[QE-users] Convergence and restart of hp.x
Dear all, I am using hp.x to compute the U value. I have two types of Hubbard atoms in my unit cell. I am using qe-6.5. For my system with a 2x2x2 q-grid, the calculations converge. However, the moment I use a denser grid, the calculations don't converge, even after playing around with the mixing parameter. Can someone please indicate other input parameters that can be tuned to reach convergence? Also is there a way to restart the hp.x calculations? With regards, Prasenjit -- PRASENJIT GHOSH, IISER Pune, Dr. Homi Bhabha Road, Pashan Pune, Maharashtra 411008, India Phone: +91 (20) 2590 8203 Fax: +91 (20) 2589 9790 ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] Possible typo: ecutrho, & ! suggested cut-off for the change
Hi, In the following file: atomic/src/ld1inc.f90, the line 96 is as follows: ecutrho, & ! suggested cut-off for the change I'm afraid that is a typo in it, maybe it should be written as follows: ecutrho, & ! suggested cut-off for the charge Best wishes -- Hongyi Zhao ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Subject: RE: Slab doesn't converge
Dear Moraga Your feedback is very helpful . I have another questions :Is it fine to use the default value (1xe-03) for force-conv-thr ? If i use this default value , what is the suitable value conv-thr ? I’m familiar with this asymmetric slab, so How to use the dipole background You stated?In order to study various monolayer coverages , i have yo place multiple S atoms on the surface, ao is there any criteria for placing these atoms on asymmetric slab ? How far they should be from each other to avoid lateral interactions between Adsorbed atoms ? Sulfur (S⁻²) adsorption, should i set the net charge of the system to -1 ? If i set it to 0 (default) , does that mean Sulfur will be bonded to two Ni atoms ? The Ni2P is paramagnetic, so I didn’t consider spin polarized calculation, is this correct ? Sent from Yahoo Mail for iPhone On Sunday, August 16, 2020, 11:19 PM, Lucas Nicolás Lodeiro Moraga wrote: Hi! I am not an expert in adsorption, but I thought your input has some deficiencies. Firstly, forc_conv_thr is set for the maximum force for each relax-frame, the value that you use is too low, compared with your conv_thr (which define how accurate are your scf forces). Also, for relax, stress is not necessary and you could save some time neglecting it (I assume your slab is converged after for vc-relax). About the electronic parameters... 25 Ry for ecutwfc seems to me too low for USPPs. The fermi-dirac smearing is slower than others and the results do not change appreciably. Another thing is the vdW approach, you are using a vdW-functional plus the grimme's vdW-corrections... this does not seem correct. If you use a vdW-functional you are accounting the vdW interactions by the non-local correlation term, and the grimme's empirical correction is not needed, and vice versa, if you use the grimme's empirical corrections, you have to use a non vdw-functional as LDA,GGA or hybrid. A value of 0.7 for mixing_beta could be problematic for bad behaved systems. When I have problems on the first scf, I change this value to lower values like 0.5 or 0.3 to prevent long jumps in the scf solutions. Finally, something about the system-structure is weird to me. Your distances between S-P and S-Ni are too short, ~1.5 A. Those distances are too short to heavier atoms, and seems to me a bad guess for the first geometry (but I remember, I am not an expert in adsorptions, you know more about the system).Moreover, as you are putting the adsorbate only on one face of the system, you have a net dipole in the system. If you print the electrostatic potential profile along z, on vacuum space you will have a slope on the profile. It can be fixed, using a dipole background or using symmetrical adsorbates and surfaces on both sides of the slab. Also, you are simulating the adsorption of neutral atoms (in principle) of S. If you want to simulate Sulfur (S⁻²) adsorption, your system has to be a total negative charge, this requires some 2-D method to neglect the interactions between both sides of slab. Regards ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] Subject: RE: Slab doesn't converge
