<html>
<head>
<title>v2 : README</title>
</head>
<body>

<h5>Copyright (C) 1997-2007 ABINIT group (DCA, XG)
<br> This file is distributed under the terms of the GNU General Public License, see
~abinit/COPYING or <a href="http://www.gnu.org/copyleft/gpl.txt">
http://www.gnu.org/copyleft/gpl.txt </a>.
<br> For the initials of contributors, see ~abinit/doc/developers/contributors.txt .
</h5>

<pre>
This directory, v2, contains tests which exercise parts 
of the ABINIT package, especially those features acquired
during the development of the version 2.
This file gives first a very brief description
of the tests, then a much longer one.

These tests are designed primarily to exercise parts of the code
quickly, NOT necessarily to give physically sensible results.
For tests of correctness, see directory physics (under development).
For greater speed, some tests are not run to full convergence.
Also the quality parameters (especially ecut) are sometimes minimal, i.e.
the calculations are underconverged.

A classification of the tests is provided now. Later, each test is described
in more detail, with accompanying notes and references (some of the notes
might have to be updated, due to the small possible drift related to improvement
of the accuracy of the code).

Tests 1 to 39 as well as 90 to 99 focus on RF features.
Tests 40 to 89 focus on GS features.

Tests 1 to 8 are simple tests of phonon frequency calculations,
   reproducing results obtained by using the "old" RESPFN code. 
Tests 9, 10 and 11 are further simple tests, with a local psp and q/=0,
   as well as partial occupation numbers 
Test 12 is also a simple test with q/=0, BCC.
Tests 13 to 15 treat different DDBs for SiO2 thanks to MRGDDB and ANADDB,
   reproducing results obtained by using the "old" versions of 
   these codes.
Tests 16 and 17 treat a Si DDB, and reproduce IFCs, phonon frequencies
   zero-point contribution to Helmholtz energy and entropy at 298K,
   as published in the paper by Rignanese et al, PRB53, 4488 (1996).
Tests 18 to 21 treat a BaTiO3 DDB, and reproduce the convergence
   study with sampling of the Brillouin zone for q wavevectors,
   as done in Ghosez et al, Ferroelectrics 206-207, 205 (1998),
   as well as the analysis of IFCs in term of dipole-dipole 
   contributions and short-ranged contributions. Eigenvectors are also
   analyzed.
Test 22 deal with a ZrO2 DDB.
Tests 23 to 24 reconstitute a BaTiO3 DDB for Gamma and X points,
   and then produce the phonon frequencies at these points, that
   can be compared with results of test 18.
Test 25 treats a PbZrO3 DDB.
Tests 26 to 29 compute phonon band structures for Al.
Tests 30, 31 and 32 show the complete procedure needed to generate
   a phonon band structure, from the ground-state calculations to the
   interpolation of dynamical matrices. In order to keep the CPU time
   reasonable, the parameters (ecut, nkpt and sampling of q wavevectors)
   are much too small.
Test 33 to 36 compute accurately the derivatives of the energy
   for a H2 or Al2 molecule in a big box, and compare to finite-differences.
Test 37 compares the frequencies at two different q points, one in a position
   of high symmetry, and one in a position of low symmetry.
Test 38 focuses on pseudopotentials with a non-linear core correction.
Test 39 FCC unit cell. Show how to use only the q Gamma point
     to obtain symmetric IFCs using Anaddb.

Tests 40 to 89 are related to GS computations. 

Tests 40 and 41 checks the interplay between GGA, 
   non-linear XC core correction, spin, forces and stresses.
Test 42 : TDDFT based on xOEP
Tests 43 and 44 check the k point generator and the input array angdeg
Tests 45 and 46 check SCF and numeric convergence in the metallic case
Test 47 checks the generation of charge densities for selected states
Test 48 and 49 check a modified Broyden algorithm (ionmov=3) and
   the Verlet algorithm (ionmov=6 and 7)
Test 50 and 51 test metallic, spin-polarized, non-linear XC core
   correction, GGA calculation of stress.
Test 52 to 55 checks the symmetry finder.
Test 56 tests a FHI pseudopotential with non-linear XC core correction
Test 57 checks the calculation of frequency dependent susceptibilities.
Test 58 test the mechanism for vacuum identification
Tests 59 to 60 checks the correctness of space group symbols to generate
   conventional as well as primitive cells.
Tests 61 to 73 test the different sets of k points generated automatically
   using prtkpt/=0.
Tests 74 and 75 : the effect of spin-orbit coupling
Test 76 : conversion of input wavefunctions
Test 77 to 80 : cut3d
Test 81 and 82 : Berry phase calculation of the polarization
Test 83 : FCC Ni ferromagnetic 
Tests 84, 85 and 86 : again the effect of spin-orbit coupling
Tests 87 and 88 : Nose thermostat and Langevin dynamics.

Tests 90 and beyond are related to RF computations.

Tests 90 to 97 test carefully metallic RF of phonons: H, Li and Ni in
   simple cubic structure, with double and quadruple supercells.
Test 98 present RF for insulator with non-zero ecutsm and dilatmx
Test 99 RF of Niobium, at a rather large wavelength


To run these tests, do the following:

0. Be sure that the perl script "run-standard-tests", in the main directory,
   has been generated from the primitive file "run-standard-tests.pl",
   thanks to the command "make perl"
   issued in ~abinit (see the installation notes on the Web).

