[QE-users] Comparison of k_points for two different inputs of BCC structure
KRISHNENDU MUKHERJEE
krishnendu at nmlindia.org
Mon Oct 31 11:24:56 CET 2022
Hello Xin,
Thank you for your reply. I have made a new thread. Below is the k-points obtained from kpoints.x for BCC Bravais Lattice, for a 4 4 4 0 0 0 mesh, considering the 48 symmetries. I hope this will be the same for the case with calculations for ibrav=3, with BCC primitive cell with one atom per cell.
8 (=number of k_points)
1 0.0000000 0.0000000 0.0000000 1.00
2 0.2500000 0.0000000 0.2500000 12.00
3 0.5000000 0.0000000 0.5000000 6.00
4 0.2500000 0.2500000 0.5000000 24.00
5 0.0000000 0.0000000 0.5000000 6.00
6 0.2500000 0.0000000 0.7500000 12.00
7 -0.5000000 0.5000000 0.5000000 2.00
8 0.0000000 0.0000000 1.0000000 1.00
Now let us compare with the k_points that were used in your calculation when a BCC supercell consisting of BCC Unit Cells is used as input with ibrav=0. So, I request you if you can kindly post the k_points considered in your calculation. Also you may kindly include your input script once more so that later on somebody can follow the discussion.
Thanking you,
Best regards,
Krishnendu
From: "xin tlg jin" <xin.tlg.jin at outlook.com>
To: users at lists.quantum-espresso.org
Cc: "KRISHNENDU MUKHERJEE" <krishnendu at nmlindia.org>
Sent: Monday, October 31, 2022 12:28:24 PM
Subject: Re: [QE-users] Parallel computing of QE7.1 vc-relax crashes when using large number of processors
Hello Krishnendu,
Thank you for the message
Yes, we can start a separate thread. I didn't realize that there could be such a problem.
Actually, the reason that I use ibrav=0 for BCC W is that I want to put some interstitials inside if the program works.
Best regards,
Xin
On 30/10/2022 05:35, KRISHNENDU MUKHERJEE wrote:
Dear Xin Jin,
Sorry that I have another matter to discuss. If you wish we may start a separate thread on the subject I am eager to discuss. I have some concern about your script. If you want to do calculation for BCC W you may need to use ibrav=3. Note that for ibrav=3 the (primitive) lattice vectors are in the form:
ibrav=3 cubic I (bcc)
v1 = (a/2)(1,1,1), v2 = (a/2)(-1,1,1), v3 = (a/2)(-1,-1,1) which is inbuilt in QE (so that you need not input CELL_PARAMETERS) and you only need to put an atom in the position 0.00 0.00 0.00.
With those input QE determines the k_points for BCC Bravais Lattice.
You have input the atomic positions in terms of supercell made of BCC unit cell. And you have put ibrav=0. With those input I am afraid most probably the k_points generated would be that of a Simple Cubic structure.
What we can do is next week I would generate the k_points for BCC with a mesh of 4 4 4 0 0 0 and post it. And you can see whether it is matching with the k_points considered in your calculation and discuss further.
Thank you,
Best regards,
Krishnendu
------------------------------------------------------------------------------------------------------------------------------------------
Xin Jin wrote on 28/Oct/2022
Dear Quantum Espresso Forum,
I encountered a problem related to the parallel computing using QE7.1
for vc-relax.
I was trying to perform a vc-relax for a 3*3*3 BCC tungsten super cell.
The code works fine for non-parallel computing, also works fine for
parallel computing if the number of processors is smaller than 10.
However, if the number of processors is larger than 10, I will get
following MPI error:
/*** An error occurred in MPI_Comm_free//
//*** reported by process [3585895498,2]//
//*** on communicator MPI_COMM_WORLD//
//*** MPI_ERR_COMM: invalid communicator//
//*** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,//
//*** and potentially your MPI job)/
For parallel computing, I am using /OpenMPI/3.1.4-gcccuda/. (In
addition, it seems that If I use OpenMPI V4, the simulation speed will
be much slower than that of V3.)
Another thing is that, if I decrease the size of the supper cell, for
example to 2*2*2, then there is no problem in the parallel computing
even if I use more than 30 processors.
