quantum-espresso/TODO

TODO LIST - May 2007

BUGS TO EXAMINE

- Zr pp buggy?
- strange empty states with C blyp ?
- O pseudopotentials?
- Zval is wrong for pseudopotentials generated on a ionic
  configuration, when semicore states are in the valence
- discrepancies between spin-unpolarized and spin=0 spin-polarized
  results with gradient corrections
- mysterious bfgs crash(-es) with ndim=3
- Lisenkov:
> Did anybody else report that pw.x hangs sometimes without any explicit 
> reason? I see it very often recently (after realizing 3.2 version) on 
> different machines (xeon em64t, cray xd1, cray xt3, altix,..). It usually
> happens when the number of scf steps is being large (~100). I didn't see
> such problem with earlier versions, but I need to use the latest one due 
> to using LDA+U method
- Sara Furlan:
> sto usando cpvib.x dell'espresso versione 3.2.
> Mi sono accorta che se in input si danno le ATOMIC_POSITIONS
> con un ordine diverso dagli ATOMIC_SPECIES si ottengono delle
> frequenze improbabili.

- Problems found by Eduardo: 
  - pw => cp with atoms out of the cell
  - extrapolation of wfcs

CVS/INSTALLATION/PORTING

- restructuration of the CVS server, dividing it into "base libs",
  "base QE", additional packages - release.sh is clumsy, simplify
- remove obsolete stuff from install/
- put misc scripts into a Utils/ subdirectory

COMMON

- reduce usage of nonscalable static and dynamical memory:
  qfunc, qrad, uspp, etc

- use qexml for xml file processing so that a single, easily
  exportable routine, is used everywhere. Requires some though
  on how to deal with parallelism without filling qexml with
  parallel stuff

- uniform and clear values for constants in Modules/constants.f90
  Almost done: now they have to be used everywhere!

- convert all single-precision constants to double precision
  (mostly done)

- pseudopotentials: support new format (xml) if/when ready
  add default cutoff values as well as radii of augmented charge
  in pseudopotential files

- adding/removing/modifying input variables is too complex
  Why are some checks on input variables performed in read_namelist,
  while others apparently similar are in */input.f90?

- time steps in ps/fs (both in input and in output)

- fix shift in eigenvalues between CP/FPMD and PW
  The eigenvalues should be the same
  Total energies for charged systems should also be the same

- merge spherical harmonics between CP-PW and FPMD
  merge integration routines, random number generators

- spherical bessel functions: only one function calculating dj_l/dx
  j_l with l=-1 should not be needed

- error is in flib/ but requires modules: not a good solution,
  things in flib/ should not use modules except maybe kinds 
  and other very low level modules 
- too many confusing error messages are still around

- replace calls to level-1 blas copy, scal, axpy, with f90 syntax

- CP "small boxes" vs PW "real-space" approach

- It would be nice to merge "cell_base_init" from CPV to PW,
  and also "cellmd" module of PW into "cell_base" of CPV

- there should be a single function or routine for periodic boundary 
  conditions (i.e. bringing all atoms inside the unit cell)

PW

- inversion symmetry should allow real hamiltonian and wavefunctions (Dario)

- nscf calculations are slow. There must be a way to make a better usage
  of the available information from the scf calculation: wavefunctions
  are just discarded. Same for phonon calculation: it shouldn't be needed 
  to recalculate everything almost from scratch at each different q-points

- atomic coordinates should be read from file and only optionally 
  from input in non-scf calculations: the present method is error-prone

- DIIS still doesn't work!

- Fermi-Dirac: pass T instead of "broadening", make it possible
  to use it on top of smearing for free-energy calculations

- add possibility to read atomic positions from file

- Use more standard choices for crystal axis (if any)

- make hpsi/spsi/newd/addusdens/CG/DIIS  faster
  - remove complex factor i**l from beta fct and q(r) (preliminary)
  - shift structure factor from beta to psi when computing becp (reduce memory)
  - use real BLAS routine instead of COMPLEX one in hpsi/spsi 
    (at least 2 times faster).
  - use only half of the G's when computing real integrals (2 times faster)
  - seek for CG and DIIS algoritms that only use (H-eS)|psi> and not
    the two vector separately ... compute it in one single call. (this way
    S|psi> is  inexpensive)

- output should be more informative and less confused
  the printout with UPF PP is incomplete or misleading (lloc=0)

- write explicitly and unambiguosly the units of output quantities
  and also what are exactly the output quantites!