1. Submit the "run-standard-tests" script, specifying a machine, and
   the keyword "v2" (for Tests_v2) and either
   the index of a test case, i.e.
   (run-standard-tests name_of_machine v2 22) >& log_file
   or a whole range of test cases (two indices), i.e.
   (run-standard-tests name_of_machine v2 02 08) >& log_file
   or, if you want to run all the test cases of this directory
   (run-standard-tests name_of_machine v2) >& log_file
   This will send stdout and stderr to log_file.
   The script 'run-standard-tests', will create a subdirectory of ~abinit/tests/v2,
   with the name_of_machine and the
   date, where all the results will be placed.

2. In that directory, you will find for each test case that you have
   run, a log file (with the name of the test case), an output
   file, but also a 'diff.xxx' file, automatically created by making
   a 'diff' with respect to the "Refs" subdirectory output files.
   It contains output files from a recent version of the ABINIT code.
   There may be large differences in timing but there should only
   be minor differences in the output of physical quantities.

3. There is also a global report file, generated by the use of the
   fldiff script. Its name is fldiff.report . See the last
   version of the  ~abinit/doc/install_notes/install** file
   in the Infos directory for information about the use of this file.
   This file is the most convenient for a quick look at the correctness
   of results. When the results are not correct, one has often to
   rely on the 'diff.xxx' file to understand what was going wrong.


**********

Test cases:

  1. Linear chain of Ge2 molecules (2 atoms per unit cell),
     using a local (Starkloff-Joannopoulos) pseudopotential,
     and no exchange-correlation (ixc=0). Uses 4 k-points.
     Computation of the second derivative of the total energy
     with respect to a atomic displacement along the chain,
     with q(0 0 0) wavevector.
     (see test.ge.xc0 of RESPFN)
     
  2 and 3. Linear chain of Ge2 molecules (2 atoms per unit cell),
     using a local (Starkloff-Joannopoulos) pseudopotential,
     with exchange-correlation (ixc=3). Uses 2 k-points.
     Computation of the second derivative of the total energy
     with respect to a atomic displacement along the chain,
     with q(0 0 0) wavevector.
     Same as test 1, except ixc and the k-points, and
     except that it doesn't uses multi-data mode, so GS and RF
     calculations are separated.
     (see test.ge.xc5 of RESPFN)

  4. Linear chain of Si2 molecules (2 atoms per unit cell),
     using a separable pseudopotential, with ixc=3.
     Computation of the second derivative of the total energy
     with respect to a atomic displacement perpendicular to 
     the chain, with q(0 0 1/2) wavevector.
     (see test.si.chain of RESPFN)

  5. Linear chain of Si2 molecules (2 atoms per unit cell),
     using a separable pseudopotential, with ixc=3.
     Computation of the second derivatives of the total energy
     with respect to a atomic displacement along the chain,
     with q(0 0 0) wavevector, as well as to an homogeneous
     electric field. The computed derivatives include the
     mixed derivative wrt the two perturbations.
     (see test.si.elfd of RESPFN)

  6. Si crystal (diamond structure), 2 atoms per unit cell. 
     Computation of responses to atomic displacements
     and homogeneous electric field at Gamma,
     giving access to the dynamical matrix at q(0 0 0), including
     its non-analytical behaviour (effective charges
     do not vanish completely due to lack of convergence), 
     the associated phonon frequencies, and the macroscopic 
     dielectric constant.
     (see test.si.gamma of RESPFN)

  7. GaAs crystal (zinc-blende structure), 2 atoms per unit cell.
     Computation of responses to atomic displacements
     with X-point wavevector, giving the dynamical matrix and
     the associated phonon frequencies.
     Can be compared to the test gaas.x of RESPFN, provided
     the cut-off energy is changed (here 3Ha, in RESPFN, 6Ha).

  8. Linear chain of Si2 molecules (2 atoms per unit cell),
     using a separable pseudopotential, with ixc=5 
     and non-linear XC core correction.
     Computation of the second derivatives of the total energy
     with respect to a atomic displacement along the chain,
     with q(0 0 0) wavevector.
     (see test.si.core0 of RESPFN)

  9. Same as test 2-3 (Germanium linear chain, with a local psp),
     but at non-zero 0 : q=(0 0 0.5). 
     Similar to test4, actually, with Si changed to Ge.

 10. Similar to test 9 (Germanium linear chain, with a local psp),
     with a badly commensurate q vector : q(0 0 0.21).

 11. Orthorhombic Al system, q(1/2 0 0), partial occupation numbers.
     Compares with frozen-phonon calculations. Examine two cases:
     varying occupation numbers, and fixed occupation numbers.
     A. Varying occupation numbers (occopt=4): 
        RF calculation in dataset 5 gives
        2DE equal to -3.812577 Ha, while finite difference of "TOTAL"
        energy (datasets 6 and 7, including entropy term) gives -3.812230 Ha
        (the agreement could be better with better finite difference)
     B. Fixed occupation numbers (occopt=2): 
        RF calculation in dataset 8 gives
        2DE equal to +5.431807 Ha, while finite difference of total
        energy (no entropy term is present with occopt=2) gives +5.431857 Ha
        (the agreement could be better with better finite difference)

 12. BCC Helium q close to Gamma, along Gamma-H
     There is a problem with kptopt=1 or 3 for the NSC step.

 13. Analyze a simple DDB for quartz, but do not test interatomic
     force constants (see test 15 for this). Compute phonon frequencies
     at gamma with and without LO-TO splitting.

 14. Generates a DDB for quartz containing the 2DTEs at three 
     q-points : (0 0 0), (0 0 1/4) and (1/2 0 1/4).
     This is a standalone test of MRGDDB.

