[gpaw-users] DFT+U-Failed to converge

Marcin Dulak Marcin.Dulak at fysik.dtu.dk
Tue May 29 10:52:29 CEST 2012


Hi,

i would disable the DipoleCorrectionPoissonSolver in the optimization.
It's know to cause convergence troubles.

Best regards,

Marcin


On 05/24/12 18:53, Monica Garcia Mota wrote:
> Dear GPAW-users,
>
> I would like to calculate the binding energies of adsorbate species on MnO2(110) surface with RPBE+U (U=2).
>
> The jobs are starting from the optimized RPBE geometries but they can not get convergence when the U parameter is applied.
> I successfully converged the structure using U=1 for some cases, but the convergence fails again if I increase the U to U=2.
>
> I would appreciate any advice you can give me.
>
> Here is the script:
>
> atoms = read('../O-MnO2-001-2layu.gpw')
> calc = GPAW('../O-MnO2-001-2layu.gpw',poissonsolver=DipoleCorrectionPoissonSolver(PoissonSolver(),2))
>
> atoms.set_pbc((1,1,0))
>
> mask=[atom.tag<1 for atom in atoms]
> atoms.set_constraint(FixAtoms(mask=mask))
>
> calc.set(eigensolver='dav', maxiter=480,mixer=BroydenMixerSum(0.1,6),txt=calc_name,convergence={'eigenstates': 1e-6})
> e1 = calc.get_potential_energy()
>
> calc.write(gpwdat0)
>
> l=2
> U_ev=[2.0]
>
> scale=1
> store=0
>
> for j in range(len(U_ev)):
>      for a in range(len(atoms)):
>     	if atoms[a].symbol == 'Mn':
>       	   calc.hamiltonian.setups[a].set_hubbard_u(U_ev[j]/Hartree,l,scale,store)
>
> calc.scf.reset()
> calc.attach(calc.write, 2,name0+str(U_ev[j])+'.gpw')
>
> atoms.set_calculator(calc)
> traj = PickleTrajectory(name0+str(U_ev[j]),'w',atoms)
>
> qn = QuasiNewton(atoms,logfile = name0+str(U_ev[j])+'.log')
> qn.attach(traj)
> qn.run(fmax=0.05)
>
> e2 = calc.get_potential_energy()
> niter2 = calc.get_number_of_iterations()
> calc.write(name0 + str(U_ev[j]) + '.gpw')
>
> write(name0+str(U_ev[j])+'.xyz',atoms)
>
>
> and the last part of the output:
>
> iter: 353  03:45:45  -2.4   -1.8     -176.26914   5      214      -0.6522
> iter: 354  03:47:08  -2.8   -1.8     -176.77445   3      136      -0.8145
> iter: 355  03:48:58  -2.4   -1.8     -177.26662   4      264      -1.4467
> iter: 356  03:50:47  -2.0   -1.6     -175.59455   4      259      -0.0897
> iter: 357  03:52:11  -2.2   -1.8     -176.64283   3      145      -0.4278
> iter: 358  03:53:34  -2.8   -1.8     -175.27459   4      136      -0.3667
> iter: 359  03:54:58  -3.0   -1.9     -174.91091   4      144      -0.5833
> iter: 360  03:56:34  -2.8   -2.0     -174.93828   3      199      -1.4358
> iter: 361  03:57:57  -2.8   -2.0     -175.19963   4      140      -1.8777
> iter: 362  03:59:35  -2.5   -2.1     -175.90281   4      204      +0.3426
> iter: 363  04:00:47  -2.8   -1.8     -175.68958   4      90       +0.1266
> iter: 364  04:02:40  -2.5   -1.9     -178.40883   3      269      -0.0667
> iter: 365  04:04:21  -2.2   -1.6     -174.72345   5      226      -0.5828
> iter: 366  04:05:43  -3.4   -2.1     -174.99554   3      132      -1.9902
> iter: 367  04:07:06  -3.2   -2.1     -175.45046   3      137      -2.1659
