock kernel (3.2.0-4 amd64).

I've tested the RAM using memtest86+  and found no errors, the rig uses a 700 W Corsair PSU which /should/ provide enough power, and I see no evidence of overheating based on a cronjob which runs every 2 minutes. Anyway, the first step in troubleshooting is finding a good way of reproducing the error reliably, and prime95 is what the windows overclockers use to stresstest.

Turns out prime95 (actually GIMPS) can run in a few different modes which tests different aspects of you system, which makes it sound like a pretty good program for my purposes.

See here for more information: http://www.mersenne.org/freesoft/

---

mkdir ~/tmp/mprime -p
cd ~/tmp/mprime
wget http://www.mersenne.info/gimps/p95v279.linux64.tar.gz
tar xvf p95v279.linux64.tar.gz
./mprime

Welcome to GIMPS, the hunt for huge prime numbers.  You will be asked a 
few simple questions and then the program will contact the primenet server 
to get some work for your computer.  Good luck!

Attention OVERCLOCKERS!!  Mprime has gained a reputation as a useful 
stress testing tool for people that enjoy pushing their hardware to the 
limit.  You are more than welcome to use this software for that purpose.  
Please select the stress testing choice below to avoid interfering with 
the PrimeNet server.  Use the Options/Torture Test menu choice for your 
stress tests.  Also, read the stress.txt file.

If you want to both join GIMPS and run stress tests, then Join GIMPS and 
answer the questions.  After the server gets some work for you, stop 
mprime, then run mprime -m and choose Options/Torture Test.

Join Gimps? (Y=Yes, N=Just stress testing) (Y): N
Number of torture test threads to run (3): 2
Choose a type of torture test to run.
1 = Small FFTs (maximum FPU stress, data fits in L2 cache, RAM not tested 
much).
2 = In-place large FFTs (maximum heat and power consumption, some RAM 
tested).
3 = Blend (tests some of everything, lots of RAM tested).
11,12,13 = Allows you to fine tune the above three selections.
Blend is the default.  NOTE: if you fail the blend test, but can pass the 
small FFT test then your problem is likely bad memory or a bad memory 
controller.
Type of torture test to run (3): 1
Accept the answers above? (Y): Y
[Main thread Sep 20 11:06] Starting workers.
[Worker #1 Sep 20 11:06] Worker starting
[Worker #1 Sep 20 11:06] Setting affinity to run worker on any logical CPU.
[Worker #2 Sep 20 11:06] Worker starting
[Worker #2 Sep 20 11:06] Setting affinity to run worker on any logical CPU.
[Worker #1 Sep 20 11:06] Beginning a continuous self-test to check your computer.
[Worker #1 Sep 20 11:06] Please read stress.txt.  Hit ^C to end this test.
[Worker #2 Sep 20 11:06] Beginning a continuous self-test to check your computer.
[Worker #2 Sep 20 11:06] Please read stress.txt.  Hit ^C to end this test.
[Worker #1 Sep 20 11:06] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
[Worker #2 Sep 20 11:06] Test 1, 180000 Lucas-Lehmer iterations of M580673 using AMD K10 type-1 FFT length 28K, Pass1=112, Pass2=256.
CTRL+C

And so on.

516. Potential energy diagrammes in latex using pgfplots

Given these recent posts, a post on how to represent reaction coordinates in latex was inevitable.

Note: a much prettier example is found here: https://quantumchemistryniser.wordpress.com/2013/05/25/potential-energy-profile-using-pgfplots/

However, I initially couldn't reproduce it, and after spending a few hours with pgfplots I wanted my own style anyway.

---

But first, a recap:
Once you've completely left office-like wordprocessors behinds and fully embraced the wonder of typesetting in LaTeX you find yourself wanting to produce more and more of the figures directly in latex, rather than simply importing figures.

While nothing beats gnuplot for plotting, and pov-ray (via e.g. gdis) for fancy figures and bkchem for simple 2d structures, there are some nice and simple latex packages that make life easier: I touch on a few in this post: http://verahill.blogspot.com.au/2012/07/debian-texlive-and-style-files-making.html

For chemical formulae, use mhchem -- I think it's standard in debian for texlive. It's indispensable. Also see http://www.ctan.org/pkg/mhchem

For very, very simple chemical structures you can look at chemfig -- I don't use it much though. http://www.ctan.org/tex-archive/macros/latex/contrib/chemfig/

For drawing simple MO diagrammes, modiagram is very pretty: http://www.ctan.org/pkg/modiagram

For drawing NMR pulse sequences there are a few options: http://verahill.blogspot.com.au/2013/08/499-briefly-drawing-nmr-sequences-using.html and http://verahill.blogspot.com.au/2013/08/498-briefly-drawing-nmr-pulse-sequences.html

---

A dead end: endiagram
It looks very promising, but it didn't work out for me: http://ctan.mirrorcatalogs.com/macros/latex/contrib/endiagram/

Installation:

mkdir ~/texmf/tex/latex/endiagram -p
cd ~/texmf/tex/latex/endiagram
wget http://ctan.mirrorcatalogs.com/macros/latex/contrib/endiagram/endiagram.sty
sudo texhash

You'll need siunitx and tikz as well, but it seems to come with a standard texlive install on debian. Out of the box I kept getting:

! Undefined control sequence.

<argument> \clist_count:N

\l__endiagram_points_clist

l.16 \ENcurve{1,4,0}

In the end I couldn't get it to work properly, even after installing l3kernel -- I don't think it would've been impossible to figure it out, but I was running out of patience.

---

The solution: pgfplots
So I kept looking, and found this: https://quantumchemistryniser.wordpress.com/2013/05/25/potential-energy-profile-using-pgfplots/

While I couldn't get it to work 'out of the box' (after deleting the nodes depending on the pdf files), I ended up reading the pfgplots manual: http://pgfplots.sourceforge.net/pgfplots.pdf

pgfplots is not meant for drawing potential energy diagrammes, but it can be made to do it -- it's 'dum' in the sense that it's not aware of what a PE diagram is, so you'll have to draw all objects expliticly. In the end, it's actually quite simple though.

I eventually ended up with these two examples:

The code is here:

\documentclass[10pt,a4paper]{article}
\usepackage{pgfplots}
\usepackage{tikz}
\begin{document}

%\begin{center}
%\begin{tikzpicture}[scale=1.0]
%\draw (0,0) --(1,2);
%\end{tikzpicture}
%\end{center}

\begin{tikzpicture}
\begin{axis}[
 axis x line= bottom,
 axis y line= left,
 xmin=-1,   xmax=10,
 ymin=-1,   ymax=6,
 xlabel=Reaction coordinate,
 ylabel=$\Delta$E (kcal/mol),
 xtick=\empty
]
 \addplot[smooth,solid,blue] coordinates { (0,0)(1,0) (2,4)(3,4) (4,3)(5,3) (6,4)(7,4) (8,1)(9,1)};
 \node at (axis cs:4.5,2.5) {Intermediate};
\end{axis}
\end{tikzpicture}

\begin{tikzpicture}
\begin{axis}[
 axis x line= bottom,
 axis y line= left,
 xmin=-1,   xmax=6,
 ymin=-1,   ymax=15,
 xlabel=Reaction coordinate,
 ylabel=$\Delta$E (kcal/mol),
 xtick=\empty
]
 \addplot[smooth,solid,blue] coordinates { (0,0)(1,0)};
 \addplot[smooth,dotted,black] coordinates {(1,0)(2,11.7)};
 \addplot[smooth,solid,blue] coordinates {(2,11.7)(3,11.7)}; 
 \addplot[smooth,dotted,black] coordinates {(3,11.7)(4,10.5)};
 \addplot[smooth,solid,blue] coordinates {(4,10.5)(5,10.5)};

\end{axis}
\end{tikzpicture}
\end{document}

515. Very briefly: boot issues and nvram (anecdote)

I had to move offices at work last Friday (which partly explains my radio silence).

One of my nodes, with a Biostar 880G+ motherboard, wouldn't boot. The issue manifested itself in a few different ways (and not always at the same time):

Most of the time the bios portion of the boot would be very slow, and I'd get a loooooong beep when the voltages were echoed during what I presume is POST. I never made it to the GRUB screen. I'd had this happen once or twice before, especially when the computer had been turned off and/or unplugged for a longer period of time.

Occasionally I'd get a message about NVRAM being checked..

