Please try to disabled the Diag.DivideAndConquer option, which is ture by
default, and run the job again from scratch.
Cheers.
/Guangping Zhang
2014-11-08
发件人:邵德喜
发送时间:2014-11-08 04:31
主题:Re: [SIESTA-L] Error: Failure to converge standard eigenproblem
收件人:"siesta-l"
抄送:
I have tested w
Dear Pu Zhang,
In addtion to Fei, you can specify the same size basis set for all the atoms in
the calculation, or you can used a different size for different atom species
(or even more for the same atom species you can specify a different basis set.
e.g. atoms at surface should have a larger b
Dear siesta users:
I have submitted a job using 18 cores to a computing cluster. However, after
500 SCFs it failed to converge. Therefore, I resubmitted this same job from
scratch also using 18 cores. This time, it only use 16 SCFs to converge. I can
not understand why this happen. I can not gu
This error is expected since you use "Diag.ParallelOverK T" in your *fdf
input file.
Try to use "Diag.ParallelOverK F", I think it will be OK for 16 ,32 and more
cores.
PS:
Diag.ParallelOverK (logical): For the diagonalisation there is a choice in
strategy about
whether to parallelise ove
:"zgp121"
发送时间:2014-08-13 17:37
主题:Re: [SIESTA-L] can not constrain atom using constr.f (with attched constr.f)
收件人:"siesta-l"
抄送:
Dear Nick,
Thanks very much for your suggestion.
I will try it.
With best regards.
Guangping
2014-08-13
发件人:Nick Papior Andersen
发送时间:20
s.
2014-08-13 8:19 GMT+00:00 zgp121 :
Dear all,
I wan to do molecular dynamics for part of my system. So I use constr.f to
constrain some of the atoms. As attached my constr.f, during the molecular
dynamics, I want only my first 10 and last 10 atoms rigidly move along z axis.
So, ther
Dear all,
I wan to do molecular dynamics for part of my system. So I use constr.f to
constrain some of the atoms. As attached my constr.f, during the molecular
dynamics, I want only my first 10 and last 10 atoms rigidly move along z axis.
So, there only two dimensions unstricted for the 20 atom
Dear all,
I wan to do molecular dynamics for part of my system. So I use constr.f to
constrain some of the atoms. As attached my constr.f, during the molecular
dynamics, I want only my first 10 and last 10 atoms rigidly move along z axis.
So, there only two dimensions unstricted for the 20 atom
nt (0 given)
On 25-Jul-2014, at 8:07 pm, "zgp121" wrote:
Please give more information on how you use the EigenChannels commands. That
would be much of useful. Otherwiese, nothing can be deduced.
2014-07-25
发件人:Akshu
发送时间:2014-07-25 21:03
主题:[SIESTA-L] EigenChannels proble
Please give more information on how you use the EigenChannels commands. That
would be much of useful. Otherwiese, nothing can be deduced.
2014-07-25
发件人:Akshu
发送时间:2014-07-25 21:03
主题:[SIESTA-L] EigenChannels problem in Inelastica
收件人:"siesta-l@uam.es"
抄送:
Dear siesta users
When I try t
Your input file is OK for my trunk 444.
2014-07-22
发件人:Sergio Cruz
发送时间:2014-07-21 22:54
主题:[SIESTA-L] ERROR PARSE module Siesta trunk
收件人:"siesta-l"
抄送:
Dear:
I am performing some simulations of the DNA chains with Siesta trunk 433, and
using vdW, but I am getting an error in all simul
Figure 3a is the dependence of projected DOS on E-Ef.
I never get figures like figure 3b. Figure 3c can be obtained from inelastica.
Cheers.
Guangping
2014-06-30
发件人:Jingxian Yu
发送时间:2014-06-30 11:08
主题:Re: [SIESTA-L] Non equilibrium density of states in Transiesta
收件人:"zgp121&quo
f us. Could you please explain its
outputi got one eigen value and remaining all zeroes.
Regards
Akshu
On Saturday, June 28, 2014 6:42 PM, zgp121 wrote:
Yes, it is possible. But in TranSIESTA itself not. You can achive this in
inelastica package. But if you are only interested in the
the code with all of us. Could you please explain its
outputi got one eigen value and remaining all zeroes.