 15. Analyze the DDB for quartz build in test 14.
     Compute the interatomic force constants, the phonon Density of States,
     different thermodynamical functions of temperature,
     phonon frequencies in the whole Brillouin zone,
     oscillator strengths and mode effective charges at q(0 0 0),
     the electronic dielectric tensor, and the (full) dielectric tensor
     at zero frequency (so including atomic displacements and 
     related polarization).

 16. Analyze a DDB for Silicon. Lattice parameter = 10.18 Angstrom.
     Here, IFCs are obtained, and can be compared to those
     published in Table I of Rignanese et al, PRB53, 4488 (1996).
     For example, the atoms numbers for NN=0,1,2,5 in Table I
     are found as atoms 1,5,17,20. The numbers are equal in the 
     paper or in the output of the code.
     The frequencies at X and L points are also computed. They
     can be compared with those provided in TABLE II (TA(X) mode
     at 140.466 cm-1) and TABLE III (TA(L) mode at 108.626 cm-1)
     in the above-mentioned paper. The numbers are equal in the
     paper or in the output of the code. 

     Thermodynamic quantities are also obtained. The input
     parameters of the code does not give particularly well
     converged quantities (one needs reasonable CPU time for testing !),
     but it is OK for comparison with the paper.
     In the output of the code, a mole corresponds
     to the number of Avogadro times one CELL. As a Silicon cell contains
     2 atoms, caution must be taken in the comparison with experiment !
     We will refer to the numbers given by the code as being given
     for a mole-cell (in short a mol-c). The distinction between
     the usual definition of a mole (Avogadro number times 1
     Silicon atom) and a mole-cell (Avogadro number times 2 Silicon atom) 
     was unfortunately the source of errors in the paper by 
     Rignanese et al (our thanks to Steve Erwin for noticing 
     this - an errata should be written). From the output of the code 
     ( zero Kelvin is approximated by results at 1 Kelvin) : 
     the zero point contribution to the Helmholtz free energy
     is 11.90 kJ/mol-c , so 5.95 kJ/mole ; the entropy at 298.15K
     is 38.25 J/(mol-c.K), so 19.12 J/(mole.K) ; the constant-volume
     specific heat at 298.15K is 39.59 J/(mol-c.K), so 19.80 J/(mole.K) ;
     the change in F (Helmholtz free energy) from 1 K to 298.15 K is
     -4.91 kJ/mol-c, so -2.45 kJ/mole ; the change in E (internal energy) 
     from 1 K to 298.15 K is 6.49 kJ/mol-c, so 3.25 kJ/mole. Supposing 
     that we take the usual definition of a mole (Avogadro number times 1 
     Silicon atom), then, in the above-mentioned paper, the FIG. 3, 5 and 8
     should be rescaled, as well as the zero-point contribution to the
     Helmholtz free energy (5.95 kJ/mole, and not 12 J/mole - the 'k' was 
     also missing), that is only 1.25% of the cohesive energy.
     The output of the code is now more clear than in 1995, and this
     kind of error should not appear anymore.
     

 17. Analyze a DDB for Silicon. Lattice parameter = 10.18 Angstrom.
     IFCs are computed as in test 16, but by limiting artificially
     the number of allowed interacting shells to 2. The total number
     of atoms to be taken into account is 17 (1 for the shell 0,
     4 for the shell 1, and 12 for the shell 2).
     The frequencies at X and L points are again computed. They
     can be compared with those provided in TABLE II (TA(X) mode
     at 161.684 cm-1) and TABLE III (TA(L) mode at 127.508 cm-1)
     in the above-mentioned paper. The number are close, with
     small differences at the level of 0.001 cm-1 .
     
 18. Analyze a DDB for BaTiO3 (see also tests 19 and 20). Here, 
     no use of IFCs is done. Phonon frequencies at the
     Gamma, X, M, R and along the Gamma-R line are obtained
     directly from the DDB. They are to be compared
     with the data in the TABLE II of Ghosez et al, Ferroelectrics,
     206-207, 205 (1998), as well as FIGURE I. 
     At q(1/8 1/8 1/8), the lowest frequency is  i 136.7 cm-1 .

 19. Analyze a DDB for BaTiO3, same as test18, but IFCs are
     obtained from a 2x2x2 non-shifted grid (referred to as M1
     in the above-mentioned paper by Ghosez et al).
     At q(1/8 1/8 1/8), the lowest frequency is  i 178.8 cm-1 .

 20. Analyze a DDB for BaTiO3, same as test18, but IFCs are
     obtained from a 2x2x2 BCC grid (referred to as M2
     in the above-mentioned paper by Ghosez et al).
     At q(1/8 1/8 1/8), the lowest mode is  i 153.5 cm-1 .
     The error, compared with the reference result of test 18,
     is still 10%, but much less than with the grid of test 19.
     Also, the IFCs are obtained and can be compared with those
     provided in TABLE III-V of the Ghosez et al paper.
     Note the decomposition between dipole-dipole contribution
     and short-ranged contribution.