Dear Moraga Your feedback is very helpful . I have another questions :Is it fine to use the default value (1xe-03) for force-conv-thr ? If i use this default value , what is the suitable value conv-thr ? I’m familiar with this asymmetric slab, so How to use the dipole background You stated?In order to study various monolayer coverages , i have yo place multiple S atoms on the surface, ao is there any criteria for placing these atoms on asymmetric slab ? How far they should be from each other to avoid lateral interactions between Adsorbed atoms ? Sulfur (S⁻²) adsorption, should i set the net charge of the system to -1 ? If i set it to 0 (default) , does that mean Sulfur will be bonded to two Ni atoms ? The Ni2P is paramagnetic, so I didn’t consider spin polarized calculation, is this correct ? Sent from Yahoo Mail for iPhone On Sunday, August 16, 2020, 11:19 PM, Lucas Nicolás Lodeiro Moraga wrote: Hi! I am not an expert in adsorption, but I thought your input has some deficiencies. Firstly, forc_conv_thr is set for the maximum force for each relax-frame, the value that you use is too low, compared with your conv_thr (which define how accurate are your scf forces). Also, for relax, stress is not necessary and you could save some time neglecting it (I assume your slab is converged after for vc-relax). About the electronic parameters... 25 Ry for ecutwfc seems to me too low for USPPs. The fermi-dirac smearing is slower than others and the results do not change appreciably. Another thing is the vdW approach, you are using a vdW-functional plus the grimme's vdW-corrections... this does not seem correct. If you use a vdW-functional you are accounting the vdW interactions by the non-local correlation term, and the grimme's empirical correction is not needed, and vice versa, if you use the grimme's empirical corrections, you have to use a non vdw-functional as LDA,GGA or hybrid. A value of 0.7 for mixing_beta could be problematic for bad behaved systems. When I have problems on the first scf, I change this value to lower values like 0.5 or 0.3 to prevent long jumps in the scf solutions. Finally, something about the system-structure is weird to me. Your distances between S-P and S-Ni are too short, ~1.5 A. Those distances are too short to heavier atoms, and seems to me a bad guess for the first geometry (but I remember, I am not an expert in adsorptions, you know more about the system).Moreover, as you are putting the adsorbate only on one face of the system, you have a net dipole in the system. If you print the electrostatic potential profile along z, on vacuum space you will have a slope on the profile. It can be fixed, using a dipole background or using symmetrical adsorbates and surfaces on both sides of the slab. Also, you are simulating the adsorption of neutral atoms (in principle) of S. If you want to simulate Sulfur (S⁻²) adsorption, your system has to be a total negative charge, this requires some 2-D method to neglect the interactions between both sides of slab. Regards ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] Slab doesn't converge
Dear All I'm attempting to calculate the adsorption energy Sulfur (Monolayer coverage= 0.444) in 4-layer slab of Ni2P(001).The clean surface of Ni2P(001)(without Sulfur adsorbate) is converging well, but my problem is that the complex system (Sulfur + slab)is not converging even after 200 step. I've visualized the system in XcrysDen , and every thing looks fine.Is there any distance regulation for placing multiple adsorbate atoms in the surface ?Please help me , what is wrong with my input file attached ? Please find My input file ATTACHED input file Description: Binary data ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] ?==?utf-8?q? Same run not accelerated when starting from converged rho and wfc
6.d-9 is still too large.. it should be something like 1d-13 to aim at a smaller scf estimate. are you really starting from the scf charge and wfcs of the same conficuration ? stefano On 16/08/20 22:05, Antoine Jay wrote: Dear Stephano, adding diago_thr_int=1.0D-8 does not change the first conv_thr (exept the average#of iterations) As you said, the first value 1.0D-2 is detected to be too large and is updated to 6.0D-9 so I don't see why changing manually the first value would change something if it is already automatically changed... Antoine Jay LAAS-CNRS Toulouse, France Le Samedi, Août 15, 2020 17:10 CEST, Stefano de Gironcoli a écrit: Hi Antoine, don't know exactly why