- some symmetry-related routines such as sgama are too complex
  Verify that a group is a group (call multable again in sgama)

- printout of cpu times not very informative

- better memory estimate

- add traceback in error (error_handler module)

POSTPROCESSING

- complete postprocessing in Gamma case (only average missing),
  and with CP/FPMD data

- stm_wfc_matching not working

- Move to a xml-based format for the dynamical matrix files

- fix pp.x in parallel execution

- pwtools: add example of a plot of phonon dispersions
  Fix fqha.f90

PH, D3, Gamma

- C(R) might be calculated at the end of a dispersion run

- any possibility to merge the various solve_* ?

- use charge mixing instead of potential mixing

- move all phonon-specific stuff (symmetry etc) from pw into phonon

- clean up the logic of symmetry: there shouldn't be any longer the
  present confusion on whose symmetries we are dealing with. We should
  either keep unchanged the crystal symmetries and use indices to
  locate symmetries of the small group of q, or keep different copies 
  for the symmetries of the crystal and of the small group of q

- same input for Gamma and PH

- replace "use pwcom" into more "use" statements

- stop with 'prefix.EXIT' and restart (PH now has it; D3 & Gamma do not
  have any infrastructure for stopping like this)
  Phonon restart is inefficient in some cases

DOCUMENTATION

- move to wiki-based documentation

- single source (xml?) for text file, GUI help file, user guide

- better automated testing procedure, quick and distinct from examples!

- add some benchmarks

CP/FPMD:

- PWGui for CP/FPMD?

- check on small box size

- Proposals from Princeton:
  * default set up for molecular calculations
  * default choice for spin restricted and spin unrestricted configurations
    for neutral systems (maybe already there?)
  * automatic FFT grids from cutoff
  * automatic box grid for USPP from radii of augmented charge
  * default values for emass(emass=300), for dt (dt=7), for preconditioning
    cutoff (3)
  * the code should consider a common error:
      dt^2/emass too large does not allow ortho to converge
      or cause energy to increase as time step evolve
  * Electronic minimization: SD should be discouraged: use damp as default,
    introduce an automatic default schedule, something as:
    1st step sd followed by 5 steps with with damp= 0.8,
                followed by 5 steps with damp=0.5,
                followed by 10 steps with damp=0.3,
                followed by 10 steps with damp=0.2,
                followed by as many steps as necessary
                to achieve the required convergence with damp=0.1
   A max number of steps should be included to ensure program termination
   The other option allowed should be conjugate gradients: 
   see Nicola - eventually it could become the default
  * Ionic minimization: again SD should be discouraged
    A default scheme for simultaneous damped dynamics should be given
    (to be tested) for example: zero damp on ions and start with damp=0.5
    on electrons to become then 0.1 or perhaps the values should be set
    given the forces on the ions
    When moving ions and electrons simultaneously an important parameter
    is the ratio between electron and ion masses - For minimization it 
    is better to set up all the ion masses equal - A default value for
    the ion masses (considering the defaults for emass and dt) is perhaps
    10 AMU (we should do some test to see if 20 AMU is s a safer value)
  * Default values for randomization should be given
    amprp=0.1 is a decent value - amprp=0.01 is too small
  * Car-Parrinello dynamics: the proper masses for the ions, an optimal value
    for emass and dt should be set up by the code, based on the smaller atomic
    mass and the default value used in the minimizationi
    e.g. Amass_default=10 AMU. If the minimum physical AMASS is 20 then
    dt=sqrt(2) dt_default and emass should be increased so to keep
    emass^2/dt constant
  * defaults for the Nose thermostat