 21. Analyze a DDB for BaTiO3, same as test18.
     Compute the phonon frequencies at Gamma, and also output
     the corresponing eigenvectors. These data can be compared
     with those published by Ghosez et al, Ferroelectrics 194, 39 (1997),
     TABLE II, although the normalisation factor is different.
     For example, the z displacement of mode 1 given by the code
     is ( .25d-4 , .158d-2 , -.261d-2, -.118d-2 , -.118d-2 ), that
     multiplied by -60.54 and rounded to the third digit, 
     gives values that appear in the first line of that TABLE II :
     ( -0.002 , -0.096 , 0.158 , 0.071 , 0.071 )

 22. Analyze a DDB for ZrO2, containing only the dynamical
     matrices at Gamma, X and L.
     Generate the IFC using a rough sampling, based on Gamma and X
     only, then produces the phonon band structure along Gamma-X,
     as well as in L.
     The degeneracies can be compared with those of 
     Detraux et al, Phys. Rev. Lett. 81, 3297 (1998).
     The errors produced with this sampling are still large, as can be
     judged from the comparison at L. The frequencies (in cm-1) from the DDB are
     128.8 (2-deg), 271.4, 413.7 (2-deg), 518.2, 521.8 (2-deg), 598.1 ;
     while from the IFCs, using the interpolation, they are
     152.5 (2-deg), 290.8, 429.6 (2-deg), 434.6 (2-deg), 551.9, 576.0 .

     The same DDB allows to generate a slightly better sampling, by using
     nqshft=2 and two shifts (0.0 0.0 0.0) and (0.5 0.5 0.5).
     In this case, the L point is used for the interpolation. The comparison
     can be done at the level of the mid-point between Gamma and X :
     from the GX grid, one gets
     107.0 (2-deg), 260.3, 295.5 (2-deg), 394.3, 577.0 (2-deg), 682.1  ;
     from the GXL grid, one gets
     106.3 (2-deg), 262.6, 302.7 (2-deg), 403.8, 582.5 (2-deg), 684.2  ;
     while directly from the ABINIT code, one gets
     106.3 (2-deg), 270.0, 277.9 (2-deg), 443.2, 581.6 (2-deg), 681.6  .
     This is not yet satisfactory.
     More points are needed for better results, the next interesting grid
     being obtained by changing ngqpt from 2 2 2 to 4 4 4 in the t22.in file.
     Two additional dynamical matrices are then needed.

 23. Generates a DDB for BaTiO3 containing only the 2DTEs at the 
     (0 0 0) and (1/2 0 0) q-points.
     This is a standalone test of MRGDDB.

 24. Analyze the DDB of test23. Uses asr=0. 
     If asr is set to 1, the results of test 18 are recovered.
     The effect of asr choice is important in the case
     of BaTiO3, with the pseudopotentials that were used,
     see Ph. Ghosez's thesis.
     The option asr is shown to work on both Gamma and X points :
     the correction is obtained at Gamma, and transferred to X.
     Note that a DDB MUST include the Gamma point.

 25. Analyze a DDB for PbZrO3. It is also a ABO3 compound, like
     BaTiO3 (see tests 18 to 21), but the DDB contains the q-wavevectors
     on a 2x2x2 FCC grid, finer than those tested for BaTiO3.
     Output : analysis of IFCs, as well as a few phonon frequencies.

 26. Generate first-order responses for FCC Aluminum.
     Very low cut-off, to keep CPU the lowest possible
     Aim at a regular sampling of phonon wavevectors, needed to
     interpolate the dynamical matrix over the whole Brillouin Zone,
     in test 28. The chosen grid (too coarse, though) is
     (0 0 0), (1/4 1/4 0), (1/2 1/2 0), (1/2 0 0),
     (1/2 1/4 1/4), (1/2 -1/4 1/4)  (in reduced coordinates).
     The ecut and nkpt parameters are really too low (see test 28), 
     but parameters suitable for physical results (see test 29) 
     are also provided in the input files.

 27. Combines the DDBs of test 26.

 28. Phonon band structure of Al, from DDB of test 27.
     The parameters were really too low in test 26, so that some 
     phonon unstabilities are present close to Gamma. These
     do not appear anymore in test 29.

 29. Phonon band structure of Al. Similar to test 28,
     except that the DDB was generated with parameters much better
     than those of test 26, the better parameters are mentioned 
     in the t26.in file, for information. The convergence is not complete
     though, but the frequencies compare already rather well with those
     in Quong and Klein, PRB 46, 10734 (1992), except
     close to Gamma (still, no instability is observed, unlike in test 28).

 30. Generate first-order responses for GaAs (zinc-blende).
     Aim at a regular sampling of phonon wavevectors, needed to
     interpolate the dynamical matrix over the whole Brillouin Zone,
     in test 32. The chosen grid (too coarse, though) has only two 
     special q points : 1/4 1/4 1/4 and 1/4 1/2 1/2 (in reduced coordinates). 
     The Gamma point is also needed, especially for the computation of 
     dielectric matric and effective charges, needed to get the asymptotic
     behaviour of the interatomic force constants. To generate
     the different responses, the following steps are followed, 
     with each of them corresponding to a different dataset (the multi-dataset
     mode allows to have only one input file) :
     1) ground state calculation, with a k-point sampling in the irreducible
       Brillouin zone only; 
     2) using the density of 1), computation of the wavefunctions for the
       grid of k-points in the full Brillouin zone; (not really needed in v3.0)
     3) computation of the ddk response, in preparation to the electric field 
       response ;
     4) computation of the dynamical matrix at Gamma, as well as the
       Born effective charges, and the dielectric tensor ;
     5) using the density of 1), computation of the GS wavefunctions
       at k+q where q is 1/4 1/4 1/4 ;
     6) computation of the dynamical matrix at 1/4 1/4 1/4 ;
     7) using the density of 1), computation of the GS wavefunctions
       at k+q where q is 1/4 1/2 1/2 ;
     8) computation of the dynamical matrix at 1/4 1/2 1/2 .
   