you get this result but one thing you can try is to set diag_thr_init ~ conv_thr/Nelec/10 so the first diagonalization is pushed tighter (if the wfcs are already very good it should not take too many iterations) and the computed dr2 estimate should be more faithful now diag_thr_int is 1.d-2 then updated to 6.e-9 which is consistent with conv_thr ~6.d-5... idk. you can try stefano On 14/08/20 17:09, Antoine Jay wrote: Dear all, I'am doing two consecutive scf calculations with exactly the same structure and parameters by calling qe6.5 as a library (attached output files). For the second call, I use the options: startingwfc='file' and input_rho ='rho.in' where these inputs are the converged wfc1.dat and charge-density.dat of the first step. Here I face two problems: -I expected that the initial scf accuracy is 10^-11 as obtained at the end of the first step, but it is only 10^-4. How is it possible to explain such a decrease? I generally loose only 2 orders of magnitude by doing this. -Even with less scf iterations, the cpu time is greater. Is it possible that some extra memory is allocated by qe when input rho and wfc are asked, and not desallocated? Note that until now, I have these troubles only when I use paw pseudopotentials on big systems. Regards, Antoine Jay LAAS-CNRS Toulouse France ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing listus...@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
Re: [QE-users] ?==?utf-8?q? Same run not accelerated when starting from converged rho and wfc
Dear Stephano, adding diago_thr_int=1.0D-8 does not change the first conv_thr (exept the average#of iterations) As you said, the first value 1.0D-2 is detected to be too large and is updated to 6.0D-9 so I don't see why changing manually the first value would change something if it is already automatically changed... Antoine Jay LAAS-CNRS Toulouse, France Le Samedi, Août 15, 2020 17:10 CEST, Stefano de Gironcoli a écrit: Hi Antoine, don't know exactly why you get this result but one thing you can try is to set diag_thr_init ~ conv_thr/Nelec/10 so the first diagonalization is pushed tighter (if the wfcs are already very good it should not take too many iterations) and the computed dr2 estimate should be more faithful now diag_thr_int is 1.d-2 then updated to 6.e-9 which is consistent with conv_thr ~6.d-5... idk. you can try stefano On 14/08/20 17:09, Antoine Jay wrote:Dear all, I'am doing two consecutive scf calculations with exactly the same structure and parameters by calling qe6.5 as a library (attached output files). For the second call, I use the options: startingwfc='file' and input_rho ='rho.in' where these inputs are the converged wfc1.dat and charge-density.dat of the first step. Here I face two problems: -I expected that the initial scf accuracy is 10^-11 as obtained at the end of the first step, but it is only 10^-4. How is it possible to explain such a decrease? I generally loose only 2 orders of magnitude by doing this. -Even with less scf iterations, the cpu time is greater. Is it possible that some extra memory is allocated by qe when input rho and wfc are asked, and not desallocated? Note that until now, I have these troubles only when I use paw pseudopotentials on big systems. Regards, Antoine Jay LAAS-CNRS Toulouse France ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] Subject: RE: Slab doesn't converge
Hi! I am not an expert in adsorption, but I thought your input has some deficiencies. Firstly, forc_conv_thr is set for the maximum force for each relax-frame, the value that you use is too low, compared with your conv_thr (which define how accurate are your scf forces). Also, for relax, stress is not necessary and you could save some time neglecting it (I assume your slab is converged after for vc-relax). About the electronic parameters... 