> iter: 368  04:08:39  -2.9   -2.1     -177.11007   5      186      -1.7775
> iter: 369  04:10:28  -2.1   -1.8     -175.60064   6      259      -0.2767
> iter: 370  04:11:48  -2.7   -1.9     -174.99349   5      124      -0.6728
> iter: 371  04:13:38  -2.5   -2.0     -177.39503   3      264      -0.3327
> iter: 372  04:15:20  -2.4   -1.6     -174.78822   4      224      -0.9620
> iter: 373  04:16:41  -2.9   -2.2     -174.96188   4      130      -1.6781
> iter: 374  04:18:03  -3.4   -2.3     -175.26695   3      132      -2.1362
> iter: 375  04:19:25  -3.6   -2.1     -175.49114   3      134      -2.1599
> iter: 376  04:20:50  -3.2   -2.1     -175.00059   4      146      -2.0644
> iter: 377  04:22:19  -2.8   -2.3     -175.43969   3      166      -0.4797
> iter: 378  04:23:34  -2.9   -1.9     -175.67681   4      99       -0.5623
> iter: 379  04:25:24  -2.5   -1.9     -175.07616   4      226      -1.8010
> iter: 380  04:26:46  -3.0   -2.0     -176.01589   3      131      -2.0533
> iter: 381  04:28:20  -2.6   -2.0     -178.57608   3      191      -0.6120
> iter: 382  04:29:57  -2.5   -1.6     -175.23117   3      203      -2.0527
> iter: 383  04:31:22  -2.3   -2.1     -175.21228   4      150      -2.0948
> iter: 384  04:32:55  -2.3   -2.2     -175.24795   4      185      -0.7070
> iter: 385  04:34:10  -2.9   -1.9     -175.86214   4      101      -0.6654
> iter: 386  04:35:35  -2.8   -1.9     -175.10458   4      146      -1.2730
> iter: 387  04:36:57  -3.0   -1.9     -175.14400   3      134      -1.2789
> iter: 388  04:38:39  -2.4   -1.9     -177.79251   7      222      -1.3896
> iter: 389  04:39:59  -2.5   -1.6     -178.98921   3      129      -1.5875
> iter: 390  04:41:48  -2.2   -1.6     -176.86945   4      256      -2.0411
> iter: 391  04:43:24  -2.1   -1.6     -178.29883   7      202      -0.5392
> iter: 392  04:45:36  -1.3   -1.6     -177.09284   14     360      -0.7396
> iter: 393  04:47:02  -1.4   -1.6     -180.72445   7      151      -0.0044
> iter: 394  04:48:25  -2.0   -1.6     -180.58141   4      140      -0.1527
> iter: 395  04:50:16  -0.7   -1.6     -126.44499   11     269      -2.4162
> iter: 396  04:52:01  -0.7   -1.6     -78.36229    6      239      -2.5894
> iter: 397  04:53:45  -1.0   -1.6     -135.72346   7      235      -2.7295
> iter: 398  04:55:47  -0.7   -1.6     -174.27179   12     317      -1.8428
> iter: 399  04:57:18  -0.9   -1.7     -175.23055   5      176      -1.9691
> iter: 400  04:59:03  -1.6   -1.8     -177.02439   6      243      -2.1142
>
>
> Best regards,
> Monica
>
>
>
> ---
> Monica Garcia Mota, Ph.D.
>
> Department of Chemical Engineering
> Stanford University
> Stanford, CA 94305-5025
>
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>


-- 
***********************************

Marcin Dulak
Technical University of Denmark
Department of Physics
Building 307, Room 229
DK-2800 Kongens Lyngby
Denmark
Tel.: (+45) 4525 3157
Fax.: (+45) 4593 2399
email: Marcin.Dulak at fysik.dtu.dk

***********************************



More information about the gpaw-users mailing list