 31. The three derivative databases generated at steps 4, 6 and 8 of test 30
     are merged, using the MRGDDB utility.

 32. The derivative database generated by case 31 is now analyzed, and serves
     to compute a phonon band structure.
     There is a very small breaking of the symmetry-induced degeneracy
     of modes along the (x x x) direction. This is due to an incomplete 
     convergence. Indeed, if one makes tolwfr more stringent 
     for response calculations (from 1.0d-16 to 1.0d-20), the 
     lowest acoustic modes at (0.1 0.1 0.1), that are 4.042cm-1 and
     4.067cm-1, becomes 4.05398 cm-1 and 4.05400 cm-1 . 
     Of course, this effect cannot be seen for other modes, since only
     the acoustic modes are sensitive to that level of accuracy ! 
 
 33. H2 molecule in a big box : compute VERY accurately
     the derivatives of the energy, by both symmetric finite-differences and
     direct computation of forces and 2DTE.
     Also test the interplay between istwfk/=1 in the GS calculation
     and istwfk==1 in the RF calculation (istwfk/=1 is not yet-991020-
     allowed for RF, which is a shame)
     1) Computation of the first-order derivative of the total energy
     With delta(xred)=0.0002, one gets delta(etot)/delta(xred)=-3.145846551
     With delta(xred)=0.0001, one gets delta(etot)/delta(xred)=-3.145836932
     The combination of both results, in a higher-order finite difference
     formula gives -3.145833726 . The direct computation of forces
     at the target geometry gives -3.145833725869 . The agreement is perfect,
     taking into account the "limited" number of digits (10) of the 
     finite-difference result.
     2) Computation of the second-order derivative of the total energy
     With delta(xred)=0.0002, one gets delta(dedt)/delta(xred)=188.73875
     With delta(xred)=0.0001, one gets delta(dedt)/delta(xred)=188.73837
     The combination of both results, in a higher-order finite difference
     formula gives 188.73824613 . The direct computation of 2DTE
     at the target geometry gives 188.73824613046 . The agreement at the
     level of 11 digits is also perfect.

 34. Again H2 molecule in a big box (like test 33). 
     With the same configuration and parameters as test 33, 
     investigate the treatment of unoccupied states : 
     use nband 2 and occopt 1 , causing occ 2.0 1.0 . The same results
     as with test 33 are obtained.

 35. Al2 molecule in a big box. Treat 8 bands, with some of them
     partially occupied. The occupation numbers are fixed, with occopt 0 .
     Computation of the second-order derivative of the total energy.
     With delta(xred)=0.0001, one gets delta(etot)/delta(xred)=3.32914893
     The direct computation of 2DTE at the target geometry 
     gives 3.3291477145164 . The agreement is good, and can be improved
     if a higher-order finite difference estimation is used.
     
 36. Al2 molecule in a big box. Treat 8 bands, occupied using occopt=4.
     For a metallic occopt, the frozen-phonon (finite-difference) approach
     is strictly equivalent to the RF approach only at q/=Gamma.
     Here, computes the result at q=(0 0 1/2).
     With delta(xred)=0.0001 (DATASET 5), one gets 
     delta(etot)/delta(xred)=0.42807994 .
     The direct computation of 2DTE at the target geometry
     gives 0.428080350 .

 37. MgO FCC crystal, with very low cut-off, and 32 k points. 
     Pseudopotentials WITHOUT a non-linear core correction.
     Compute the dynamical matrices at q(0.25 0.5 0.498) and q(0.25 0.5 0.5). 
     The results should be very close to each other, but were not in pre v2.1
     versions, because the symmetry was not treated correctly for the
     highest-symmetry q vector q(0.25 0.5 0.5) (test case found by PTepesch).

 38. MgO FCC crystal, with very low cut-off, and 32 k points. 
     Pseudopotentials WITH a non-linear core correction.
     Compute the dynamical matrix at q(0.5 0.5 0.5) .
     Also test the non-type-ordering of atoms.

 39. BaSr(TiO3)2 FCC unit cell. Show how to use only the q Gamma point
     to obtain symmetric IFCs using Anaddb.

 40. O2 (nsppol=1, non-linear XC core correction, GGA) 
     Computation of forces in the GGA, and comparison with a finite
     difference of energy.
     The direct computation of force (dataset 2) gives 0.627251486 Ha/Bohr
     A simple finite-difference estimation (dataset 1 and 3) gives
      0.627251265 Ha/Bohr. The agreement can be improved if a better
     finite-difference estimation is used.
     
 41. O2 (nsppol=1, non-linear XC core correction, GGA)
     Computation of uniaxial stresses in the GGA, 
     and comparison with a finite difference of energy.
     The direct computation of sigma(zz) stress (dataset 2) 
     gives 0.964267876d-3 Ha/Bohr**3
     A simple finite-difference estimation (dataset 1 and 3) gives
     a difference in total energy of 0.085048409d-3 Ha,
     for a difference of volume of 0.0882 Bohr**3 leading
     to a stress estimation of 0.96426768d-3 Ha/Bohr**3
     The agreement can be improved if a better finite-difference 
     estimation is used.

 42. (HeH)+ in a big box. Computation of excitation energies in TDDFT,
     in the following approximations : TDxOEP/xOEP, TDLDA/xOEP,
     BPG hybrid/xOEP.