25 Ry for ecutwfc seems to me too low for USPPs. The fermi-dirac smearing is slower than others and the results do not change appreciably. Another thing is the vdW approach, you are using a vdW-functional plus the grimme's vdW-corrections... this does not seem correct. If you use a vdW-functional you are accounting the vdW interactions by the non-local correlation term, and the grimme's empirical correction is not needed, and vice versa, if you use the grimme's empirical corrections, you have to use a non vdw-functional as LDA,GGA or hybrid. A value of 0.7 for mixing_beta could be problematic for bad behaved systems. When I have problems on the first scf, I change this value to lower values like 0.5 or 0.3 to prevent long jumps in the scf solutions. Finally, something about the system-structure is weird to me. Your distances between S-P and S-Ni are too short, ~1.5 A. Those distances are too short to heavier atoms, and seems to me a bad guess for the first geometry (but I remember, I am not an expert in adsorptions, you know more about the system). Moreover, as you are putting the adsorbate only on one face of the system, you have a net dipole in the system. If you print the electrostatic potential profile along z, on vacuum space you will have a slope on the profile. It can be fixed, using a dipole background or using symmetrical adsorbates and surfaces on both sides of the slab. Also, you are simulating the adsorption of neutral atoms (in principle) of S. If you want to simulate Sulfur (S⁻²) adsorption, your system has to be a total negative charge, this requires some 2-D method to neglect the interactions between both sides of slab. Regards ___ Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
[QE-users] Slab doesn't converge
Dear AllI'm attempting to calculate the adsorption energy Sulfur (Monolayer coverage= 0.444) in 4-layer slab of Ni2P(001).The clean surface of Ni2P(001)(without Sulfur adsorbate) is converging well, but my problem is that the complex system (Sulfur + slab)is not converging even after 200 step. I've visualized the system in XcrysDen , and every thing looks fine.Is there any distance regulation for placing multiple adsorbate atoms in the surface ?Please help me , what is wrong with my input file shown below ? Input file: calculation = "relax" prefix = 'ML=0.444' outdir = './outdir' pseudo_dir = '' restart_mode = 'from_scratch' forc_conv_thr = 1.0e-04 nstep = 200 tprnfor = .TRUE. tstress = .TRUE. / ibrav =0 ecutrho = 250 ecutwfc = 25 nat = 76 ntyp = 3 occupations='smearing',smearing='fermi-dirac',degauss=0.002 input_dft = "vdw-df" vdw_corr = 'grimme-d2' !nspin = 2 !tot_magnetization = 0 / conv_thr = 2.0e-05 electron_maxstep = 200 mixing_beta = 0.7 / ion_dynamics='bfgs' / K_POINTS {automatic} 5 5 1 1 1 0 ATOMIC_SPECIES Ni 58.69340 Ni.pbe-n-rrkjus_psl.0.1.UPF P 30.97376 P.pbe-n-rrkjus_psl.1.0.0.UPF S 32.065 S.pbe-n-rrkjus_psl.1.0.0.UPF ATOMIC_POSITIONS {angstrom} S 2.832675042 4.906337849 6.314453602 S 1.73011915912.99665591766.314453602 S -1.1040486364 7.90468975556.314453602 S 7.39768571822.99621228196.314453602 S 4.56286766787.90311984366.314453602 Ni 2.272856395 0.0 1.604817867 Ni 2.272856395 0.0 4.814453602 Ni -0.559819954 4.906337849 1.604817867 Ni -0.559819954 4.906337849 4.814453602 Ni 7.938207617 0.0 1.604817867 Ni 7.938207617 0.0 4.814453602 Ni 5.105531268 4.906337849 1.604817867 Ni 5.105531268 4.906337849 4.814453602 Ni 1.696246347 2.937986477 1.604817867 Ni 1.696246347 2.937986477 4.814453602 Ni -1.136429833 7.844324033 1.604817867 Ni -1.136429833 7.844324033 4.814453602 Ni 7.361598075 2.937986477 1.604817867 Ni 7.361598075 2.937986477 4.814453602 Ni 4.528921896 7.844324033 1.604817867 Ni 4.528921896 7.844324033 4.814453602 Ni -1.136428713 1.968351665 1.604817867 Ni -1.136428713 1.968351665 4.814453602 Ni -3.969105062 6.874689513 1.604817867 Ni -3.969105062 6.874689513 4.814453602 Ni 4.528922678 1.968351665 1.604817867 Ni 4.528922678 1.968351665 4.814453602 Ni 1.696246329 6.874689513 1.604817867 Ni 1.696246329 6.874689513 4.814453602 Ni -2.095321633 3.629202410 0.00 0 0 Ni -2.095321633 3.629202410 3.209635735 Ni -4.927997813 8.535539966 0.00 0 0 Ni -4.927997813 8.535539966 3.209635735 Ni 3.570029758 3.629202410 0.00 0 0 Ni 3.570029758 3.629202410 3.209635735 Ni 0.737353578 8.535539966 0.00 0 0 Ni 0.737353578 8.535539966 3.209635735 Ni 0.737354798 1.277135585 0.00 0 0 Ni 0.737354798 1.277135585 3.209635735 Ni -2.095321635 6.183473580 0.00 0 0 Ni -2.095321635 6.183473580 3.209635735 Ni 6.402706020 1.277135585 0.00 0 0 Ni 6.402706020 1.277135585 3.209635735 Ni 3.570029587 6.183473580 0.00 0 0 Ni 3.570029587 6.183473580 3.209635735 Ni 4.190642299 0.0 0.00 0 0 Ni 4.190642299 0.0 3.209635735 Ni 1.357965950 4.906337849 0.00 0 0 Ni 1.357965950 4.906337849 3.209635735 Ni 9.855993352 0.0 0.00 0 0 Ni 9.855993352 0.0 3.209635735 Ni 7.023317003 4.906337849 0.00 0 0 Ni 7.023317003 4.906337849 3.209635735 P 0.0 0.0