FLAIR/GEM - an Alternative to Wien 2K

Guenter's new Flair/Gem page with Si tutorial, basic instructions.

The purpose of this page is to solicit input/feedback from the students in Ph 575 on the use of Wien 2K as a teaching tool in Ph575. Please add/comment on the strength and weaknesses of Wien 2K and add/comment on proposed features of the new system (flair/gem) described below. It would help to mark your edits with your name, but you don't have to. You can also email Guenter at Guenter.Schneider@physics.oregonstate.edu and Janet at tate@physics.oregonstate.edu.

• web interface (both input/output) (gs)
  • almost no computer background necessary (gs)
• real research tool (gs)
  • potentially useful in your graduate/post-graduate research (many examples) (gs)
• shows the difficulties in doing research (gs)
• knows cif files

• limited access during project crunch time (gs)
  • see next point (not free)

• not free (gs)
  • can't use Wien 2K on your own computer (gs)
  • access restricted during graduate research (gs)
  • access requires license purchase during post-graduate research (gs)

• web interface not always easy or intuitive (gs)
  • if the defaults don't work you need to consult a very extensive handbook which was not written for beginners (gs)
  • there is no graded access to the input parameters. You have to go through all of them every single time. A scheme like essential, intermediate, advanced, and expert would allow focus on relevant parameters and you could expand the number of parameters as needed.(gs)

• not easy to figure out what is calculated besides the basics (band structure, DOS, Fermi energy) For example Wien 2K calculates the magnetic moment of a ferromagnetic material but you need to figure out yourself where you can find this number (gs)

The band structure program which will replace Wien 2K as a teaching tool in Ph 575 is named flair and is free. flair is based on the same technology (FLAPW = full-potential linearized augmented plane wave method)) and has pretty much the same strengths in terms of accuracy and speed as Wien 2K. flair exists and is used in research. A few features will hopefully be added before Spring term 2012.

We will use a python based interface (called gem) to interact with flair. The part of gem which handles the setup of a calculation mostly exists and it can read in cif files. The part of gem which handles output exists in parts but needs much more work. Eventually all output of flair will be directly available in python as data structures. gem is the system we use in my research group to do the same calculations you did in class (and some more complicated ones). It is a complex system but it is much easier and more logical than dealing directly with Wien 2K input and output.

On top of gem we will build a web interface unless I hear from you that that is really not needed. A web interface (really a form based interface) will always be limited in its scope (or really expensive) but there is always gem the python layer to access more interesting data.

• all computers in Weniger 412 + 497 (currently >30) can run flair+gem. No more access problems during crunch time.
• It is free and it can run on your Mac and Windows machines too.
• gem has graphics (2D/3D) built in.
• No need to access all parameters of flair. You can concentrate on the important ones.

This is your chance to shape future teaching! Please add here or email (see top of page for email addresses).

• What features would you like to see?
  • Automatice k-path generation …
  • tool-tips, direct help access.
  • examples?
  • add your ideas …
• Something you tried to do but couldn't?
  • add yours …

I would like the students to learn to use a real research tool that will be useful to them as they pursue research in solid state physics as experimentalists (as most will become). Obviously, as theorists, they will use this type of tool, too, but theorists will very quickly HAVE to go beyond the level presented in this course.

There must be a reasonable compromise between frustration (a barrier right now) and a black box (no learning).

Important things that should NOT be black-box:

• learn to interpret a cif file (at least find the lattice positions, lattice parameters, and know what's important) and set up a structure from crystallographic information. Distinguish between different origins (e.g. Si)
• Be able to monitor the convergence process
• Plot the density of states, partial density of states - same breakdown as in Wien (including interstitial)
• Understand that the size of the atom radii matters, and be able to find out how many electrons are “lost” by particular choice of atom radii
• Know the difference between core states and valence electron states
• Understand that the density of k-points is important and how to choose so that accuracy is maximized and computation time is minimized.
• Plot the band structure (“spaghetti diagram”), and be able to choose a k-path for the band structure plot (as in x-crysden)

Things that should be black-box, but aren't in Wien:

• swapping the k-files for DoS and spaghetti should be automatic
• fewer error messages
• pointing and clicking through the scf is pointless
• Having to find the Fermi energy and manually substitute it in one file is bad

Capabilities of the new program

• Be able to calculate magnetic moments (need a sub list here to see how automated this should be)
• Be able to calculate optical absorption (need a sub list here to see how automated this should be)
• Be able to plot electron density (need a sub list here to see how automated this should be)
• Be able to plot Fermi surface (need a sub list here to see how automated this should be)
• find total energy minimum, and hence optimize lattice constants