RSPt Site - The best RSPt in the world!

The best RSPt in the world!

Topics

Home

General News (14/0)

GeekLog (12/0)

User Functions

Don't have an account yet? Sign up as a New User
Lost your password?

Events

There are no upcoming events

Older Stories

Friday25-Apr

RSPt workshop in Sandia (0)

Tuesday01-Apr

Welcome to RSPt Site
Sunday, September 05 2010 @ 04:39 AM CEST

First | Previous | 1 2 | Next | Last

Home »

Use »

testsuite rspt920

testsuite rspt920

patrice

Registered: 10/20/07
Posts: 23

Thursday, May 06 2010 @ 10:39 AM CEST

After succesful compilation (see forum -install) with IBM (cineca) architecture, I went tru 2 tetst in the testsuite : Pr and bcc_Fe
The latter is ok.
For the former I got very different magnetic moments . I attach a list of comparison.

INFO : test has been done by hand (i.e. ./rspt in the Pr directory)

out.ref = reference_output
out = present calculation_output
MPI RSPT
rspt version number: rspt.6.0102221136

-------------------------TEST--------------------

TEST rspt920 CINECA , May 5, 2010 -

testsuite/Tests/Pr

out.ref
MPI RSPT
rspt version number: rspt.6.0102221136
MPI Task Configuration
Total MPI tasks: 0
FFT Configuration
single-threaded FFTW3

RSPT
OUTPUT for :
LABEL
time date
22.08 31 Mar 2010

Eigenvalue sum: -8.73226126248728
fermi energy = 4.2600356503220E-01
D(ef) = 1.0308699058037E+02
TIME: pdos0: 0.094m in, 0.002m added, 0.097m total

out
MPI RSPT
rspt version number: rspt.6.0102221136
MPI Task Configuration
Total MPI tasks: 0
FFT Configuration
single-threaded FFTW3

Readin: Please, set the env variableRSPT_SCRATCH to point to your scratch direc
tory.
RSPT
OUTPUT for :
LABEL
time date
16.5905 May 2010

Eigenvalue sum: -8.67030700075630
fermi energy = 4.4021483916294E-01
D(ef) = 1.1337754739590E+02
TIME: pdos0: .061m in, .002m added, .063m total

out.ref

Muffin tin charge from density matrix:
type spin Q
1 down .99422365536039
total down .99422365536039

1 up .00000000000000
total up .00000000000000

Energies from state 1
Full charge density :
Vhartree * nsic : 1.817816130
Vexc * nsic : -0.987742243
Eigenvalue shift : -0.830073888
Hartree energy : 0.908908065
XC energy : -0.911582196

out

Muffin tin charge from density matrix:
type spin Q
1 down 1.0005663706084
total down 1.0005663706084

1 up .00000000000000
total up .00000000000000

Energies from state 1
Full charge density :
Vhartree * nsic : 1.911989296
Vexc * nsic : -1.008940992
Eigenvalue shift : -0.903048304
Hartree energy : 0.955994648
XC energy : -0.952230698

out.ref

condng0:
Interstitial charge:
spin Q dQ/Q fix
down .628139324374855 -.000084575723761 1.00008458287741
up .574494217857561 -.000005441194917 1.00000544122452

out

condng0:
Interstitial charge:
spin Q dQ/Q fix
down .631194758660053 .000718316472096 .999282199136087
up .611779203791776 .000220339447641 .999779709091135

out.ref

4 -3 -7.487858134 6.000 1.225 0.0000 -0.000819
5 -1 -2.156653588 0.000 1.874 0.0046 0.040281
5 1 -0.991149576 0.000 2.289 0.0177 -0.071956
5 -2 -0.776214890 0.000 2.406 0.0263 -0.085458
core electrons: 46.0 fix = 1.00000000000000
spin down: 22.9999999999999
spin up: 22.9999999999999
mt: 45.9999894882164 leakage: .000010511783507
spin down: .000005255891754 22.9999947441082
spin up: .000005255891754 22.9999947441082

out

4 -3 -7.487858134 6.000 1.225 0.0000 -0.000819
5 -1 -2.156653588 0.000 1.874 0.0046 0.040281
5 1 -0.991149576 0.000 2.289 0.0177 -0.071956
5 -2 -0.776214890 0.000 2.406 0.0263 -0.085458
core electrons: 46.0 fix = 1.00000000000000
spin down: 22.9999999999999
spin up: 22.9999999999999
mt: 45.9999894882164 leakage: .000010511783522
spin down: .000005255891761 22.9999947441082
spin up: .000005255891761 22.9999947441082

out.ref

Magnetic moment ... -2.7366099 Bohr magnetons

potential constant r**2 * n0: true pseudo
.835028106 .239260483

out

Magnetic moment ... -1.9009698 Bohr magnetons

potential constant r**2 * n0: true pseudo
.838050418 .232417290

out.ref

nsph Hartree mt : 0.18292110546320E+01 1.829 211 054 6
nsph E_xc mt : -0.18511738926036E+01 -1.851 173 892 6
nsph E_sic mt : -0.21962837971656E-01 -0.021 962 837 9
Total Energy
E : -0.18472103636542E+05 -18,472.103 636 541
Other total energies
LDA/GGA energy : -0.18471960510708E+05 -18,471.960 510 708
Full-CD SIC E : -0.18471938547870E+05 -18,471.938 547 870

out

nsph Hartree mt : .19447403929440E+01 1.944 740 392 9
nsph E_xc mt : -.19473406151109E+01 -1.947 340 615 1
nsph E_sic mt : -.26002221668540E-02 -0.002 600 222 1
Total Energy
E : -.18472100361529E+05 -18,472.100 361 528
Other total energies
LDA/GGA energy : -.18471930416731E+05 -18,471.930 416 731
Full-CD SIC E : -.18471927816509E+05 -18,471.927 816 509