Regards
Akshu
On Saturday, June 28, 2014 6:42 PM, zgp121 wrote:
Yes, it is possible. But in TranSIESTA itself not. You can achive this in
inelastica package. But if you are on
抄送:
Hi,
Is it possible to calculate homo and lumo (mpsh eigen states) of the molecule
sandwiched between two
Electrodes using transiesta ??
On 28-Jun-2014, at 5:39 pm, "zgp121" wrote:
Dear Jin,
Projected density of states at nonequilibrium (under non zero bias) in
TranSIESTA sho
oment
we are experiencing some problems with basis sets/PPs. Once we overcome these
problems, we will test your suggestion and let you know the outcome.
Many thanks again.
Best regards, Vaida
From: siesta-l-requ...@uam.es [siesta-l-requ...@uam.es] on behalf of zgp121
[zgp...@126.com]
Sent: Thu
Hi, Yuehua:
I guess you used different number of cores to run these two jobs. If it is
true, I think the coverge of your task does not arrive at the real converge or
converge to different electron structures. You can see that the enery of
Ion-ion is exactly the same. So the compling of SIESTA c
If you do not want your system to be relaxed, and MD.TypeOfRun is CG, of course
the MD.NumCGsteps should be 0. You can have a try for a small system to verify
your guess.
2014-06-05
发件人:xu yuehua
发送时间:2014-06-05 23:01
主题:[SIESTA-L] the MD.TypeRun choice for the total energy calculation
收
Hi, Vaida:
First, I am not familar with the LDA+U method and the use in SIESTA. But I
think if you are right, you still can achieve what you want to do. Maybe the
following hint is useful.
%block Chemical_Species_Label
1 26 Fe
2 26 Fe_1
3 26 Fe_2
4 26 Fe_3
5 26 Fe_4
6
Dear all,
I suddenly got the answer why the job can run with 8 cores but not 16 cores.
The optical part calculation is very memory consuming. It can not work due to
lack of memory.
With best regards.
Guangping
2014-03-29
发件人:"zgp121"
发送时间:2014-03-29 00:57
主题:[SIESTA-L
Dear siesta users,
I find there is one problem on the parallel calcuation of optical properties.
When I use 16 cores, the calcuation stops when arriving the optical part. But
it can run normally when I use 8 cores for the same job. So, I wonder what is
the parallel based on during the optical c
od' opened at (1) is not a GFORTRAN module file
make: *** [iogrid_netcdf.o] Error 1
Do you know the problem?
On Thu, Mar 27, 2014 at 9:10 PM, zgp121 wrote:
That is enough. Do not forget to add -DCDF to FPPFLAGS and aslo the right path
for NETCDF_LIBS and NETCDF_INCLUDE.
Good luc
lease tell me what have you included in the arch.make
file to get the support. I followed like in this:
http://sourceforge.net/apps/mediawiki/inelastica/index.php?title=Installation_notes_%28Guangping_Zhang%29
Thanks again,
Behnaz
On Thu, Mar 27, 2014 at 8:40 PM, zgp121 wrote:
So, I think y
In addition, if your transiesta is a NetCDF supported one, you will see this at
the head of the output file.
PARALLEL version
TRANSIESTA support
NetCDF support
2014-03-28
发件人:"zgp121"
发送时间:2014-03-28 01:40
主题:Re: Re: [SIESTA-L] EigenChannels problem in Inelastica
收件人:"siest
ave NetCDF supported transiesta but it doesn't produce me ion.nc!
I don't know why?
On Thu, Mar 27, 2014 at 6:30 PM, zgp121 wrote:
You need the files named *ion.nc, where contains the basis for each element
used. The *ion.nc file can be generated by a NetCDF supported transiesta or
sies
You need the files named *ion.nc, where contains the basis for each element
used. The *ion.nc file can be generated by a NetCDF supported transiesta or
siesta executable.
Wish it helps.
2014-03-27
发件人:Behnaz Bagheri
发送时间:2014-03-27 23:16
主题:[SIESTA-L] EigenChannels problem in Inelastica
On Wed, Mar 26, 2014 at 3:00 PM, zgp121 wrote:
Dear all,
Can anyone tell me the relationship between the dielectric function and the
polarizability for the molecular system in SIESTA code?