 43. BCC Molybdenum. Test the k point generator : Monkhorst-Pack
     grid, then band structure.

 44. Arsenic in rhombohedral structure.
     Optimization of unit cell size and shape.
     With 4x4x4 k point grid, ecut 3.0, occopt 4 and tsmear 0.06d0,
     one get acell 3*7.325  angdeg 3*57.36 and xred 0.2277
     These input parameters are too small, but still give a realistic
     geometry, since with the much better parameters
     12x12x12 k point grid, ecut 12.0, occopt 4 and tsmear 0.04d0,
     one gets acell 3*7.633  angdeg 3*54.95 and xred 0.2296, while
     the experimental values are
     acell 3*7.751   angdeg 3*54.554 and xred 0.2276 

 45. Aluminum in FCC structure.
     For occopt=4,5,6,7 , determination of the total energy
     at slightly different tsmear values, and check of the
     dE/d(tsmear)=-kT.entropy relationship.
     For example, for occopt=4, the difference between tsmear=0.101 and
     tsmear=0.099 is -11.149450d-6 Ha , while the value of -kT.entropy
     is -5.574709d-4Ha, that is about 500 larger, as expected. 
     
 46. Aluminum in FCC structure.
     At fixed tsmear (=0.04), check the convergence with respect
     to the number of k points, for different occopt.
     The grids that are tested have respectively 10, 28, and 60 k points
     in the IZB (ngkpt values are 4 4 4 4, 6 6 6 4 and 8 8 8 4
     For occopt=4, the total energy (or the enthalpy) 
     is -2.079770 Ha, -2.080534 Ha, -2.080672 Ha ;
     for occopt=5, it is -2.079787 Ha, -2.080563 Ha, -2.080693 Ha ;
     for occopt=7, it is -2.084211 Ha, -2.084861 Ha, -2.084916 Ha.
     The latter values can be corrected by computing the mean of the
     total energy and the internal energy, as described in Marzari's 
     thesis, for example. Thanks to this procedure, one gets :
     -2.079732 Ha, -2.080470 Ha, and -2.080626 Ha.
     For information, with a 12 12 12 4 grid (182 k points) 
     and occopt=4, one gets -2.080648 Ha.
 
 47. Si in diamond structure. 2 k points, low ecut.
     Output the SCF density, then partial densities that
     correspond to the 1st, 2nd, 3rd and 4th valence bands, 
     then the density that corresponds to the 1st conduction band,
     then the density of the lowest conduction state at 1/4 1/4 1/4, then
     the density of the highest valence state at 0 0 0 .
     Also test the symmetry finder.

 48. H2 molecule in a big box.
     Comparison of the modified Broyden algorithm (ionmov=3) with the
     original one (ionmov=2). Start with different values 
     of xcart, from 0.6 to 1.1, by step of 0.1 . The number of Broyden
     steps needed to reach acceptable residual forces
     with the ionmov=3 algorithm are : 3, 2, 2, 3, 3, 4, while
     with the ionmov=2 algorithms, one get : 4, 3, 2, 5, and then,
     either the algorithm does not converge within 8 steps, or it
     converges to a saddle point of the energy !
     This test was hard to make portable. This is why the tolerance
     for fldiff is very large.

 49. Si2 molecule, static, spin-polarized.
     Same system as test 17 of fast, except lower ecut.
     Test ionmov=6 (Verlet) as well as ionmov=7.
    
 50. O2 (nsppol=2, non-linear XC core correction, GGA)
     Some similarities with test 41, except that nsppol=2 instead of 1.
     Computation of uniaxial stresses in the GGA,
     and comparison with a finite difference of energy.
     The estimation of sigma(zz) stress  at acell(3)=9.0
     from acell(3)=8.9991 and 9.0009
     gives 1.62867d-3 Ha/Bohr**3
     A simple finite-difference estimation gives
     a difference in total energy of 0.123861d-3 Ha,
     for a difference of volume of 0.07605 Bohr**3 leading
     to a stress estimation of 1.62867d-3 Ha/Bohr**3
     
 51. O2 (nsppol=2, non-linear XC core correction, GGA)    
     Similar to test50, except use iscf=6 .

 52. Test the symmetry finder for all the Bravais lattices,
     with different input formats (rprim or angdeg), and for
     non-conventional choices of axes as well.
     Uses only one atom, placed at (0 0 0)

 53. Test the symmetry finder for all the Bravais lattices,
     with different input formats (rprim or angdeg), and for
     non-conventional choices of axes as well.
     Uses two different atoms, placed at (0 0 0) and (0.1 0 0), thus
     breaking many symmetries.

 54. Test the symmetry finder for all the Bravais lattices,
     with different input formats (rprim or angdeg), and for
     non-conventional choices of axes as well.
     Uses two atoms of the same type, placed at (0 0 0) and (0.1 0 0), thus
     breaking many symmetries, while the inversion needs a non-symmorphic
     translation.

 55. Test different additional features of the symmetry finder :
     - handling non-primitive cells
     - handling glide planes
     - handling screw axes

 56. Test a FHI pseudopotential for Chromium, with non-linear XC 
     core correction. The ecut is too low, the box is too small.
     Use metallic occupation numbers.
     The convergence is not enough to make finite-difference of energy.

 57. Test the calculation of frequency dependent susceptibility matrix. For a 
     He atom in a box, calculate the LDA groundstate and then the Kohn-Sham 
     susceptibility. From that compute the interacting susceptibility
     within the random phase approximation (RPA). Output the polarizability, 
     and certain traces over the susceptibility matrix that contribute
     to the correlation energy. For the RPA, the Coulomb interaction is
     cutoff at half the maximum box dimension.