After
1) having checked that nothing had vibrated loose during the transport
2) having removed all cards and drives except a single RAM stick
3) having checked that RAM stick
I still wasn't getting any closer to getting booted. So I went into the BIOS. Having seen the message about nvram and having done a bit of reading online I selected to erase the NVRAM.

And suddenly booting worked...once.

After that I kept getting messages about nvram being corrupted and boot fail, or 'NVRAM...Update failed'.

I then set the CMOS jumpers to pin 2 and 3 (small jumper next to the CMOS battery)

And then it worked! It booted just fine! But I kept getting the following error echoing on my screen:

[  120.718683] hub 3-0:1.0: unable to enumerate USB device on port 2
[  120.932378] usb 3-2: new full-speed USB device number 28 using ohci_hcd
[  121.072497] usb 3-2: device descriptor read/64, error -62
[  121.316673] usb 3-2: device descriptor read/64, error -62
[  121.556904] usb 3-2: new full-speed USB device number 29 using ohci_hcd
[  121.697019] usb 3-2: device descriptor read/64, error -62
[  121.941224] usb 3-2: device descriptor read/64, error -62
[  122.181425] usb 3-2: new full-speed USB device number 30 using ohci_hcd

Unplugged all the USB devices. Turned out to be the KVM (ritmo). Turning it off (i.e. unplugging all USB input to the KVM so that it lost power) and plugging it in again solved it.

And ever since then all has been fine.

Presumably it's time to change the CMOS battery.

514. Extracting Frequency data from a gaussian 09 calculation for gnuplot

This is another python script.

Say you've done a computation along the lines of this:

#P rBP86/GEN 5D Pseudo(Read) Opt=() Freq=() SCF=(MaxCycle=999 ) Punch=(MO) Pop=()

and want the data in a neat data file, like this:

33.237 0.0023 0.0536
39.9976 0.0043 0.8305
69.7345 0.0129 0.3348
84.7005 0.0173 0.7027
[..]
3133.0068 6.2938 0.6114
3143.8021 6.3551 0.3775
3164.9242 6.4685 0.8829
3221.8787 6.6972 4.6005

Then you can use the following python (2.x) script, g09freq:

#!/usr/bin/python
# Compatible with python 2.7 
# Reads frequency output from a g09 (gaussian) calculation
# Usage ex.: g09freq g09.log ir.dat
import sys 

def ints2float(integerlist):
    for n in range(0,len(integerlist)):
        integerlist[n]=float(integerlist[n])
    return integerlist

def parse_in(infile):
    g09output=open(infile,'r')
    captured=[]
    for line in g09output:
        if ('Frequencies' in line) or ('Frc consts' in line) or ('IR Inten' in line):
            captured+=[line.strip('\n')]
    g09output.close()
    return captured
    
def format_captured(captured):
    vibmatrix=[]
    steps=len(captured)
    for n in range(0,steps,3):
        freqs=ints2float(filter(None,captured[n].split(' '))[2:5])
        forces=ints2float(filter(None,captured[n+1].split(' '))[3:6])
        intensities=ints2float(filter(None,captured[n+2].split(' '))[3:6])
        for m in range(0,3):
            vibmatrix+=[[freqs[m],forces[m],intensities[m]]]
    return vibmatrix

def write_matrix(vibmatrix,outfile):
    f=open(outfile,'w')
    for n in range(0,len(vibmatrix)):
        item=vibmatrix[n]
        f.write(str(item[0])+'\t'+str(item[1])+'\t'+str(item[2])+'\n')
    f.close()
    return 0

if __name__ == "__main__":
    infile=sys.argv[1]
    outfile=sys.argv[2]

    captured=parse_in(infile)

    if len(captured)%3==0:
        vibmatrix=format_captured(captured)
    else:
        print 'Number of elements not divisible by 3 (freq+force+intens=3)'
        exit()
    success=write_matrix(vibmatrix,outfile)
    if success==0:
        print 'Read %s, parsed it, and wrote %s'%(infile,outfile)

Run it as e.g.

g09freq g09.log test.out

The output is compatible with gnuplot:

gnuplot> set xrange [3500:0]
gnuplot> set yrange [10:-1]
gnuplot> plot './test.out' u 1:2 w impulse

It's trivial to add gaussian broadening (see e.g. this post)

513. Extracting data from a PES scan with gaussian

There are a few reasons to like gaussian, and many reasons not to. Gaussian is fast, and their whitepapers are great resources for learning computational techniques.

Without going into discussions about the commercial behaviour of Wavefunction inc., the things I don't like about gaussian is the clunky input format (nwchem has a much more readable syntax), the inscrutable error messages, and the unreadable output. Well, it's not unreadable in a literal sense, but it could certainly be clearer. On the other hand, I've having issues with running some of my PES scans in nwchem -- and I can't find a solution (more about that in a later post)

Anyway, here's a python script for extracting optimized structures and energies from a relaxed PES scan in Gaussian 09.

First, an example of a simple scan:

%nprocshared=2
%Chk=methanol.chk
#P rB3LYP/6-31g 6D 10F Opt=(modredundant) NoSymm  Punch=(MO) Pop=() 

methanol

0 1 ! charge and multiplicity
 C     0.0351714     0.00548884     0.0351714
 H     -0.617781     -0.634073     0.667983
 H     0.667983     -0.634073     -0.617781
 H     -0.605139     0.646470     -0.605139
 O     0.839603     0.818768     0.839603
 H     1.38912     0.201564     1.38912

 1 5 S 10 0.1

And here's the script, pes_parse_g09:

#!/usr/bin/python
import sys

def getrawdata(infile):
        f=open(infile,'r')
        opt=0
        geo=0
        struct=[]
        structure=[]
        energies=[]
        energy=[]
        for line in f:
                
                if opt==1 and geo==1 and not ("---" in line):
                        structure+=[line.rstrip()]
                
                if 'Coordinates (Angstroms)' in line:
                        if opt==0:
                                opt=1
                                structure=[]
                        
                if opt==1 and "--------------------------" in line:
                        if geo==0:
                                geo=1
                        elif geo==1:
                                geo=0
                                opt=0
                if 'SCF Done' in line:
                        energy=filter(None,line.rstrip('\n').split(' '))
                if      'Optimization completed' in line and (opt==0 and geo==0):
                        energies+=[float(energy[4])]
                        opt=0
                        geo=0
                        struct+=[structure]
                        structure=[]
        
        return struct, energies

def periodictable(elementnumber):
        ptable={1:'H',2:'He',\
        3:'Li', 4:'Be',5:'B',6:'C',7:'N',8:'O',9:'F',10:'Ne',\
        11:'Na',12:'Mg',13:'Al',14:'Si',15:'P',16:'S',17:'Cl',18:'Ar',\
        19:'K',20:'Ca',\
        21:'Sc',22:'Ti',23:'V',24:'Cr',25:'Mn',26:'Fe',27:'Co',28:'Ni',29:'Cu',30:'Zn',\
        31:'Ga',32:'Ge',33:'As',34:'Se',35:'Br',36:'Kr',\
        37:'Rb',38:'Sr',\
        39:'Y',40:'Zr',41:'Nb',42:'Mo',43:'Tc',44:'Ru',45:'Rh',46:'Pd',47:'Ag',48:'Cd',\
        49:'In',50:'Sn',51:'Sb',52:'Te',53:'I',54:'Xe',\
        55:'Cs',56:'Ba',\
        57:'La',58:'Ce',59:'Pr',60:'Nd',61:'Pm',62:'Sm',63:'Eu',64:'Gd',65:'Tb',66:'Dy',67:'Ho',68:'Er',69:'Tm',70:'Yb',71:'Lu',\
        72:'Hf', 73:'Ta', 74:'W',75:'Re', 76:'Os', 77:'Ir',78:'Pt', 79:'Au', 80:'Hg',\
        81:'Tl', 82:'Pb', 83:'Bi',84:'Po',85:'At',86:'Rn',\
        87:'Fr',88:'Ra',\
        89:'Ac',90:'Th',91:'Pa',92:'U',93:'Np',94:'Pu',95:'Am',96:'Cm',97:'Bk',98:'Cf',99:'Es',100:'Fm',101:'Md',102:'No',\
        103:'Lr',104:'Rf',105:'Db',106:'Sg',107:'Bh',108:'Hs',109:'Mt',110:'Ds',111:'Rg',112:'Cn',\
        113:'Uut',114:'Fl',115:'Uup',116:'Lv',117:'Uus',118:'Uuo'}
        element=ptable[elementnumber]
        return element

def genxyzstring(coords,elementnumber):
        x_str='%10.5f'% coords[0]
        y_str='%10.5f'% coords[1]
        z_str='%10.5f'% coords[2]
        element=periodictable(int(elementnumber))
        xyz_string=element+(3-len(element))*' '+10*' '+\
        (8-len(x_str))*' '+x_str+10*' '+(8-len(y_str))*' '+y_str+10*' '+(8-len(z_str))*' '+z_str+'\n'
 
        return xyz_string

def getstructures(rawdata):
        
        n=0
        for structure in rawdata:
                
                n=n+1
                num="%03d" % (n,)
                g=open('structure_'+num+'.xyz','w')
                itson=False
                cartesian=[]
                        
                for item in structure:
                        