I think the data stored in Systemlabel.EPSIMG is the frequency-dependent
polarizability not die
Dear all,
Can anyone tell me the relationship between the dielectric function and the
polarizability for the molecular system in SIESTA code?
I think the data stored in Systemlabel.EPSIMG is the frequency-dependent
polarizability not dielectric function. Or can someone give some formula used
i
at this is a very rough
estimate!
It is only the memory used by transiesta, whatever is also allocated in SIESTA
is ontop of this memory usage.
Kind regards Nick
2014/1/2 zgp121
By the way, can I estimate the memory use during my transiesta run before the
job runs. Recently, I h
LAPACK
against these versions.
2. Use the ACML library from AMD. Compile NetLIB ScaLAPACK and link against
ACML.
I would recommend you follow step 2 or 1a.
2. is easiest, then 1a and lastly 1b can be quite difficult.
Kind regards Nick
2014/1/2 zgp121
Dear Nick,
Thanks for your kindly
ACML.
I would recommend you follow step 2 or 1a.
2. is easiest, then 1a and lastly 1b can be quite difficult.
Kind regards Nick
2014/1/2 zgp121
Dear Nick,
Thanks for your kindly reply.
In both case, I compile the TranSIESTA using MKL and Intel compiler.
According what you mentione
available
optimizations for your CPU
cluster 2. looses in SCF due to lower clock-frequency
cluster 2. wins in TS routine as that relies heavily on the libraries used, and
if MKL is used this will most certainly use all available optimizations.
Kind regards Nick
2014/1/2 zgp121
Dear a
Dear all,
I have encountered a situation where the TranSIESTA performance is very hard to
understand.
I have two systems, which are very similar and have the same number of atoms
and basis orbitals. And I run the two systems on two different clusters using
the same number of cores, let's say c
file to initialize the SCF?
收件人:"siesta-l"
抄送:
Dear Guangping,
On Mon, 09 Dec 2013 12:10:48 +0100, zgp121 wrote:
Dear Nick,
I now see the effect of Rho.IN.grid.nc on the SCFs of SIESTA part. Indeed, this
would not accelerate the SIESTA SCF, in turn it would make more SCFs
e explicit notification about using it.
On Mon, 09 Dec 2013 11:49:18 +0100, zgp121 wrote:
Dear Nick,
You mean, in the SIESTA part for *.DM in TranSIESTA calculation, it can use
Rho.IN.grid.nc file? It just not give out the explicit information on the
standard output?
Thanks for your reply.
On Mon, 09 Dec 2013 10:36:42 +0100, zgp121 wrote:
Dear Nick,
I have tried it in TranSIESTA. If TranSIESTA does not find a *TSDE or *DM, then
it will use SIESTA method to generate a *.DM despite of the exist of a
Rho.IN.grid.nc file. That is what I have experienced in trunk-444. So, I think
t the code it does seem like it
should be possible. Why do you know it is not available in TranSIESTA?
Kind regards Nick
2013/12/6 zgp121
Dear TranSIESTA developers:
In SIESTA, one can use SCF.Read.Charge.NetCDF or
SCF.Read.Deformation.Charge.NetCDF options to use the Rho.IN.
Dear TranSIESTA developers:
In SIESTA, one can use SCF.Read.Charge.NetCDF or
SCF.Read.Deformation.Charge.NetCDF options to use the Rho.IN.grid.nc or
DeltaRho.IN.grid.nc file to initialize the SCF. This is very important for
increasing the basis size for the same systems. However, it is not avai
Dear siesta users:
I wonder to know, in the optical calculation in siesta, whether the dielectric
function has the contribution from transitions between different k values for
bulk systems? I guess it just calculates the transitions within the same k
value and averages it over the k points.
Is
ically you need not worry about this as the self-consistency takes care
of the potential drop in the cell.
Kind regards Nick
2013/11/20 zgp121
Dear Transiesta developers and users,
In Transiesta, how the possion equation is solved? I see in the code,
transiesta call the same subrouti
Dear Transiesta developers and users,
In Transiesta, how the possion equation is solved? I see in the code,
transiesta call the same subroutine to solve the possion eqution, where a
periodic boudary condition in three directions are used. Under a bias,
-V(z/Lz-0.5) is added by calling ts_voltag
Dear siesta users,
I want to set up the Hamiltonian of a system based on the spatial electron
density and then output the total potential and the corresponding wavefunction.
