Avogadro for NWchem periodic calculations

the killer feature will be allowing user to get rid of tedious coordinate operation manually. This is really important. xyz for simple calcs is not as important, as manipulating crystal structure. if user can import cif, cut it to the right plane, make it 2-D slab, add some adsorbed molecules and export it to output that would be best. basis set etc is almost extraneous to the amount of work needed to setup input in fractional coordinates. that’s why I am advocating it

I am using both: plane wave and “regular” DFD calcs so I know how painful is setup former when compared to latter. probably good example of good solid state data manipulation and input prep package would be Accelryss Materials Studio.

Keep in mind, that packages, such as Quantum-Espresso, CPMD and NWCHem don;t have hassle free input manipulation and visualization program. They can be linked to some other visualization packages, but definitely not simple and efficient manner

— On Thu, 4/2/09, Marcus D. Hanwell marcus@cryos.org wrote:

From: Marcus D. Hanwell marcus@cryos.org
Subject: Re: [Avogadro-devel] NwChem input
To: “Geoffrey Hutchison” geoff.hutchison@gmail.com
Cc: jasius_1@yahoo.com, avogadro-devel@lists.sourceforge.net
Date: Thursday, April 2, 2009, 8:16 AM
Geoffrey Hutchison wrote:

Also, is there a way to setup multijob inputs
(Gaussian has those,
for example, as well as NWChem)

You can do this manually, but no, we don’t have an
automatic way to do
this.

This becomes a question of how far the input generation
should go too.
There comes a time when people are generating very complex
input files -
do people just tend to enter these manually? I know I do,
but find it
convenient to have a starting point in Avogadro.

One killer feature I would like to put in at some point for
advanced
users is syntax highlighting of input files, and this is
not very hard
to do. I am just not sure how I feel about trying to make
an input deck
generator do everything…

Another good one for advanced users is input file templates
with
replacement of keywords. That would allow people to
generate templates
for complex jobs and have keywords replaced with the atom
coordinates,
unit cell parameters etc. Would this be a better direction
than trying
to create a dialog to expose every possible function in a
quantum code?
It would allow a simple dialog that is personalized to each
user,
classes could distribute input files etc.

These are all just ideas. I think I would prefer to add job
templates
and syntax highlighting to provide a natural progression as
people start
to use more advanced features.

Thanks,

Marcus

Jonas Baltrusaitis wrote:

the killer feature will be allowing user to get rid of tedious coordinate operation manually. This is really important. xyz for simple calcs is not as important, as manipulating crystal structure. if user can import cif, cut it to the right plane, make it 2-D slab, add some adsorbed molecules and export it to output that would be best. basis set etc is almost extraneous to the amount of work needed to setup input in fractional coordinates. that’s why I am advocating it

That would be the killer feature for you and the work you do. Quite a
few components are already there, but I am unlikely to use those
features in my work. That is why we went to the extra effort to create a
framework and a clear API - every researcher has different needs. I have
never touched a plane wave calculation but can appreciate where you are
coming from. Many other people feel that a z matrix editor is the big
missing feature…

I am using both: plane wave and “regular” DFD calcs so I know how painful is setup former when compared to latter. probably good example of good solid state data manipulation and input prep package would be Accelryss Materials Studio.

Keep in mind, that packages, such as Quantum-Espresso, CPMD and NWCHem don;t have hassle free input manipulation and visualization program. They can be linked to some other visualization packages, but definitely not simple and efficient manner

This is exactly why we have gone to a lot of trouble to create a library
and I have been working hard on making the API clear and intuitive as
well as the application. I have not used any of those packages but would
love to see support added. I am not likely to be the one doing it though
as I don’t have access to the packages and there are several other
features we need more for the work we are currently engaged in.

Many of the building blocks are already in Avogadro. I would certainly
help as much as I could, but do not have the time to learn about plane
wave calculations, get software set up and implement support. I have
another group of users who want to see far better support for large
systems, MD and trajectories for example.

On Thu, Apr 2, 2009 at 8:02 PM, Jonas Baltrusaitis jasius_1@yahoo.com wrote:

David,

a vague question: which, if any, software from Microsoft DreamSpark package can be used for C++ coding? I have access to it via university

I have no clue – this is my first attempt at getting anything to
compile in windows, so I’m of no help here. Myself (as well as most of
the devs from what I gather) use the free MSVC express that can be
found here:

Visual Studio: IDE and Code Editor for Software Developers and Teams

HTH,

Dave

How do I load source code into the Visual? In particular, how do I load vibration visualization part of code into the Visual? I would like to work on the last bug you recommended but don;t know how to load the code

JOnas

Gents,

do take a look at

http://www-organik.chemie.uni-wuerzburg.de/lehrstuehlearbeitskreise/bringmann/specdis/

This would partially answer some of the problems David has with my constant nagging to implement CD and UV spectra. If something similar can be done with Avogadro that would be great. I can be a liaison and answer all the technical questions in visualizing excited state spectra

Jonas

Hi,

I am trying to calculate electrostatic potential of the molecule. Actually, I did that with Gaussian to a cube file, now I want to display it. I load cube file, go to Create surfaces, surface type>Electrostatic Potential (or cubefile.cub Electrostatic potential), color by >Electrostatic Potential (or cubefile.cub Electrostatic potential), click Calculate and… nothing happens in either case.

I check Task Manager, avogadro.exe sits there with no CPU usage, only some memory, I don;t think it’s calculating anything.