                        coords=filter(None,item.split(' '))
                        coordinates=[float(coords[3]),float(coords[4]),float(coords[5])]
                        element=coords[1]
                        cartesian+=[genxyzstring(coordinates,element)]
                g.write(str(len(cartesian))+'\n')
                g.write('Structure '+str(n)+'\n')
                for line in cartesian:
                        g.write(line)
                g.close()
                cartesian=[]
        return 0
        
if __name__ == "__main__":
        infile=sys.argv[1]
        rawdata,energies=getrawdata(infile)
        structures=getstructures(rawdata)
        g=open('energies.dat','w')
        for n in range(0,len(energies)):
                g.write(str(n)+'\t'+str(energies[n])+'\n')
        g.close()

And here's what we get from the output:

g09 methanol.in |tee methanol.out
pes_parse_g09 methanol.log
cat structure* > meoh_traj.xyz

And here's a plot of energies.dat:

512. Briefly: zmatrices in nwchem -- methanol

And another update:
I can now confirm that using your own z matrix still does not constrain the geometry during a PES scan, which was the original impetus for this post: http://verahill.blogspot.com.au/2013/09/511-when-nwchem-pes-scans-fail-to.html

Another update:
the gaussian run failed after 14 geometry steps during the first PES point.

NTrRot=    -1 NTRed=   628 NAtoms=    34 NSkip=   532 IsLin=F
 Error in internal coordinate system.
 Error termination via Lnk1e in /opt/gaussian/g09/l103.exe at Thu Sep  5 18:17:12 2013. 
 Job cpu time:  0 days 22 hours 25 minutes 27.6 seconds.
 File lengths (MBytes):  RWF=    192 Int=      0 D2E=      0 Chk=     28 Scr=      1 

Not being an expert, to me it seems that there's something fundamentally difficult with the system I'm working on. In an ideal world I'd give the actual details, but quite apart from the risk of being scooped, doing so would also make it easier to identify me (not that it's impossible at this point).

[Suffice to say that the system holds a large polyoxoanion and a small p-block anion, both of which are symmetrical and negatively charged. The goal of the PES scan is to bring the ions closer to see whether they 'react'. Which is also a troublesome use of computational resources -- computational chemistry is good at answering well-defined questions using carefully designed computational experiments -- but not generally very good at answering ill-defined questions about synthesis (i.e. you can't generally 'mix two things together and see what happens' and expect a useful result. Anyway, regardless of that, that's exactly what I want to do.]

Update:
nwchem still gives errors about autoz in spite of using noautoz. But I also get messages about the user generated z matrix, so we'll see whether my input is respected or not.

Also, for one of the calcs I'm getting

There are insufficient internal variables: expected   95 got   96

which is really, really, really annoying since there doesn't seem to be a real fix for it -- I've tried everything suggested in http://www.nwchem-sw.org/index.php/Special:AWCforum/st/id286. I can get the same calc to run in gaussian though (gaussian has its own issues), but it'd be nice if stuff just...worked...

Original post:
Normally you don't have to fiddle with zmatrices in nwchem -- instead you'd typically supply cartesian coordinates, and nwchem would do autoz to autogenerate internal (z matrix) coordinates.

Sometimes that fails, and nwchem defaults to using cartesian coordinates. In most cases, this isn't a cause for any real concern -- the computation will continue although I think cartesian coordinates are supposed to be slightly slower.

However, if you're doing a PES scan you'll notice that it's not proceeding as intended -- the constraints are completely ignored: 511. When nwchem PES scans fail to constrain -- autoz failure

The easiest remedy is to supply the internal coordinates directly, but there honestly aren't too many examples online showing how that's done, and I kept on getting annoying failure messages along the lines of

NWChem Input Module
                                -------------------


                                       zmat
                                       ----
 THE 3-D  PIECE OF -Z- DATA FOR ATOM =    2 IS NEITHER FLOATING POINT
 NOR ALPHANUMERIC OR COULD NOT BE MATCHED WITH A VARIABLE. STOP
 IAT=    2 ZMAT=  2  1  0  0  0     0.00000     0.00000     0.00000

 ------------------------------------------------------------------------
                               JOB STOPPED

 PROGRAM STOP IN - ZDAT -

 ------------------------------------------------------------------------
 ------------------------------------------------------------------------
  CALLS IT QUIT FROM HND_HNDERR         0
 ------------------------------------------------------------------------
 This error has not yet been assigned to a category

This particular error came about because the zmatrix module is case sensitive, and my Variables couldn't be interpreted (it should be variables). Anyway, you'll understand more after this post, and it isn't important anyway.

---

Calculation using a z matrix (internal coordinates) in nwchem, with a little bit of help from openbabel:

Assuming that you set up a calculation in e.g. ECCE for a geometry optimisation of methanol you'll end up with the following input file:

scratch_dir /home/andy/scratch
Title "methanol"

Start  methanol

echo

charge 0

geometry autosym units angstrom
 C     0.0351714     0.00548884     0.0351714
 H     -0.617781     -0.634073     0.667983
 H     0.667983     -0.634073     -0.617781
 H     -0.605139     0.646470     -0.605139
 O     0.839603     0.818768     0.839603
 H     1.38912     0.201564     1.38912
end

ecce_print ecce.out

basis "ao basis" spherical print
  H library "6-31+G*"
  O library "6-31+G*"
  C library "6-31+G*"
END

dft
  mult 1
  direct
  XC b3lyp
  grid fine
  mulliken
end

driver
  default
end

task dft optimize

Take the coordinates, and paste them into a file, e.g. methanol.xyz:

6
methanol
 C     0.0351714     0.00548884     0.0351714
 H     -0.617781     -0.634073     0.667983
 H     0.667983     -0.634073     -0.617781
 H     -0.605139     0.646470     -0.605139
 O     0.839603     0.818768     0.839603
 H     1.38912     0.201564     1.38912

Next, use openbabel:

babel -ixyz methanol.xyz -ogzmat 

#Put Keywords Here, check Charge and Multiplicity.

 methanol

0  1
C
H  1  r2
H  1  r3  2  a3
H  1  r4  2  a4  3  d4
O  1  r5  2  a5  3  d5
H  5  r6  1  a6  2  d6
Variables:
r2= 1.1117
r3= 1.1117
a3= 109.74
r4= 1.1094
a4= 108.78
d4= 118.90
r5= 1.3984
a5= 110.18
d5= 238.51
r6= 0.9924
a6= 105.98
d6=  60.61

1 molecule converted
18 audit log messages

The format isn't quite right (everything in red needs to go, and the V in blue should be lower case), but we can sort that out:

 babel -ixyz ~/methanol.xyz -ogzmat |sed 's/\=//g;s/V/v/g;s/\://g' |tail -n+6 > methanol.zmat

C
H  1  r2
H  1  r3  2  a3
H  1  r4  2  a4  3  d4
O  1  r5  2  a5  3  d5
H  5  r6  1  a6  2  d6
variables
r2 1.1117
r3 1.1117
a3 109.74
r4 1.1094
a4 108.78
d4 118.90
r5 1.3984
a5 110.18
d5 238.51
r6 0.9924
a6 105.98
d6  60.61

Let's update out nwchem input file with the internal coordinates:

scratch_dir /home/andy/scratch
Title "methanol"

Start  methanol

echo

charge 0

geometry noautoz
zmatrix
C
H  1  r2
H  1  r3  2  a3
H  1  r4  2  a4  3  d4
O  1  r5  2  a5  3  d5
H  5  r6  1  a6  2  d6
variables
r2 1.1117
r3 1.1117
a3 109.74
r4 1.1094
a4 108.78
d4 118.90
r5 1.3984
a5 110.18
d5 238.51
r6 0.9924
a6 105.98
d6  60.61
end
end

ecce_print ecce.out

basis "ao basis" spherical print
  H library "6-31+G*"
  O library "6-31+G*"
  C library "6-31+G*"
END

dft
  mult 1
  direct
  XC b3lyp
  grid fine
  mulliken
end

driver
  default
end

task dft optimize

And run. Done!

