[ase-users] An Error on a NEB calculation with VASP

[Esben Leonhard Kolsbjerg]

Hi Kaoru,

As the error says images cannot share calculators. In this for-loop

    for image in images:
        image.pbc = (True, True, True)
        image.set_cell(Lattice_vector)
        image.set_calculator(calc_grad)

you attach the SAME calculator that you initially created to each image/atoms object. Each image needs its own calculator. 
The easiest thing to do is to move the initialization of your calculator inside the for-loop such that you create a new instance of the calculator for each image, like:

    for image in images:
        image.pbc = (True, True, True)
        image.set_cell(Lattice_vector)
        calc = Vasp(...)
        image.set_calculator(calc)

Secondly, I'm not sure that an atoms object can have two calculators attached at the same time which it for me seems you are trying to do when invoking "set_calculator()" the second time in the beginning of your script, but I might be wrong. 

Hope it helps,

Esben

D. 24/04/2018 08.41 skrev "ase-users-bounces at listserv.fysik.dtu.dk på vegne af Kaoru Yamazaki via ase-users" <ase-users-bounces at listserv.fysik.dtu.dk på vegne af ase-users at listserv.fysik.dtu.dk>:

    Dear All,
    
    I have a trouble to run a NEB calculation using ASE 3.16 and VASP 5.44. 
    Even though I follow the NEB tutorial (https://wiki.fysik.dtu.dk/ase/tutorials/neb/diffusion.html),
     I received an error message as follows:
    
    Traceback (most recent call last):
      File "ceo-b-mfi_vasp_neb_180420.py", line 115, in <module>
        qn.run(fmax=0.03)
      File "/home/users/uv5/anaconda2/lib/python2.7/site-packages/ase/
    optimize/optimize.py", line 174, in run
        f = self.atoms.get_forces()
      File "/home/users/uv5/anaconda2/lib/python2.7/site-packages/ase/neb.py
    ", line 150, in get_forces
        raise ValueError(msg)
    ValueError: One or more NEB images share the same calculator.  Each 
    image must have its own calculator.  You may wish to use the ase.neb.
    SingleCalculatorNEB class instead, although using separate calculators 
    is recommended.
    
    My ASE-NEB input is as follows:
    
    from ase.neb import NEB
    from ase.build import bulk
    from ase import Atoms, Atom
    from ase.optimize import BFGS
    from ase.constraints import UnitCellFilter
    from ase.calculators.vasp import Vasp
    from ase.io import read, write
    from ase.neb import NEBTools
    
    #Configuration of vasp
    ## setup vasp for zeolite
    calc_guess = Vasp(prec='normal',
                algo='fast',
                xc='PBE',
               encut= 550,
                setups ={'Ce': '_3'}, # use Ce^3+ PP
                ivdw=11,
                ldau      = 'TRUE',
                ldautype  = 2,
    	ldau_luj={'Ce':{'L':3, 'U':4.5, 'J':0.0}, 
    		   'O':{'L':-1, 'U':0.0, 'J':0.0}, 
    		   'Si':{'L':-1, 'U':0.0, 'J':0.0}, 
                                 'B':{'L':-1, 'U':0.0, 'J':0.0},
                                 'C':{'L':-1, 'U':0.0, 'J':0.0}, 
                                 'H':{'L':-1, 'U':0.0, 'J':0.0}},
                icharg=12,
                ldauprint = 2,
                lmaxmix   = 6, 
                ncore      = 32,
                npar=8,
                lplane = 'TRUE',
                lscalu = 'FALESE',
                nsim = 8,
                ispin=2,
                nbands = 2560,
                laechg= 'TRUE', #Use DFT-D3 and GGA+U, 32 core 
                lreal='AUTO')
    
    
    calc_grad = Vasp(prec='normal',
                algo='fast',
                xc='PBE',
               encut= 550,
                setups ={'Ce': '_3'}, # use Ce^3+ PP
                ivdw=11,
                ldau      = 'TRUE',
                ldautype  = 2,
    	ldau_luj={'Ce':{'L':3, 'U':4.5, 'J':0.0}, 
    		   'O':{'L':-1, 'U':0.0, 'J':0.0}, 
    		   'Si':{'L':-1, 'U':0.0, 'J':0.0}, 
                                 'B':{'L':-1, 'U':0.0, 'J':0.0},
                                 'C':{'L':-1, 'U':0.0, 'J':0.0}, 
                                 'H':{'L':-1, 'U':0.0, 'J':0.0}},
                ldauprint = 2,
                lmaxmix   = 6, 
                ncore      = 32,
                npar=8,
                lplane = 'TRUE',
                lscalu = 'FALESE',
                nsim = 8,
                ispin=2,
                nelm=1000,
                nbands = 2560,
                laechg= 'TRUE', #Use DFT-D3 and GGA+U, 32 core 
                lreal='AUTO')
    
    #Lattice vector (optimized one without surface adsobate)
    Lattice_vector =[(20.090705234, 0.00135092434476, -0.0117978794464),
              (-0.00835128432838, 39.9536468604, -0.00596798669412),
              ( 0.0593384875412, -0.0040107523377, 13.3610212492)]
    
    # initial structure
    initial = read('initial.xyz')
    initial.pbc = (True, True, True)
    initial.set_cell(Lattice_vector)
    initial.set_calculator(calc_guess)
    initial.get_potential_energy()
    initial.set_calculator(calc_grad)
    initial.get_potential_energy()
    
    
    #Create final state
    final = read('final.xyz')
    final.pbc = (True, True, True)
    final.set_cell(Lattice_vector)
    
    #Generate blank images
    images = [initial]
    
    for i in range(5):
        images.append(initial.copy())
    
    for image in images:
        image.pbc = (True, True, True)
        image.set_cell(Lattice_vector)
        image.set_calculator(calc_grad)
    
    images.append(final)
    
    #Run IDPP interpolation
    neb = NEB(images, climb=True,method='eb')
    neb.interpolate('idpp')
    
    #Run NEB calculation
    qn = BFGS(neb, trajectory='neb_idpp.traj', logfile='neb_idpp.log',  
    restart='neb.pckl')
    qn.run(fmax=0.03)
    
    # print neb results
    images = read('neb_idpp.traj at -7:')
    nebtools = NEBTools(images)
    Ef, dE =  nebtools.get_barrier(fit=False, raw=True)
    print Ef
    print de
    write('neb.xyz', neb)
    
    I guess there is something wrong on the specification on the VASP 
    calculator on each NEB image but I have not fount what is wrong. Could 
    you kindly let me know possible solutions on this problem?
    
    Best regards,
    
    
    Kaoru Yamazaki, Ph.D.
    Assistant Professor
    Institute for Materials Research, 
    Tohoku University
    E-mail: kaoru.yamazaki at imr.tohoku.ac.jp
    
    
    _______________________________________________
    ase-users mailing list
    ase-users at listserv.fysik.dtu.dk
    https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users