Data manager

Development
1

I want to get back to my feature request of adding pwscf calculated DOS plotting. Actually, what I meant was getting back to my favorite topic: data manager. As an exampled here http://www.wien2k.at/features/img93.png can be seen. The truth is that lots of software packages calculate lots of data but manipulating it is really challenging. I mentioned that before and I’ll say again: creating a data manager that can not only plot IR spectra, but plot various calculated properties from the file. This would avoid adding one by one (IR,NMR) but instead would recognize various data and allow plot them, instead of asking David implement one or another. Once a layout is created for it, formats can be added as needed

As an example, I have bunch of files calculated with pwscf and I need to sum up selected files and plot them (DOS). I can’t find anything automated, everything would be semi manual. I am sure more people would find that functionality useful

Jonas

2

Jonas,

It’s not as if I’m ignoring your request. It’s just not as easy as you
think. If I could just sit down and have a DOS plot feature in a few
hours, it would be done by now. And a “data manager” (or in this case,
a “spectrum plot” manager) is a good idea.

But let me give a brief summary of what needs to happen. So let’s say
I want to add a plot of Density of States (DOS):

  1. I need to code something which reads the orbital (or state)
    energies from a file. Keep in mind, that needs to happen for every
    single QM program (i.e. Gaussian, MOPAC, Molpro, NWChem, Q-Chem,
    GAMESS) since each program’s file format is different. Much less work
    if you’re just doing a GUI for one program.
    NOTE: this is why Open Babel is a lot of help – lots of the file
    parsing code exists there, plus the data structures to appropriately
    store it.

  2. OK, now I have the energies. Most people want to see them plotted
    in eV, but most programs report them in Hartree. Except a few
    screwball ones. So you’ll need to convert the energies somewhere along
    the way. I include step #2 as a discrete step, because for some data
    (i.e., NMR) there’s some work to convert from QM data (isotropic
    shifts) into something a user wants (ppm). In the case of DOS, you may
    want to set the Fermi energy as the zero point.

  3. Now I probably want to add some Gaussian or Lorentzian broadening.
    (Keep in mind that the energy scales are very different for IR vs. NMR
    vs. DOS).

  4. Finally plot.

So there are two parts to a feature you describe – what you (the
user) sees, and what the programmer sees. David now has a nice
“Spectra” window, which can handle all sorts of plots. It can probably
become the “data manager” you describe, with a nicer interface than
I’ve seen in the screenshots you post.

But the hard part is the backend grunt work – coding the data
structure for IR or NMR or UV/Vis or DOS, then writing file parsers
and standardizing the data. If Avogadro were a commercial product with
several full-time coders, it might get done faster. But saying the
code would “recognized various data” … “instead of implementing one
or the other” sounds like you don’t understand how much work it takes
for each type of data and for each QM product.

In short, I don’t think any of us are blowing off your suggestions.
They’ll just take some time to get done, and so far all the active
coders have long To Do lists.

Cheers,
-Geoff

On May 19, 2009, at 11:27 PM, Jonas Baltrusaitis wrote:

I want to get back to my feature request of adding pwscf calculated
DOS plotting. Actually, what I meant was getting back to my favorite
topic: data manager. As an exampled here http://www.wien2k.at/features/img93.png
can be seen. The truth is that lots of software packages calculate
lots of data but manipulating it is really challenging. I mentioned
that before and I’ll say again: creating a data manager that can not
only plot IR spectra, but plot various calculated properties from
the file. This would avoid adding one by one (IR,NMR) but instead
would recognize various data and allow plot them, instead of asking
David implement one or another. Once a layout is created for it,
formats can be added as needed

As an example, I have bunch of files calculated with pwscf and I
need to sum up selected files and plot them (DOS). I can’t find
anything automated, everything would be semi manual. I am sure more
people would find that functionality useful

Jonas

3

Hi,

On Tue, May 19, 2009 at 08:27:57PM -0700, Jonas Baltrusaitis wrote:

I want to get back to my feature request of adding pwscf calculated
DOS plotting. Actually, what I meant was getting back to my favorite
topic: data manager. As an exampled here
http://www.wien2k.at/features/img93.png can be seen.

I see a DOS plot here, not a data manager. I think it is agreed upon
that plotting the DOS is a desired feature. If wien2k has a good data
manager, can you describe it and/or post screenshots of it, please?

The truth is that lots of software packages calculate lots of data but
manipulating it is really challenging.

What do you mean with manipulating? Do you want to modify some values
of your DOS calculation to make it look different? Do you want to
modify some input values (like geometry) of your system and have the DOS
recomputed?

I mentioned that before and I’ll say again: creating a data manager
that can not only plot IR spectra, but plot various calculated
properties from the file. This would avoid adding one by one (IR,NMR)
but instead would recognize various data and allow plot them, instead
of asking David implement one or another. Once a layout is created for
it, formats can be added as needed

I think it does not make sense to plot arbitrary data, e.g. plotting the
dipole moment would be pretty pointless I guess?

Having a generic some-energy-scale-vs-some-intensity-scale plot for
various data would be nice though. But I guess it would already break
down for e.g. orbital energy vs. occupation number.

As an example, I have bunch of files calculated with pwscf and I need
to sum up selected files and plot them (DOS). I can’t find anything
automated, everything would be semi manual. I am sure more people
would find that functionality useful

OK, so now I probably understand a bit better what you meant by
“manipulating” above, you want to merge/blend data from different
sources for your plot?

Maybe this could be achieved through some Avogadro extension. Maybe it
makes more sense to do this as an external OpenBabel-based application
and then use Avogadro to render the results, not sure. In any case I
think this is rather an advanced use-case compared to general-purpose
plotting.

Michael