511. When nwchem PES scans fail to constrain -- autoz failure

Another update:
My jobs have run long enough now that I can confirm that using your own z matrix still does not constrain the bond lengths i.e. the link at the end of the post is useless.

Update:
I'm not sure using a zmatrix actually solved this -- for each step in the optimization it seems that nwchem attempts to generate a new zmatrix, and probably ignoring my input (and yes, I'm using noautoz). I'll let my calcs run for a little while to see whether the constraints are honoured or not.

But I'm getting really frustrated with nwchem right now, especially since gaussian isn't having any issues running these particular jobs (there are other issues with gaussian, such as the format of the output, etc.)

Original post:
I set up PES scans in nwchem as shown in this post: http://verahill.blogspot.de/2013/08/503-relaxed-pes-scanning-in-nwchem.html

I was noticing that while almost all of my potential energy surface scans were working out just fine, some of them would...not. What would happen is that there would be no error messages, but for some reason the e.g. atom-atom distance that was defined (and defined using constant) would not remain constant during the geometry optimization in each step.

I saw this when looking trying to move an anion (9 atoms) step-wise closer to a large, negatively charged metal oxide ion (25 atoms).

I took a while to chase this down. First I though that well maybe the distances weren't really set as immutable, but were just associated with a certain force constant -- and that the anion-anion repulsion somehow overcame this. That wasn't the case.

Instead it was something that I should've paid attention to: the zmatrix generation.
If you find that for some reason the PES scan is not constrained at all, look for something along the lines of the following in your output:

NWChem Input Module
                                -------------------


                                 molecules_def2_svp
                                 ----------------

 Scaling coordinates for geometry "geometry" by  1.889725989
 (inverse scale =  0.529177249)


          ------
          auto-z
          ------
      warning. autoz generated   10 bonds for atom   24
      warning. autoz generated   10 bonds for atom   25
      warning. autoz generated   10 bonds for atom   26
      warning. autoz generated   10 bonds for atom   27
      warning. autoz generated   10 bonds for atom   28
      warning. autoz generated   10 bonds for atom   29
  autoz: The atoms group into disjoint clusters
 cluster   1:    1    2    3    4
 cluster   2:    5    6    7    8    9   10   11   12   13   14   15
                16   17   18   19   20   21   22   23   24   25   26
                27   28   29
 Connecting clusters   1   2 via atoms    3    7 r = 3.71
  autoz: regenerating connections with new bonds
      warning. autoz generated   10 bonds for atom   24
      warning. autoz generated   10 bonds for atom   25
      warning. autoz generated   10 bonds for atom   26
      warning. autoz generated   10 bonds for atom   27
      warning. autoz generated   10 bonds for atom   28
      warning. autoz generated   10 bonds for atom   29
  autoz: excessive number of variables                  2066                   81

 AUTOZ failed to generate good internal coordinates.
 Cartesian coordinates will be used in optimizations.

If that happens, cartesian coordinates will be used, and your

python
from nwgeom import *
geom = ''' 
    geometry adjust
        zcoord
            bond 1 14 %f bond constant
        end
    end 
'''

won't do anything.

The solution is to provide the coordinates as a zmatrix instead -- and that's the focus of my next post:
http://verahill.blogspot.com.au/2013/09/512-briefly-zmatrices-in-nwchem-methanol.html

Oh, and don't forget to include noautz

510. Kernel 3.11 on Debian

Here's another semi-recycled kernel building post -- the instructions are the same as for (almost) any previous version of the kernel.

There are several ways of building a kernel. The easiest (a purely subjective statement) is to use kernel-package i.e. make-kpkg. However, every now and again I see people writing that it's been deprecated.

I never spent much time with 3.10 as it caused (subjective) slow-downs on my SL410 laptop, and had issues -- at least in the early iterations -- when building the nvidia dkms module. I compiled this on a running 3.9.2-ck kernel.

So here's 3.11 -- note that this kernel too will not build the nvidia 304.88 dkms module, although it seems that 325.15 will build with a little bit of patching.

Either way, start by doing

sudo apt-get install fakeroot build-essential ncurses-dev kernel-package
mkdir ~/tmp
cd ~/tmp
wget https://www.kernel.org/pub/linux/kernel/v3.x/linux-3.11.tar.xz
tar xvf linux-3.11.tar.xz
cd linux-3.11/
cat /boot/config-`uname -r`>.config
make oldconfig