 58. Test the mechanism for checking the presence of vacuum.
     Use H2 molecule, placed in a big box.

 59. Space group number 225 : FCC Aluminum, in both conventional and
     primitive unit cells.

 60. Space group number 139 : elongated aluminum, BCT

 61. 2D hexagonal   lattice : test the sets of k points generated automatically

 62. 2D square      lattice : test the sets of k points generated automatically

 63. 2D 4 point symmetry    : test the sets of k points generated automatically

 64. 2D rectangular lattice : test the sets of k points generated automatically

 65. 2D centered    lattice : test the sets of k points generated automatically

 66. 2D oblique     lattice : test the sets of k points generated automatically

 67. 3D cP lattice : test the sets of k points generated automatically
 
 68. 3D cF lattice : test the sets of k points generated automatically
 
 69. 3D cI lattice : test the sets of k points generated automatically
 
 70. 3D tP lattice : test the sets of k points generated automatically
 
 71. 3D tI lattice : test the sets of k points generated automatically
 
 72. 3D hR lattice : test the sets of k points generated automatically
 
 73. 3D hP lattice : test the sets of k points generated automatically
 
 74. Tin in the diamond structure, with 2 special points
     Test the effect of spin-orbit coupling, especially the
     electronic structure.
     Degeneracies at Gamma are correct ! 
     (note that ecut and nkpt are too low for quantitative accuracy)
     Without spin-orbit (in eV), dataset 4:
      -8.82682  1.26955 (x3) 1.30262 3.45296 (x3)
     With spin-orbit (in eV), dataset 6:
      -8.82682  0.75225      1.30262 1.50965 (x2) 3.07128 3.62556 (x2)

 75. Bismuth, treated as a metal, with 2 special points
     Test the effect of spin-orbit coupling, especially the forces.
     The dataset 1 to 3 are related to the nspinor=1 case. The forces
     are computed from finite differences of energy :
     abs(etotal3-etotal1)=0.00090461 Ha
     abs(xcart3-xcart1)=0.089970092 (taking into account both atoms)
     Their ratio is 0.01005456 Ha/Bohr, to be compared with
     fcart2=1.0049850156E-02 Ha/Bohr 
     The dataset 4 checks the case nspinor=2, pspso=1 (no spin-orbit yet)
     The dataset 5 to 7 try to reproduce the comparison of forces
     with spin-orbit :
     abs(etotal7-etotal5)=0.000445155 Ha
     abs(xcart7-xcart5)=0.089970092 (taking into account both atoms)
     Their ratio is 3.83633E-03 Ha/Bohr ...
     A better finite-difference scheme incorporating also data from 
     half-displacement, lead to 3.81832E-03 Ha/Bohr.
     This is in excellent agreement with
     fcart6=3.8183150850E-03 Ha/Bohr

 76. Silicon, diamond structure.
     Test the ability to read an input wavefunction and to generate
     from it any other wavefunction, changing different parameters.
     One of the _DEN file is also used as starting point of
     the cut3D tests.

 77. Cut3d code. Silicon, diamond structure.
     Using the unformatted density file generated in test #76,
     compute the density along the diagonal of the primitive cell,
     accross more than one cell. Also generate a formatted
     density file, to be read in the next run.

 78. Cut3d code. Silicon, diamond structure.
     Using the formatted density file generated in test #77,
     compute the density along the diagonal of the primitive cell,
     accross more than one cell. Two other files are used :
     a "cut.in" file, and a "xyz" file.

 79. Cut3d code. Silicon, diamond structure.
     Using the unformatted density file generated in test #76,
     compute the density in a 2D plane,
     accross more than one cell.

 80. Cut3d code. Silicon, diamond structure.
     Using the unformatted density file generated in test #76,
     compute the density at two different points.

 81. Nitrogen atom in a big box, spin-polarized.
     Compute Berry phase for atom at different positions.

 82. GaAs, zinc-blende structure.
     Compute Berry phase along 0 0 1, using a 4x4x10 sampling.

 83. Ni, FCC structure.
     Ferromagnetic phase, compute the magnetic moment in GGA.
     Check the convergence, and make a restart.
     Note that the pseudopotential is NOT a GGA psp.

 84. Ta, single atom in a box (1k point). Contributed by FJollet.
     HGH pseudopotential
     Ecut and acell are NOT realistic. One would need 20 Ha
     and a 20x20x20 box, while here we have 5 Ha and a 12x12x12 box.
     The small box size changes the degeneracy of levels.
     The all-electron values derived from another code, for comparison, are
     level  degeneracy       energy (Ha)  spin-orbit splitting
     6s      2               -0.194573
     5d3/2   4               -0.141779
     5d5/2   6               -0.119933       0.0218
     6p1/2   2               -0.050542
     6p3/2   4               -0.031281       0.0193

 85. Ta, single atom in a box (1k point). Contributed by FJollet.
     Same test as t84, but with a different pseudopotential.
     HGH semi-core pseudopotential (so more bands than test 84)
     Ecut (5Ha) and acell (12 Bohr) are NOT realistic.
     The all-electron values derived from another code, for comparison, are
     level  degeneracy       energy (Ha)  spin-orbit splitting
     5s      2               -2.673078
     5p1/2   2               -1.676933
     5p3/2   4               -1.352138       0.3248
     6s      2               -0.194573
     5d3/2   4               -0.141779
     5d5/2   6               -0.119933       0.0218
     6p1/2   2               -0.050542
     6p3/2   4               -0.031281       0.0193


 86. Ta, single atom in a box (1k point). Contributed by FJollet.
     Same test as t84, but with a different pseudopotential.
     Troullier pseudopotential. Same number of bands as test 84.
     Ecut (5Ha) and acell (12 Bohr) are NOT realistic.