I see there are two subroutines on this, one is setup_H0 and the other is
setup_hamiltonian. What setup_H0 does is to con
e.
However, I would do the SCF from scratch, i.e. delete the DM and TSDE files,
then run transiesta.
Kind regards Nick
2013/11/5 zgp121
Oh, it is a bad news. You mean I should do the calcuation from the begining? I
should recalcuation the Green's function for the electrodes and do the
28
主题:Re: [SIESTA-L] Bias drop in the transiesta is incorrect
收件人:"siesta-l@uam.es"
抄送:
In principle yes, in practice no. siesta does not recognize the lattice vector
shift as such.
I would restart.
Kind regards Nick
2013/11/5 zgp121
Dear Nick,
Thanks for your reply. When I
s that the potential ramp is placed geometrically in the unit cell
as though the electrodes starts/ends at z=0, z=c.
Kind regards Nick
2013/11/3 zgp121
Dear transiesta users:
When I use the .VT for 1.0V minus the .VT for 0.0V, I should have 0.5 eV for
left electrode and -0.5eV for right
Dear Transiesta users:
Does the quantity 'electron mobility' make sense in molecular juntion. If does,
how can I obtain it from transiesta calcuation?
Guangping
2013-10-31
f new transiesta and tbtrans
收件人:"siesta-l@uam.es"
抄送:
Yes, sorry, you should do exactly the same edit for both m_ts_options and
m_tbt_options.
I have attached a patch for you to apply (that is easier).
Kind regards Nick
2013/10/29 zgp121
Dear Nick,
I think there are seveals t
do exactly the same edit for both m_ts_options and
m_tbt_options.
I have attached a patch for you to apply (that is easier).
Kind regards Nick
2013/10/29 zgp121
Dear Nick,
I think there are seveals to modified, 'm_ts_options.F90' and
'm_tbt_options.F90'.
can I j
exist=exist)
Right after that line add the following:
exist = exist .or. .not. IONode
Then it should only check the file existence for the master-node.
Kind regards Nick
2013/10/29 zgp121
Dear Nick,
Thanks for your reply.
The batach job sctript only make a copy one the master node. So i
to copy the
entire TSHS file.
Kind regards Nick
2013/10/29 zgp121
Dear Transiesta and TBtrans users:
I noticed that during the calculation of the Green's function for the leads,
the I/O nodes are not only the master node in trunk-444.
Becasue I want to parallel the executables on seve
Dear Transiesta and TBtrans users:
I noticed that during the calculation of the Green's function for the leads,
the I/O nodes are not only the master node in trunk-444.
Becasue I want to parallel the executables on several nodes, also I want to
improve the I/O performance, so I copy the files t
Hi,
I have the same transiesta executable on the same HPC. But using different
cores to run the same task to test the parallel efficiency. However, I found
when I use 1 core, it takes 11 scfs. However, 38 scfs for 12 cores. Therefore,
resulting a much longer time for 12 cores. Moreover, when I
Hi all,
Can I repeat the electric dipole using the classic description? That is I want
to generate the total charge in cube format. And using d=sum(q*r) at each mesh
grid. But the value is not equal to the result from SIESTA output.
Is there something wrong?
With best regards.
Guangping Zhang
uld be portable, it
did work by having lower case on SUSE right?
Kind regards Nick
2013/10/10 zgp121
Dear Nick,
Thanks for your advices.
Yes. I am sure. On the suse cluster, it did work for "Variables". When I
transfer this fdf file to Ubuntu, this file does not work. I t
try and goto line 1055 of file Src/m_ts_electrode.F90 and add
this (right before the "call memory" line):
print*,allocated(isa),isa
and recompile/run and return the output to me?
Kind regards Nick
2013/10/10 zgp121
Dear SIESTA users:
When I use 24 cores to run the TranSIESTA calcu
Dear SIESTA users:
When I use 24 cores to run the TranSIESTA calculation using trunk 428, I got
the error like this:
/var/spool/slurmd/job1321828/slurm_script: line 12: 19218 Bus error mpirun
transiesta < trans-0.0.fdf > trans-0.0.log
And I also find the following message in trans-0.0.log:
56 matches
Mail list logo