Compile also drivers which will not load (COMPILE_TEST) [N/y/?] (NEW) N
Kernel compression mode
> 1. Gzip (KERNEL_GZIP)
  2. Bzip2 (KERNEL_BZIP2)
  3. LZMA (KERNEL_LZMA)
  4. XZ (KERNEL_XZ)
  5. LZO (KERNEL_LZO)
  6. LZ4 (KERNEL_LZ4) (NEW)
choice[1-6?]: 1
Timer tick handling
  1. Periodic timer ticks (constant rate, no dynticks) (HZ_PERIODIC) (NEW)
> 2. Idle dynticks system (tickless idle) (NO_HZ_IDLE) (NEW)
  3. Full dynticks system (tickless) (NO_HZ_FULL) (NEW)
choice[1-3]: 2
Memory placement aware NUMA scheduler (NUMA_BALANCING) [N/y/?] (NEW) N
Simple CPU accounting cgroup subsystem (CGROUP_CPUACCT) [N/y/?] (NEW)
Group CPU scheduler (CGROUP_SCHED) [N/y/?] (NEW)
Automatic process group scheduling (SCHED_AUTOGROUP) [N/y/?] (NEW)
Choose SLAB allocator
  1. SLAB (SLAB) (NEW)
> 2. SLUB (Unqueued Allocator) (SLUB)
choice[1-2?]: 
SLUB per cpu partial cache (SLUB_CPU_PARTIAL) [Y/n/?] (NEW)
AIX basic partition table support (AIX_PARTITION) [N/y/?] (NEW)
Linux guest support (HYPERVISOR_GUEST) [N/y/?] (NEW) N
Timer frequency
  1. 100 HZ (HZ_100)
  2. 250 HZ (HZ_250) (NEW)
  3. 300 HZ (HZ_300)
> 4. 1000 HZ (HZ_1000)
choice[1-4?]: 
Enable workqueue power-efficient mode by default (WQ_POWER_EFFICIENT_DEFAULT) [N/y/?] (NEW)
Memory Hotplug (ACPI_HOTPLUG_MEMORY) [N/y/?] (NEW)
AMD frequency sensitivity feedback powersave bias (X86_AMD_FREQ_SENSITIVITY) [N/m/?] (NEW) 
Support for PCI Hotplug (HOTPLUG_PCI) [N/y/?] (NEW) 
Kernel support for scripts starting with #! (BINFMT_SCRIPT) [Y/n/m/?] (NEW)
InfiniBand media type support (TIPC_MEDIA_IB) [N/y/?] (NEW)
Network Coding (BATMAN_ADV_NC) [N/y/?] (NEW)
NETLINK: mmaped IO (NETLINK_MMAP) [N/y/?] (NEW)
NETLINK: socket monitoring interface (NETLINK_DIAG) [N/m/y/?] (NEW)
MPLS: GSO support (NET_MPLS_GSO) [N/m/y/?] (NEW)
NCI over SPI protocol support (NFC_NCI_SPI) [N/y/?] (NEW)
NFC hardware simulator driver (NFC_SIM) [N/m/?] (NEW) 
Dummy IRQ handler (DUMMY_IRQ) [N/m/y/?] (NEW) 
Generic on-chip SRAM driver (SRAM) [N/y/?] (NEW)
ME Enabled Intel Chipsets (INTEL_MEI_ME) [N/m/y/?] (NEW) 
FlashPoint support (SCSI_FLASHPOINT) [N/y/?] (NEW)
Renesas R-Car SATA support (SATA_RCAR) [N/m/?] (NEW)
Block device as cache (BCACHE) [N/m/y/?] (NEW)
Switch target support (EXPERIMENTAL) (DM_SWITCH) [N/m/?] (NEW)
Virtual netlink monitoring device (NLMON) [N/m/y/?] (NEW)
ARC devices (NET_VENDOR_ARC) [Y/n/?] (NEW)
Qualcomm Atheros AR816x/AR817x support (ALX) [N/m/y/?] (NEW)
QLOGIC QLCNIC 83XX family SR-IOV Support (QLCNIC_SRIOV) [Y/n/?] (NEW)
Renesas SuperH Ethernet support (SH_ETH) [N/m/y/?] (NEW)
SMSC LAN911x/LAN921x families embedded ethernet support (SMSC911X) [N/m/y/?] (NEW)
Realtek RTL8152 Based USB 2.0 Ethernet Adapters (USB_RTL8152) [N/m/?] (NEW) m
Atheros ath9k rate control (ATH9K_LEGACY_RATE_CONTROL) [N/y/?] (NEW) 
Atheros 802.11ac wireless cards support (ATH10K) [N/m/?] (NEW) m
Atheros ath10k PCI support (ATH10K_PCI) [N/m/?] (NEW) m
Atheros ath10k debugging (ATH10K_DEBUG) [N/y/?] (NEW)
Atheros ath10k debugfs support (ATH10K_DEBUGFS) [N/y/?] (NEW)
Atheros ath10k tracing support (ATH10K_TRACING) [N/y/?] (NEW)
rt2800usb - Include support for rt55xx devices (EXPERIMENTAL) (RT2800USB_RT55XX) [N/y/?] (NEW) y
Realtek rtlwifi family of devices (RTL_CARDS) [M/n/?] (NEW) 
Realtek RTL8188EE Wireless Network Adapter (RTL8188EE) [N/m/?] (NEW) m
CW1200 WLAN support (CW1200) [N/m/?] (NEW)
Cypress TrueTouch Gen4 Touchscreen Driver (TOUCHSCREEN_CYTTSP4_CORE) [N/m/y/?] (NEW)
IMS Passenger Control Unit driver (INPUT_IMS_PCU) [N/m/?] (NEW)
Freescale lpuart serial port support (SERIAL_FSL_LPUART) [N/m/y/?] (NEW)
Analog Devices ADT7310/ADT7320 (SENSORS_ADT7310) [N/m/y/?] (NEW) 
GMT G762 and G763 (SENSORS_G762) [N/m/?] (NEW)
National Semiconductor LM95234 (SENSORS_LM95234) [N/m/?] (NEW)
Nuvoton NCT6775F and compatibles (SENSORS_NCT6775) [N/m/y/?] (NEW)
generic cpu cooling support (CPU_THERMAL) [N/y/?] (NEW)
X86 package temperature thermal driver (X86_PKG_TEMP_THERMAL) [M/n/?] (NEW)
MEN A21 VME CPU Carrier Board Watchdog Timer (MEN_A21_WDT) [N/m/y/?] (NEW) 
ChromeOS Embedded Controller (MFD_CROS_EC) [N/m/y/?] (NEW)
Kontron module PLD device (MFD_KEMPLD) [N/m/y/?] (NEW)
Silicon Laboratories 4761/64/68 AM/FM radio. (MFD_SI476X_CORE) [N/m/?] (NEW)
System Controller Register R/W Based on Regmap (MFD_SYSCON) [N/y/?] (NEW)
TI TPS65912 Power Management chip (MFD_TPS65912) [N/y/?] (NEW)
USBTV007 video capture support (VIDEO_USBTV) [N/m/?] (NEW)
Conexant cx25821 support (VIDEO_CX25821) [N/m/?] (NEW)
Cypress firmware helper routines (CYPRESS_FIRMWARE) [N/m] (NEW)
QXL virtual GPU (DRM_QXL) [N/m/?] (NEW)
Max number of sound cards (SND_MAX_CARDS) [32] (NEW) 
Build Display HD-audio controller/codec power well support for i915 cards (SND_HDA_I915) [N/y/?] (NEW) y
M2Tech hiFace USB-SPDIF driver (SND_USB_HIFACE) [N/m/?] (NEW)
Apple infrared receiver (HID_APPLEIR) [N/m/?] (NEW)
ELO USB 4000/4500 touchscreen (HID_ELO) [N/m/?] (NEW)
Huion tablets (HID_HUION) [N/m/?] (NEW) 
Enable USB persist by default (USB_DEFAULT_PERSIST) [Y/n/?] (NEW) Y
 FUSBH200 HCD support (USB_FUSBH200_HCD) [N/m/?] (NEW)
OHCI support for PCI-bus USB controllers (USB_OHCI_HCD_PCI) [M/n/?] (NEW)
USB-Wishbone adapter interface driver (USB_SERIAL_WISHBONE) [N/m/?] (NEW) 
Infineon Modem Flashloader USB interface driver (USB_SERIAL_FLASHLOADER) [N/m/?] (NEW) 
USB Suunto ANT+ driver (USB_SERIAL_SUUNTO) [N/m/?] (NEW)
USB Physical Layer drivers (USB_PHY) [N/y/?] (NEW) 
Faraday FOTG210 USB Peripheral Controller (USB_FOTG210_UDC) [N/m/?] (NEW) 
PXA 27x (USB_PXA27X) [N/m/?] (NEW)
MARVELL PXA2128 USB 3.0 controller (USB_MV_U3D) [N/m/?] (NEW)
USB functions configurable through configfs (USB_CONFIGFS) [N/m/?] (NEW) 
LED Support for TI LP5562 LED driver chip (LEDS_LP5562) [N/m/?] (NEW)
LED Camera Flash/Torch Trigger (LEDS_TRIGGER_CAMERA) [N/m/y/?] (NEW)
QIB DCA support (INFINIBAND_QIB_DCA) [Y/n/?] (NEW) 
Mellanox Connect-IB HCA support (MLX5_INFINIBAND) [N/m/?] (NEW) 
iSCSI Extentions for RDMA (iSER) target support (INFINIBAND_ISERT) [N/m/?] (NEW)
Set system time from RTC on startup and resume (RTC_HCTOSYS) [Y/n/?] (NEW) 
Set the RTC time based on NTP synchronization (RTC_SYSTOHC) [Y/n/?] (NEW) 
 RTC used to set the system time (RTC_HCTOSYS_DEVICE) [rtc0] (NEW) 
NXP PCF2127 (RTC_DRV_PCF2127) [N/m/?] (NEW)
Synopsys DesignWare AHB DMA support (DW_DMAC_CORE) [N/m/y] (NEW)
Synopsys DesignWare AHB DMA PCI driver (DW_DMAC_PCI) [N/m/y/?] (NEW)
WIS GO7007 MPEG encoder support (VIDEO_GO7007) [N/m/?] (NEW)
DesignWare USB2 DRD Core Support (USB_DWC2) [N/m/?] (NEW)
Mediatek Bluetooth support (USB_BTMTK) [N/m/?] (NEW) 
Intel Rapid Start Technology Driver (INTEL_RST) [N/m/y/?] (NEW) 
 Intel Smart Connect disabling driver (INTEL_SMARTCONNECT) [N/m/y/?] (NEW)
 pvpanic device support (PVPANIC) [N/m/y/?] (NEW) 
Reset Controller Support (RESET_CONTROLLER) [N/y/?] (NEW) 
FMC support (FMC) [N/m/y/?] (NEW)
XFS Verbose Warnings (XFS_WARN) [N/y/?] (NEW)
Btrfs will run sanity tests upon loading (BTRFS_FS_RUN_SANITY_TESTS) [N/y/?] (NEW)
Btrfs debugging support (BTRFS_DEBUG) [N/y/?] (NEW) 
EFI Variable filesystem (EFIVAR_FS) [N/m/y/?] (NEW) 
 NFS client support for NFSv4.2 (NFS_V4_2) [N/y/?] (NEW)
Provide Security Label support for NFSv4 server (NFSD_V4_SECURITY_LABEL) [N/y/?] (NEW)
9P Security Labels (9P_FS_SECURITY) [N/y/?] (NEW) 
Wait/wound mutex debugging: Slowpath testing (DEBUG_WW_MUTEX_SLOWPATH) [N/y/?] (NEW)
torture tests for RCU (RCU_TORTURE_TEST) [N/m/y/?] (NEW)
Ring buffer startup self test (RING_BUFFER_STARTUP_TEST) [N/y/?] (NEW)
Test functions located in the string_helpers module at runtime (TEST_STRING_HELPERS) [N/m/y] (NEW)
Debug alternatives (X86_DEBUG_STATIC_CPU_HAS) [N/y/?] (NEW)
CMAC support (CRYPTO_CMAC) [M/y/?] (NEW) 
SHA256 digest algorithm (SSSE3/AVX/AVX2) (CRYPTO_SHA256_SSSE3) [N/m/y/?] (NEW)
SHA512 digest algorithm (SSSE3/AVX/AVX2) (CRYPTO_SHA512_SSSE3) [N/m/y/?] (NEW)
Camellia cipher algorithm (x86_64/AES-NI/AVX2) (CRYPTO_CAMELLIA_AESNI_AVX2_X86_64) [N/m/y/?] (NEW)
Serpent cipher algorithm (x86_64/AVX2) (CRYPTO_SERPENT_AVX2_X86_64) [N/m/y/?] (NEW)
LZ4 compression algorithm (CRYPTO_LZ4) [N/m/y/?] (NEW) 
LZ4HC compression algorithm (CRYPTO_LZ4HC) [N/m/y/?] (NEW)
KVM legacy PCI device assignment support (KVM_DEVICE_ASSIGNMENT) [Y/n/?] (NEW) 
VHOST_SCSI TCM fabric driver (VHOST_SCSI) [N/m/?] (NEW) 


make menuconfig

You can now enable any additional modules by navigating the menu structure. Note that most likely you don't have to enable anything in this step, but it can come in handy if there's a major transition (e.g. the way multimedia was handled changed between kernel 3.5 and 3.6) or if you want to enable a previously disabled option.

