FW: [Opium-talk] about opium pps

[Wu R. <g02...@nu...>]

Dear all,

I used the attached param files to generate pps, but failed and log =
files gives out a lot of warnings. Would you please check for me?
I am still not sure when an unfilled valence state is needed. In the =
case of O,S and Se, they are usually in the -2 charged state. So an =
unfilled valence state necessary or unnecessary? In fhi98pp, an unfilled =
valence state, just higher then the highest occupied valence state, is =
included.

Regards

Rongqin=20

-----Original Message-----
From: Eric J. Walter [mailto:ew...@pa...]=20
Sent: 2004=C4=EA8=D4=C23=C8=D5 0:38
To: Wu Rongqin
Subject: Re: [Opium-talk] about opium pps


Dear Rongqin,

No, just order the valence states in s,p,d order, like I did.

I put 0.5 electron into 5s because this state caused many problems when =
it
had 1 electron.  I think that the transferability tests show that it is =
fine.
You may want to try other states if you are curious.

I have located and fixed the bug related to the first problem.  I must
now run some tests.  If all goes well, I will post v1.0.2 tomorrow.

Have a good day,

Eric





On Tue, Aug 03, 2004 at 12:27:03AM +0800, Wu Rongqin wrote:
> Dear Dr Eric,
> Thank you very much.
> The first problem: must I order the orbital according to their =
energies in the current version?
> The second problem: you only put 0.5 electron in 5s state. Will this =
pp act the same as 5s1?=20
> I am going to do more test using opium.
> Thanks
> Rongqin=20
>=20
> -----Original Message-----
> From: Eric J. Walter [mailto:ew...@pa...]=20
> Sent: 2004??8??2?? 23:07
> To: Wu Rongqin
> Subject: Re: [Opium-talk] about opium pps
>=20
> Dear Ronqin,
>=20
> Thanks for sending me the ag.param file, it pointed out a bug that I =
will fix now. =20
>=20
> The first problem you were having was related to keeping your valence
> orbitals in s,p,d order.  There used to be some parts of Opium which
> relied on this, I thought I removed them.  Our interaction has shown
> me one more place where this needs to be fixed.
>=20
> Even if, you had your valence orbitals in s,p,d order, Ag is still a
> hard one to make.  I was able to get a pretty good potential though.
> I have attached the .param file.
>=20
> Thanks, let me know how it performs.
>=20
> Eric
>=20
> PS: Send me any other input, output files you are having a problem =
with and I will
> look at them as soon as I can.
>=20
>=20
>=20
>=20
>=20
> On Mon, Aug 02, 2004 at 10:08:54PM +0800, Wu Rongqin wrote:
> > Dear Dr Eric,
> >=20
> > I have difficulties in generating the Ag pseudopotential, not able =
to remove the ghost states. I wonder whether you can send the ag.param =
file for me to generate pp? the outmost 11 electrons are taken as =
valence electrons.
> > Attached is the param file I used to generate ag pp. just check =
parameters for me.
> > I want lda pp
> > Thanks=20
> > Rongqin=20
> >=20
> > -----Original Message-----
> > From: Eric J. Walter [mailto:ew...@pa...]=20
> > Sent: 2004??7??31?? 2:32
> > To: Wu Rongqin
> > Subject: Re: [Opium-talk] about opium pps
> >=20
> >=20
> > Dear Rongqin,
> >=20
> > I am not in a position to generate potentials for you.  =20
> >=20
> > I would look at the literature and see what is typical for an LDA
> > pseudopotential before you go any further.
> >=20
> > Why don't you give a look at the cu.param file in the tests
> > directory (you can download the tests at sourceforge).  Using =
running
> > opium to generate the .fhi file and then using this in Opium, I get
> > very good results for the lattice constant of Cu.  I have attached =
my
> > results as a pdf file.  I think you can see from this, it is =
possible=20
> > to get satisfactory results.
> >=20
> > What do you get and could you send me the log and rpt file?
> >=20
> > Sincerely,=20
> >=20
> > Eric
> >=20
> >=20
> >=20
> >=20
> >=20
> >=20
> > On Sat, Jul 31, 2004 at 12:44:04AM +0800, Wu Rongqin wrote:
> > > Dear Dr Eric,
> > > I almost lost my patience after playing around the opium package. =
Everything seems not so complicated but I just cannot get the =
satisfactory results.=20
> > > I have tested the ps with good convergence conditions, ecut and =
kpoints.
> > > Currently I want to do calculations on a serial material. They are =
made of copper, silver, sulfur, selenium and tellurium. I wonder whether =
you can generate them for me, or I have to send you many email and you =
have to spend a lot time answering them?
>=20
> > > Thank you for your patience for answering my questions.
> > >=20
> > > Rongqin Wu=20
> > >=20
> > >=20
> > > -----Original Message-----
> > > From: Eric J. Walter [mailto:ew...@pa...]=20
> > > Sent: 2004??7??30?? 20:56
> > > To: Wu Rongqin
> > > Subject: Re: [Opium-talk] about opium pps
> > >=20
> > > Could you get correct lattice constants with the fhi98pp =
potentials?
> > > How do your results compare with other norm-conserving psps?
> > > Are you runnning at the right ecut?  Are you converged in kpts?=20
> > > Is you transferability reasonable? =20
> > >=20
> > >=20
> > >=20
> > > Eric
> > >=20
> > >=20
> > >=20
> > >=20
> > >=20
> > > On Fri, Jul 30, 2004 at 11:12:51AM +0800, Wu Rongqin wrote:
> > > > Dear all,
> > > > I generated some pps from opium and did some tests. As I found =
that the
> > > > pps cannot produce proper lattice constants. Have you ever =
experienced
> > > > such a problem?
> > > > How to solve this problem then? I am aware of that lda =
underestimates
> > > > the lattice constant while gga go to the contrary.
> > > > =20
> > > > Rongqn wu=20
>=20
>=20