Torbjorn

Registered: 11/03/06
Posts: 74

Thursday, May 06 2010 @ 11:02 AM CEST

Sorry, my fault. I see that I have forgotten to change the out.ref file when the test changed. I will fix this.

I see that you got exactly the energy that is in tests.yaml, so your compilation is OK.

Torbj�rn

olleerik

Registered: 11/14/06
Posts: 8

Thursday, May 06 2010 @ 11:14 AM CEST

Hej !
Pr is tricky. If you have the 2 4f electrons as valence band states, you should get a spin-moment close to 1. The number of f-electrons should in addition be close to 2, inside the muffin-tin radius (look at 'projected charges'). For this calculation you need the 4f basis functions to expand you wavefunction.

If you have chosen to put the two 4f electrons in the core, you should have only 3 valence bands, and should not have 4f basis functions in the expansion of the wavefunction.

patrice

Registered: 10/20/07
Posts: 23

Thursday, May 06 2010 @ 11:24 AM CEST

Hi Torbjorn and Olle !

It's ok , don't worry Torbjorn , I figured out something like this.
Do you know of a manual to download including Section 8 ???

Thanks

Torbjorn

Registered: 11/03/06
Posts: 74

Thursday, May 06 2010 @ 12:31 PM CEST

The manual in r920 should have a chapter 8. Is it missing?

Torbj�rn

patrice

Registered: 10/20/07
Posts: 23

Thursday, May 06 2010 @ 01:57 PM CEST

Sorry for the silly question ! The manual is there.

I have an even more silly question : the file symt.inp for whatever compound MUST be re-written by hand starting from the one
in /documentation/inputs , as in the old full-manual versions ??

And : where is the input file for the exc atom to create atomdens ?
Thanks,
patrizia

Torbjorn

Registered: 11/03/06
Posts: 74

Thursday, May 06 2010 @ 02:28 PM CEST

1)
You can of course start from scratch, but the format is "semi-fixed" and must contain
1. the bravais lattice
2. the spin axis
3. the number of atoms
4. the atomic positions
The only thing that is optional is
5. a strain matrix

The example in documentations/inputs is chosen so that you can copy it for an hexagonal system. It has sufficiently many digits written out so that symt finds all the proper symmetries ("why don't I get all the symmetries for this hcp compound" .
But there is no way of automatically generating the symt.inp file, that is the one file you must supply yourself.

2)
The input files for the atom program are found in the automatically generated atom directory. They have names like data_[atom number].a. There you can change all settings for the atomic program.

Torbj�rn

patrice

Registered: 10/20/07
Posts: 23

Thursday, May 06 2010 @ 02:49 PM CEST

I still think in terms of 'old version' procedure...where everything had do be done by hand !!!

Thanks a lot

patrizia

patrice

Registered: 10/20/07
Posts: 23

Thursday, May 06 2010 @ 04:30 PM CEST

I have now almost all for fcc_Fe but the file 'data'.
I guess data, atomedens should be generated by the general RUNDIR/Makefile ,right ?

But this Makefile is not ok for xlc, right ?

SHELL=/bin/bash

include rspthome
COMPILER_IS_ON =
MATERIAL =
DTAFILES = dta/*
DATAFILES = dataForm length_scale strain_matrix

RUNSPATH = $(RSPTHOME)/runRspt
INCLUDES = -I$(RSPTHOME)/include
LIBPATHS = -L$(RSPTHOME)/lib -L./
LIBS = -ljxwac -lm
CC = icc
CFLAGS = -O2 -fno-gnu-keywords -g -traceback $(INCLUDES)

what:
echo make what?
........

So , icc must be changed to xlc ?
CCFLAGS to the ones in RSPTmake.inc ?

and
RSPTHOME sould be the directory of the compilation, right?

with this I should be 'FORMALLY' all set !!!

thanks again

Torbjorn

Registered: 11/03/06
Posts: 74

Thursday, May 06 2010 @ 04:42 PM CEST

[Quote I guess data, atomedens should be generated by the general RUNDIR/Makefile ,right ?

Not quite. atomdens is generated by going into the atom directory and type 'make'.

Actually, the flags etc in the Makefile mostly does not matter, since you don't use it to compile the code any more. I almost never touch the Makefile.

It is probably about time that Someone does Something about the Makefile, as there is a lot that is clearly outdated.

Torbj�rn