If you only want to compile modules that are currently in use (not a good idea if you want to use the same kernel on a range of computers, or have USB devices that aren't currently plugged in) you can do that by using make localmodconfig instead of make oldconfig. I wouldn't recommend it -- in most cases it won't make a faster kernel, and space and memory tends not to be much of in the way of issues these days.

Below, change -j2 to -jX, where X is the number of cores in your CPU (not cores+1 or anything funny like that. See other posts on this blog for compilation performance tests)

make-kpkg clean
time fakeroot make-kpkg -j2 --initrd kernel_image kernel_headers
sudo dpkg -i ../linux-image-3.11.0_3.11.0-10.00.Custom_amd64.deb ../linux-headers-3.11.0_3.11.0-10.00.Custom_amd64.deb

And that's it.

509. Very briefly: Send remote commands via the dropbox folder

This is probably fairly obvious to most people.

I've got a reverse ssh tunnel set up so that I can access my work computer from home. However, for the past few days I've had the connection get stuffed up on a regular basis (it doesn't get dropped, but the connection gets refused), and it frustrates me a little bit.

While a proper ssh connection is unbeatable, I would at least be able to copy files back and forth via dropbox if I only had a way of sending commands to my work computer.

And an obvious way of doing that would be to use a cronjob and a tiny bit of bash scripting. So here we go:
While we don't have to (we could just have an empty script file instead) I like the idea of testing for the presence of a specific file in the Dropbox folder, and if it exists, execute it.

Let's call the file that tests for it runremote.sh, and put it in our home folder (~/). I personally suspect that making sure that execution output and error messages get properly logged is a good thing if you're going to fly blind like this, hence the 1> and 2>

runremote.sh

if [ -e ~/Dropbox/runme.sh ]; then
    sh ~/Dropbox/runme.sh 1>> ~/Dropbox/runme.log 2>> ~/Dropbox/runme.error &
fi

Then when you want something executed, put a file called runme.sh in ~/Dropbox:

pwd
echo 'Is it working?'
cp ~/testfile.text ~/Dropbox
date

Note that any command in runme.sh is going to be run in the ~/ folder -- not in ~/Dropbox.

And set the runremote.sh file to be executed e.g. every five minutes through cron:

crontab -e

*/5 * * * * sh ~/runremote.sh

Again, you don't need to have it test for the presence of a file, but I just instinctively like the idea.

Anyway, any command you put in ~/Dropbox/runme.sh should be executed and logged within five minutes from being synced.

You CAN use sudo (echo mypassword| sudo -S ls /root )as well by providing your password in the script file, but this is obviously not terribly safe.

508. Very, very briefly: why apt-get purge exists (anecdote)

Sometimes my brain and I fall out, and we refuse to communicate. Yesterday was a day like that.

Having put my TV computer back together again (see post), I realised that I hadn't installed conky or guake on it. So I did. Having edited my default conky.conf (a bit like the one in this post) to change eth0 to eth1, I was wondering why my wireless card wasn't showing up in the ifconfig -a output. lspci showed that I had a broadcom chip, so I installed broadcom-sta-dkms. Nothing much happened, so I rebooted. And...no network card whatsoever recognised (including the ethernet one).

Face palm. I don't have a wireless pci/pci-e card on that computer. I should know -- I had just rebuilt it...

Oh well. apt-get autoremove broadcom-sta-dkms. Nothing. Reboot. Still no eth card -- also, the eth card was obviously the broadcom one (broadcom 4401). But why wasn't it working? I had uninstalled the broadcom package after all.

A bit of googling suggested loading b44 so modprobe b44 it was. Still nothing. Then a thought hit me -- ls /etc/modprobe.d/*blacklist*

And yes -- the broadcom package had installed a broadcom-wl-blacklist.conf file in /etc/modprobe.d -- and b44 was blacklisted. At this point I could either edited the file and comment out b44, delete the file entirely or purge the package and everything would be back to normal. So I did.

So remember: just because you uninstall a package (apt-get autoremove) doesn't mean that you've restore your system to the state it was before you installed the package. To at least have a chance of restoring it you need to purge your package so that settings files are removed too (apt-get purge).

507. (re)assembling a Dell C521 -- pictures

I finally got around to getting a new (well, refurbished) power supply for my old Dell C521. While the PSU worked fine, on trying to shut down the computer before it booted linux I managed to get the power button stuck. As it turns out, it's not that easy to access things on the front panel on this particular model...

So I took everything apart. EVERYTHING. Except for removing the front panel, ironically. If this post has a lesson for anyone else it's that you shouldn't be afraid to attack (gently) your desktop with a screwdriver. Computers ain't magic.

Anyway, here are some pictures from me reassembling it. For most people it will probably be more useful to look at these pictures in reverse.

Empty case. The arrow indicates where the screws for the panel go. I forget to attach them the first time I reassembled the case...

The plastic bits go to the lights on the front panel

The arrow indicates the front button thingy which I had to attack (gently) with a screw driver to get the power button to release.

The power button. Not sure why I took a picture of it, but there you go.

Front panel screws.

The (very dusty) fan swings out in the direction shown by the arrow. Because of this, you must remove the mobo before removing the fan, which you need to remove in order to remove the front panel. Not a very good design.

The mobo gets stuck in the fabric, so you might need to poke it with a narrow plastic pen or something like that in order to get it to release when removing it. Lift up this side and then slide out the mobo when dismantling.

Another odd design feature -- the heatsink must (more or less) be removed in order to reach the front panel. Also, the CPU is in dire need of a clean and some thermal paste.

The heatsink is now attached -- the plastic bit funnels the airflow from the fan at the front across the heat sink.

The PSU is now in place, but not connected

The P1 main power connector

ATX12V (labelled p2 here) in place

The HDD in place

The CD/DVD ROM in place

PCI and PCI-E in place (and nvidia graphics card and a DTV 1000S tv card)

And it's back in one piece.

506. Extracting optimized structures from a potential energy scan in nwchem

Another update:
It now dumps the energies in a file, energies.dat, as well.

Update:
some programmes, like ecce, are more picky about the xyz format than others (e.g. jmol, vmd). I've updated the code to output xyz files that ecce too can read.

Original post:
When you use scan_input() in nwchem to do a PES scan (see e.g. here: http://verahill.blogspot.com.au/2013/08/503-relaxed-pes-scanning-in-nwchem.html) you get the energies and the gradients for the optimized structures returned as the results. However, for a casual user the atomic actual coordinates is more informative.