Is there a way to display or calculate Gaussian electrostatic potential?

Jonas

PS. it opens the structure from the .cub file so this is encouraging

Hi,

I have a specific interest in CRYSTAL and TURBOMOLE visualization. While CRYSTAL’06 I uploaded input and output files in OB test repository, URBOMOLE is trickier. It produces separate file for pretty much anything (coords, spectra, MOs) and OB can parce only coord file (which it doesn’t do the good job since TURBOMOLE coord can;t be open in Avogadro, even though OB says it supports that format. I guess I added that as a bug somewhere)

But I digress. If I wanted get reading TURBOMOLE

coords
IR/Raman spectra
CD and UV spectra
constructing MOs from the MO coefficent files

implemented in Avogadro, should I request that and upload corresponding separate files in OB repository or Avogadro? I’ve been interacting with spectra guru David himself but this is much broader scope so I don;t know wich is best

Jonas

Hi Jonas,

On Sun, Nov 15, 2009 at 5:16 PM, Jonas Baltrusaitis jasius_1@yahoo.com wrote:

If I wanted get reading TURBOMOLE

coords
IR/Raman spectra
CD and UV spectra
constructing MOs from the MO coefficent files

implemented in Avogadro, should I request that and upload corresponding separate files in OB repository or Avogadro?

I ran into this problem with VASP, too – OB wasn’t designed to read
data from codes that spread their output into several files. It’s
possible, but it limits the functionality (e.g. VASP files can only be
loaded through specific OB calls, and even then only when the absolute
file path is passed). Uploading all the files to OB would be a start.
Since all of Avogadro’s file reading occurs through OB, uploading them
to Avogadro’s tracker wouldn’t neccessarily be useful.

Just my two cents – Others may have suggestions for ways to “fix” OB
to read in multiple files? (nudge, nudge :wink: )

Dave

Similar problem exists with Molpro - it writes details of geometry optimization into separate file, so OB can’t read even coordinates of last point, because no XYZ exist in output (it’s possible only if frequencies were calculated)

My suggestion: add new separate file format for every “piece of output”, and read corresponding information in separate runs of OB.
When user opens main output in Avogadro, other required files will be searched in the same directory and opened with OB. If something wasn’t found, let user to open it manually or skip it.


Regards,
Konstantin

15.11.09, 20:02, “David Lonie” loniedavid@gmail.com:

Hi Jonas,
On Sun, Nov 15, 2009 at 5:16 PM, Jonas Baltrusaitis wrote:

If I wanted get reading TURBOMOLE

coords
IR/Raman spectra
CD and UV spectra
constructing MOs from the MO coefficent files

implemented in Avogadro, should I request that and upload corresponding separate files in OB repository or Avogadro?
I ran into this problem with VASP, too – OB wasn’t designed to read
data from codes that spread their output into several files. It’s
possible, but it limits the functionality (e.g. VASP files can only be
loaded through specific OB calls, and even then only when the absolute
file path is passed). Uploading all the files to OB would be a start.
Since all of Avogadro’s file reading occurs through OB, uploading them
to Avogadro’s tracker wouldn’t neccessarily be useful.
Just my two cents – Others may have suggestions for ways to “fix” OB
to read in multiple files? (nudge, nudge :wink: )
Dave


Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what’s new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july


Avogadro-devel mailing list
Avogadro-devel@lists.sourceforge.net
avogadro-devel List Signup and Options

On Mon, Nov 16, 2009 at 11:25:48AM +0300, Konstantin Tokarev wrote:

Similar problem exists with Molpro - it writes details of geometry
optimization into separate file, so OB can’t read even coordinates of
last point, because no XYZ exist in output (it’s possible only if
frequencies were calculated)

That’s pretty annoying yes. However, we could probably regenerate the
xyz coordinates from the optimized parameters. And if you optimize in
xyz-coordinates they are directly written as parameters I believe, just
not in the usual form.

Michael

That’s pretty annoying yes. However, we could probably regenerate the
xyz coordinates from the optimized parameters. And if you optimize in
xyz-coordinates they are directly written as parameters I believe, just
not in the usual form.

Of course, it’s possible, but reading of log will be much easier. Coordinates in log are in the same format as in out after frequencies run. It will be enough to permit reading of log by OB, and coordinates on all point will be readable. Also, I see there are many programs writing to more than one file. In future, it would be useful to read PUNCH (*.dat) from Gamess US to read hessian matrix and MO vectors


Regards,
Konstantin

(CCing openbabel-devel because this is mostly an OB issue I think)

On Mon, Nov 16, 2009 at 01:53:17PM +0300, Konstantin Tokarev wrote:

Of course, it’s possible, but reading of log will be much easier.
Coordinates in log are in the same format as in out after frequencies
run. It will be enough to permit reading of log by OB, and coordinates
on all point will be readable. Also, I see there are many programs
writing to more than one file. In future, it would be useful to read
PUNCH (*.dat) from Gamess US to read hessian matrix and MO vectors

Right, it is a general problem. If you use a GUI, then it is relatively
easy to ask the user to indicate all files which belong to the
calculation; but if you just do it non-interactively in OpenBabel then
just looking e.g. at a .log file of the same name in the same directory
might be error-prone if it belongs to a different calculation. For
MOLPRO, this might be the case if you did a geometry calculation before
and now do a single-point calculation at a different geometry using the
same input filename - the .log will still belong to the old calculation.

Michael