Could you help me look at this problem, please?
The input for QE can be found below.
Thank you in advance!
Xin Jin
/&control//
//
// calculation='vc-relax' //
// restart_mode='from_scratch', //
// prefix='W_relax', //
// pseudo_dir="../../PP_files",//
// outdir='./'//
//
// ///
////
//
// &system//
// ibrav= 0, //
// celldm(1)=5.972,//
// nat= 54, //
// ntyp= 1,//
// ecutwfc = 50,//
// ecutrho = 500,//
// occupations='smearing', smearing='mp', degauss=0.06//
// ///
//
// &electrons//
// diagonalization='david',//
// conv_thr = 1.0d-8,//
// mixing_beta = 0.5,//
// ///
////
// &ions//
// ///
//
// &cell//
// press = 0.0,//
// ///
////
//ATOMIC_SPECIES//
// W 183.84 W.pbe-spn-kjpaw_psl.1.0.0.UPF//
////
//CELL_PARAMETERS {alat}//
// 3.0 0.0 0.0//
// 0.0 3.0 0.0//
// 0.0 0.0 3.0 //
////
//ATOMIC_POSITIONS {alat}//
//W 0.00000 0.00000 0.00000//
//W 0.50000 0.50000 0.50000//
//W 1.00000 0.00000 0.00000//
//W 1.50000 0.50000 0.50000//
//W 2.00000 0.00000 0.00000//
//W 2.50000 0.50000 0.50000//
//W 0.00000 1.00000 0.00000//
//W 0.50000 1.50000 0.50000//
//W 1.00000 1.00000 0.00000//
//W 1.50000 1.50000 0.50000//
//W 2.00000 1.00000 0.00000//
//W 2.50000 1.50000 0.50000//
//W 0.00000 2.00000 0.00000//
//W 0.50000 2.50000 0.50000//
//W 1.00000 2.00000 0.00000//
//W 1.50000 2.50000 0.50000//
//W 2.00000 2.00000 0.00000//
//W 2.50000 2.50000 0.50000//
//W 0.00000 0.00000 1.00000//
//W 0.50000 0.50000 1.50000//
//W 1.00000 0.00000 1.00000//
//W 1.50000 0.50000 1.50000//
//W 2.00000 0.00000 1.00000//
//W 2.50000 0.50000 1.50000//
//W 0.00000 1.00000 1.00000//
//W 0.50000 1.50000 1.50000//
//W 1.00000 1.00000 1.00000//
//W 1.50000 1.50000 1.50000//
//W 2.00000 1.00000 1.00000//
//W 2.50000 1.50000 1.50000//
//W 0.00000 2.00000 1.00000//
//W 0.50000 2.50000 1.50000//
//W 1.00000 2.00000 1.00000//
//W 1.50000 2.50000 1.50000//
//W 2.00000 2.00000 1.00000//
//W 2.50000 2.50000 1.50000//
//W 0.00000 0.00000 2.00000//
//W 0.50000 0.50000 2.50000//
//W 1.00000 0.00000 2.00000//
//W 1.50000 0.50000 2.50000//
//W 2.00000 0.00000 2.00000//
//W 2.50000 0.50000 2.50000//
//W 0.00000 1.00000 2.00000//
//W 0.50000 1.50000 2.50000//
//W 1.00000 1.00000 2.00000//
//W 1.50000 1.50000 2.50000//
//W 2.00000 1.00000 2.00000//
//W 2.50000 1.50000 2.50000//
//W 0.00000 2.00000 2.00000//
//W 0.50000 2.50000 2.50000//
//W 1.00000 2.00000 2.00000//
//W 1.50000 2.50000 2.50000//
//W 2.00000 2.00000 2.00000//
//W 2.50000 2.50000 2.50000//
//
//K_POINTS {automatic}//
//4 4 4 0 0 0//
/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: [ http://lists.quantum-espresso.org/pipermail/users/attachments/20221028/87b8c648/attachment.html | <http://lists.quantum-espresso.org/pipermail/users/attachments/20221028/87b8c648/attachment.html> ]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221031/31e10298/attachment.html>
More information about the users
mailing list