Here's a very simple parser written in python (2.7) which extracts the optimized structures from the output file:

#!/usr/bin/python
import sys

def getrawdata(infile):
        f=open(infile,'r')
        opt=0
        geo=0
        energy=[]
        energies=[]
        struct=[]
        structure=[]
        for line in f:
                if "Total DFT" in line:
                        line=filter(None,line.rstrip('\n').split(' '))
                        energy=float(line[4])
                if 'Optimization converged' in line:
                        opt=1
                if opt==1 and 'Geometry' in line:
                        geo=1
                if      'Atomic Mass' in line and (opt==1 and geo==1):
                        opt=0
                        geo=0
                        struct+=[structure]
                        energies+=[energy]
                        structure=[]
                if opt==1 and geo==1:
                        structure+=[line.rstrip()]
        return struct,energies

def genxyzstring(coords,element):
        x_str='%10.5f'% coords[0]
        y_str='%10.5f'% coords[1]
        z_str='%10.5f'% coords[2]
 
        xyz_string=element+(3-len(element))*' '+10*' '+\
        (8-len(x_str))*' '+x_str+10*' '+(8-len(y_str))*' '+y_str+10*' '+(8-len(z_str))*' '+z_str+'\n'
 
        return xyz_string

def getstructures(rawdata):
        
        n=0
        for structure in rawdata:
                
                n=n+1
                num="%03d" % (n,)
                g=open('structure_'+num+'.xyz','w')
                itson=False
                cartesian=[]
                        
                for item in structure:
                        
                        if itson and not(item==""):
                                coords=filter(None,item.split(' '))
                                coordinates=[float(coords[3]),float(coords[4]),float(coords[5])]
                                element=coords[1]
                                cartesian+=[genxyzstring(coordinates,element)]
                                #cartesian+=[coords[1]+'\t'+coords[3]+'\t'+coords[4]+'\t'+coords[5]+'\n']
                
                        if "---" in item:
                                itson=True
                        if item=="" and itson==True:
                                itson=False
                                if not(len(cartesian)==0):
                                        g.write(str(len(cartesian))+'\n')
                                        g.write('Structure '+str(n)+'\n')
                                        for line in cartesian:
                                                g.write(line)
                                        g.close()
                                cartesian=[]
        return 0
        
if __name__ == "__main__":
        infile=sys.argv[1]
        rawdata,energies=getrawdata(infile)
        structures=getstructures(rawdata)

        g=open('energies.dat','w')
        for n in range(0,len(energies)):
                g.write(str(n)+'\t'+str(energies[n])+'\n')
        g.close()

Presuming that you've saved it as pes_parse.py you can then generate a series of xyz files with the structures, catenate them into a trajectory file, and open it in e.g. jmol. I'm using the output from example 1 in http://verahill.blogspot.com.au/2013/08/503-relaxed-pes-scanning-in-nwchem.html as the example:

chmod +x pes_parse.py
./pes_parse.py nwch.nwout
ls

nwch.nwout    structure_001.xyz  structure_003.xyz  structure_005.xyz  structure_007.xyz  structure_009.xyz  structure_011.xyz  structure_013.xyz  structure_015.xyz  structure_017.xyz  structure_019.xyz
pes_parse.py  structure_002.xyz  structure_004.xyz  structure_006.xyz  structure_008.xyz  structure_010.xyz  structure_012.xyz  structure_014.xyz  structure_016.xyz  structure_018.xyz

cat structure_*.xyz >> trajectory.xyz 
jmol trajectory.xyz

You can go through the structures by clicking on the arrows indicated by the white arrow:

Finally, using VMD it's easy to make videos -- note that they for some reason look awful here (seems like a lot of frames are removed, in particular from the beginning of the run):

And here's the SN2 reaction from post 503:

505. Testing KDE on Debian (if you're a gnome user) -- getting set up

Every now and again I get ticked off with something to do with planned 'features' in gnome 3. I'm still not happy about gnome-screenshot, but I patched and compiled my own version, so I'm doing alright. I don't really like the transition of menus from the, well, menu bar of applications to the top bar of the gnome desktop, but it's not a great issue.

Oh, and I don't really use vanilla gnome 3 anyway -- I've got a desktop set up with a top panel, and bottom panel, static desktops etc. The latest 'feature' which is making me want to move to a different desktop is the planned change in the use of the middle-button so that it no longer can be used to paste text.

(of course there's a bit of inconsistency here: KDE turns out to have a remapped middle-click, the screenshot application needs a few more clicks than the old gnome-screenshot etc.)

I've toyed with the idea of changing desktops in the past, and for a while was running nothing but KDE on my laptop (I'd rather switch on my 'casual' computer first, before making the transition on my work computer), but for some reason I switched back. I actually can't remember why, but something really annoyed me about KDE.

Before I forget: a good reason to try another desktop is simply to reassure yourself that you do have alternatives if your chose DE suddenly turns into something you can't work with.

Anyway, time to give it another go. But what desktop? I like the idea of going for KDE since it's 'the other desktop' -- it's been around longer than gnome, whereas lxde and xcfe are mere babies in comparison. Alternatively, one could always piece together a custom desktop using e.g. fluxbox. Or switch to something completely different, like xmonad. Or something a little bit different, like Enlightenment. I think I should probably be going for LXDE (I instinctively don't like XFCE), but let's go for KDE first.

Finally, the main danger is switching desktops and expecting the new one to act exactly like the old one (minus the bad parts, obviously). Which is pretty much how windows users (myself included once upon a time) behave when trying out linux for the first time.

Anyway, time for a week in KDE (4.10.5).

Here's my gnome desktop at the time of switching:

In addition, I'm using Guake, and my network manager is wicd.

So let's see what KDE will look like at the end -- whether I'll be able to use KDE as KDE, instead of as an ersatz gnome.

Step 1. Install KDE

sudo apt-get install kde-full wicd-kde kdesudo yakuake

I selected kdm.

Step 2. First boot
Reboot and make sure to click the little sun icon to select the desktop you want to use, in my case KDE/plasma.

2.1 Switching workspaces
I next set up proper workspace switching so that it behaves like in...gnome (falling into the old trap of making the new desktop just like the old one)...i.e. ctrl+alt right/left arrow.

Go to System Settings/WorksSpace behaviour, Virtual Desktops -- I incrased the number of desktops to 6 (1 row), then clicked on the Switching tab -- I set shortcuts for Switch One Desktop to the Left and Switch One Desktop to the Right to ctrl+alt+left and ctrl+at+right.

2.2 Disable nepomuk
At this point I noticed that I was using 100% of the CPU. virtuoso-t was chewing up a lot, as was nepomukservices. I had a look at this post: https://forum.kde.org/viewtopic.php?f=66&t=98892

I first went to Search and Indexing (in the menu), unchecked Monitor file and directory changes, Enable when running on battery, Only when computer is not being used.  You can also go to System Settings/Desktop Search and uncheck Enable Nepomuk Semantic Desktop.

2.3 konqueror crashing
I also had repeated issues with konqueror crashing when clicking on links, but it seems to have resolved itself after disabling nepomuk.

2.4 guake/yakuake
Guake worked from the beginning -- I didn't have to set it up at all. However, I still ended up changing to yakuake since I figure it's a bit more KDE like.To get yakuake to remember changed settings (e.g. height, width) you'll need to close it manually. Restarting it will use the new settings. Not sure how I got yakuake to autostart -- I think it just happened.

2.5 The bottom panel
I had a difficult time figuring out how to add shortcuts to the panel, and once I had accidentally figured it out, I promptly screwed up the entire bottom panel, then managed to fix it again.

2.6 Keyboard shortcuts
I then had to figure out how to bind shortcuts to opening applications -- right-click on the menu ('kicker'), click Edit Applications, select your application, advanced, Advanced, Current Shortcut Key.

A more general approach to creating shortcuts was shown to me by an anonymous commentator (see below this post) -- I've posted the screenshot provided to me below:

shortcuts, gestures and whatnots

2.7 Update notification
I promptly removed the Update Notification widget -- it was driving me crazy and reminds me too much of windows. I first tried simply disabling the update notifier:

I then did apt-get autoremove update-notifier-kde. The problem -- even after a reboot I still get notifications about updates! I then did System Settings/Software Management, and click on the tiny, tiny button on the far right (level with the search field), and then click on settings and set Check for updates to never.  This finally seems to have done it.

2.8 Theme
I also changed from the default 'air' theme to 'oxygen'.

2.9 Gnome/GTK programs in KDE
To make gnome (well, gtk) programs look acceptable under KDE, install gtk2-engines-oxygen and gtk3-engines-oxygen and reboot.

2.10 Tap-to-click with synaptics
I can't get synaptics tap to click to work. Based on the odd post I tried gpointing-device-settings, but while there's a checkbox to disable tap to click, it won't enable it. I then installed kde-config-touchpad and was happy again. Funnily enough this made me discover that middle-click is mapped in KDE -- the one reason I wanted to explore something other than gnome...oh well.