 
 87. Ge liquid. Test of Nose dynamics. 2 atoms in a cell.
     Allows 4 time steps. Contributed by JY Raty.


 88. Ge liquid. Test of Langevin dynamics. 2 atoms in a cell.
     Allows 4 time steps. Contributed by JY Raty.


 90. H, compressed simple cubic, to test accurately RF in the metallic case. 
     No linear XC core correction.
     The total energy of the 1-atom cell is -.559361014386 Ha
     The 2DTE with respect to a q(0 0 0) phonon is 4.6E-10 .
     The 2DTE with respect to a transverse q(1/2 0 0) phonon
     is 1.75340771040435E-02. The 2DTE with respect to a transverse
     q(1/4 0 0) phonon is 9.38555654208462E-03.
 
 91. H, compressed simple cubic, frozen-phonon calculations corresponding
     to test 90. No linear XC core correction.
     The total energy of the quadruple cell is -2.237444057546 Ha,
     in excellent agreement with the result of test 90.
     The frozen-phonon (from forces) calculation of the 2DTE for the q(1/2 0 0)
     case gives 1.7535588d-2 . A better finite-difference scheme, using
     a doubled displacement to cancel the finite-difference error, gives
     1.7534196d-2. 
     The frozen-phonon (from forces) calculation of the 2DTE for the q(1/4 0 0)
     case gives 9.386061d-3 . A better finite-difference scheme, using
     a half displacement to cancel the finite-difference error, gives
     9.385565d-3 .


 92. Li, simple cubic, to test accurately RF in the metallic case.
     With non-linear XC core correction.
     The total energy of the 1-atom cell is -.29350020247929 Ha
     The 2DTE with respect to a q(0 0 0) phonon is 1.557063518d-7 .
     The 2DTE with respect to a transverse q(1/2 0 0) phonon
     is -0.1426599618 Ha. The 2DTE with respect to a transverse
     q(1/4 0 0) phonon is -0.6978404630E-01 Ha.


 93. Li, simple cubic, frozen-phonon calculations corresponding
     to test 92.
     The total energy of the quadruple cell is -1.1740008099174 Ha,
     in excellent agreement with the result of test 92.
     The frozen-phonon (from forces) calculation of the 2DTE for the q(1/2 0 0)
     case gives -0.142648808 Ha . A better finite-difference scheme, using
     a doubled displacement to cancel the finite-difference error, gives
     -0.142653130 Ha . This is a bit better, but not sufficiently.
     The frozen-phonon (from forces) calculation of the 2DTE for the q(1/4 0 0)
     case gives -6.9781822d-2 Ha. A better finite-difference scheme, using
     a half displacement to cancel the finite-difference error, gives
     -6.9783661d-2 Ha. 


 94. Ni, simple cubic, with a 4x4x4 grid, many bands test RF in the metallic case.
     With non-linear XC core correction. q(1/2 0 0)
     The total energy of the 1-atom cell is -41.644007152348 Ha .
     The 2DTE with respect to a longitudinal q(1/2 0 0) phonon is 1.90867182 Ha .
    

 95. Ni, supercell of simple cubic, corresponding to test 94.
     The total energy of the double cell is -83.288014304694 Ha, 
     in excellent agreement with the result of test 94.
     The frozen-phonon (from energies) calculation of the 2DTE for the q(1/2 0 0)
     case gives 1.903623 Ha . A better finite-difference scheme, using
     a doubled displacement to cancel the finite-difference error, gives
     1.908658 Ha .
     The frozen-phonon (from forces) calculation of the 2DTE gives
     1.8985 Ha. A better finite-difference scheme, using
     a doubled displacement to cancel the finite-difference error, gives
     1.9086 Ha . The agreement is excellent.

 96. Li, simple cubic, to test accurately RF in the metallic case.
     With non-linear XC core correction and non-zero ecutsm and dilatmx.
     Similar to test 92, except ecutsm and dilatmx, and no q(1/4 0 0).
     The total energy of the 1-atom cell is -.29337493523588 Ha
     The 2DTE with respect to a q(0 0 0) phonon is 5.5619d-8 .
     The 2DTE with respect to a transverse q(1/2 0 0) phonon
     is -0.12068188927 Ha. 


 97. Li, simple cubic, frozen-phonon calculations corresponding
     to test 96 (with non-zero ecutsm and dilatmx).
     The total energy of the double cell is -0.58674987047187 Ha,
     in excellent agreement with the result of test 96.
     The frozen-phonon (from energy) calculation of the 2DTE for the q(1/2 0 0)
     case gives -0.1206807 Ha .
     The frozen-phonon (from forces) calculation of the 2DTE for the q(1/2 0 0)
     case gives -0.1206794 Ha . Both are in rather good agreement with
     the result of test 96.

 98. Generate first-order responses for GaAs (zinc-blende).
     Compute responses at Gamma, in the presence of a non-zero ecutsm and dilatmx.
     1) ground state calculation, with a k-point sampling in the irreducible
       Brillouin zone only;
     2) computation of the ddk response, in preparation to the electric field
       response ;
     3) computation of the dynamical matrix at Gamma, as well as the
       Born effective charges, and the dielectric tensor ;

 99. Nb BCC along the Gamma-H direction
     Remnant of the bug search of winter 2000-2001. 

</pre>