2.11 Conky
To get conky to work properly I only had to make some minimal changes. /etc/conky/conky.conf:

background no
own_window yes
own_window_type normal
own_window_argb_visual true
own_window_transparent yes
own_window_hints undecorated,below,sticky,skip_taskbar,skip_pager
double_buffer yes


alignment top_right
background yes
border_width 1
cpu_avg_samples 2
default_color white
default_outline_color white
default_shade_color white
default_graph_size 20 200
draw_borders no
draw_graph_borders yes
draw_outline no
draw_shades no
use_xft yes
xftfont DejaVu Sans Mono:size=12
gap_x 20
gap_y 60
minimum_size 5 5
net_avg_samples 2
no_buffers yes
out_to_console no
out_to_stderr no                                                       
extra_newline no                                                       
own_window_class Conky                                                 
own_window_transparent yes                                             
stippled_borders 0                                                     
update_interval 1.0                                                    
uppercase no                                                           
use_spacer none                                                        
show_graph_scale no                                                    
show_graph_range no                                                    

TEXT
${exec lsb_release -ds}
${execi 5 acpi|gawk '{print $3,$4}'}
${addr eth0}/${addr wlan0}
$hr
${color grey}Uptime:$color $uptime
CPU:$alignc $cpu%
$alignc$color ${cpugraph 10,200 ffff00 ff0000 -t}  
RAM:$alignc $mem/$memmax 
$alignc$color ${memgraph 10,200 ffff00 ff0000 -t}
I/O:$alignc $diskio_read/$diskio_write
$alignc$color ${diskiograph 10,200 ffff00 ff0000 -t}
${color grey}Frequency (in GHz):$color
        ${freq_g 1}, ${freq_g freq_g2}
${color grey}Temperature:
$color CPU: ${acpitemp}°C
$hr
${color grey}File systems:$alignr $color/ ${fs_used /}/${fs_size /}
$alignr /home $color${fs_used /home}/${fs_size /home}
${color grey}Networking:
${color grey}eth0 ${color red} ${upspeed eth0}/${color green} ${downspeed eth0} ${color grey}
${color grey} ${upspeedgraph eth0 10,100 ffff00 ff0000 -t} ${color grey} ${downspeedgraph eth0 10,100 0000ff 00ff00 -t} ${color grey}
${color grey}wlan0 ${color red} ${upspeed wlan0}/${color green} ${downspeed wlan0} ${color grey}
${color grey} ${upspeedgraph wlan0 10,100 ffff00 ff0000 } ${color grey} ${downspeedgraph wlan0 10,100 0000ff 00ff00 } ${color grey}
$hr
${color grey}Name                       CPU%   MEM%
${color} ${top name 1} ${top cpu 1} ${top mem 1}
${color} ${top name 2} ${top cpu 2} ${top mem 2}
${color} ${top name 3} ${top cpu 3} ${top mem 3}
$hr
${color}City            Ping $alignr Time  
$font${color}San Francisco ${execi 60 ping -c 1 sfgate.com -n|grep icmp_seq|sed 's/=/\t/g'|gawk '{print $10,$11}'}       $alignr${tztime America/Los_Angeles %a %H:%M}
$font${color}Copenhagen          $alignr${tztime Europe/Copenhagen %a %H:%M}
$font${color}Shanghai    $alignr${tztime Asia/Shanghai %a %H:%M}
$font${color}Melbourne   $alignr${tztime Australia/Melbourne %a %H:%M

And I've discovered that I really like the hot corner (top left) in gnome. There is a hot corner in kde too, but it does slightly different things.

KDE is pretty but it still gives me the occasional feeling that I'm using windows due to the bottom panel, the menu on the left and the task tray on the right, but these are things I'll hopefully grow out of. But yes, KDE is pretty -- very pretty. And it's pretty fast too.

My old gnome desktop (same as at the beginning of this post)

and my new KDE desktop

And one just to show how versatile KDE is -- you can even make it look (almost) like gnome 3

So...now it just remains to see how I feel after a week or so. The baby duck syndrome is pretty powerful...and I'm already well on my way to turning KDE into GNOME. But then, isn't the versatility of KDE the real selling point?

Postscript:
I ended up re-enabling nepomuk and spent some time struggling with akonadi to get kmail working -- the key to getting akonadi to work was to install the same server as the backend. In my case I did:

sudo apt-get install akonadi-backend-mysql mysql-server

504. Swap file and hibernation on debian

I've got an SSD and 4 gb ram on my laptop, and have no desire to use to actual swap during normal use. However, I'd like to be able to let my laptop go into hibernation -- even suspend seems to be draining my battery pretty fast (at maybe a quarter of the rate of keeping the laptop on).

Does it make sense hibernating a laptop with SSD i.e. one which boots in ten seconds flat? Probably not. But we humans are greedy by nature.

So I need to set up a swap file, disable swapping, and see if I can use it for hibernation.

As usual, the best source of information is the archlinux wiki.
https://wiki.archlinux.org/index.php/Swap#Swap_file
https://wiki.archlinux.org/index.php/Pm-utils#Using_Swap_file_instead_of_regular_swap_partition

Anyway, having made it work and tried it out I would say this is really not worth the hassle IF you have an SSD. 

---

1. The swap file
This is my file system layout:

Filesystem      Size  Used Avail Use% Mounted on
/dev/sda1        39G   20G   18G  53% /
udev             10M     0   10M   0% /dev
tmpfs           380M  816K  380M   1% /run
tmpfs           5.0M     0  5.0M   0% /run/lock
tmpfs           760M  1.1M  759M   1% /run/shm
/dev/sda2       109G   71G   33G  69% /home

In my case I'd say that / has enough space to handle a swap file. My RAM is '4 gb' (really 3.71 Gb) -- according to this and this my swap should equal my RAM, which makes intuitive sense.

Since 4,000,000,000 byte is 3.76 gb, and 4000*1024*1024 (i.e. 4000M) is 3.91 Gb, I think 4000M should be ok:

su -
fallocate -l 4000M /swapfile
chmod 600 /swapfile
mkswap /swapfile

Setting up swapspace version 1, size = 4095996 KiB
no label, UUID=2a8de3d1-14f6-473f-b40f-31618fd81169

echo 'vm.swappiness=1' >> /etc/sysctl.d/50-local.conf
echo '/swapfile none swap defaults 0' >> /etc/fstab

---

2. "PM: Swap header not found"

To try it out without rebooting:

sudo sysctl -w vm.swappiness=1
sudo swapon /swapfile
pm-is-supported --hibernate 
echo $?

0

If you got 0, then you're good to go.
sudo pm-hibernate

Trying it out the first time I got "PM: swap header not found" and some weird behaviour. This has been mentioned e.g. here. A step-by-step guide is here: https://ubuntuforums.org/showthread.php?t=1042946

Get the UUID of the partition on which the swapfile is located:

mount | grep " / "

/dev/disk/by-uuid/8adf424c-c375-4035-8d5d-181489b4461b on / type ext4 (rw,noatime,nodiratime,discard,errors=remount-ro,data=ordered)

sudo filefrag -v /swapfile | grep "First block:"

The latter command gave nothing, so I then did:

sudo filefrag -v /swapfile|less

Filesystem type is: ef53
File size of /swapfile is 4194304000 (1024000 blocks of 4096 bytes)
 ext:     logical_offset:        physical_offset: length:   expected: flags:
   0:        0..       0:    7182336..   7182336:      1:            
   1:        1..    6143:    7182337..   7188479:   6143:             unwritten
   2:     6144..    8191:    7190528..   7192575:   2048:    7188480: unwritten

So now we have the UUID (8adf424c-c375-4035-8d5d-181489b4461b) and the offset (7182336).

su -
echo "resume=UUID=8adf424c-c375-4035-8d5d-181489b4461b resume_offset=7182336" | sudo tee /etc/initramfs-tools/conf.d/resume
exit

Edit /etc/default/grub and add the same line to GRUB_CMDLINE_LINUX_DEFAULT:

GRUB_CMDLINE_LINUX_DEFAULT="quiet resume=UUID=8adf424c-c375-4035-8d5d-181489b4461b resume_offset=7182336"

Run

sudo update-grub
sudo update-initramfs -u

Reboot.

---

3. Hibernating.

sudo pm-hibernate

On my lenovo sl410 what I see is the screen go blank save for a blinking "-", and the little crescent light at the front of the laptop starting to blink (next to the battery light).

After 5-10 seconds the laptop turns off.

Hitting the power button starts up the laptop -- you get the bios screen, then the grub menu, and at this point you're thinking that the whole exercise has failed -- but it ends up booting into the same situation as you had when you did pm-hibernate i.e. it worked. I mean, this is how hibernation works -- but I had honestly never used anything other than suspend before, so...well..there's a first time for everything.

Overall, on a laptop with an SSD, resuming from hibernation is about as slow as a cold start -- with entering hibernation taking longer than a shutdown. On a laptop with a spinning disk this could presumably be worth it.