-
Notifications
You must be signed in to change notification settings - Fork 30
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Convergence issues with LASSIS #87
Comments
Is this occurring in the |
Please be aware that I am by no means guaranteed to merge |
In the lsi.kernel() Here is the output file #INFO: **** input file is /project/lgagliardi/jangidbhavnesh/LASSIS_NIL/Cu2O2/bis/bisoxo.py **** mol = gto.Mole() mf = scf.ROHF(mol).x2c() from pyscf.mcscf import avas molden.from_mo(mol, 'bisoxo.molden', mo_coeff) from mrh.my_pyscf.mcscf.lasscf_o0 import LASSCF las = asyn.LASSCF(mf, (10,12,10),((5,4),(5, 5), (4, 5)),spin_sub=(2,1,2)) from mrh.my_pyscf.tools import molden lsi = lassi.LASSIS(las1) from mrh.my_pyscf import mcpdft #INFO: ******************** input file end ******************** System: uname_result(system='Linux', node='midway3-0552.rcc.local', release='4.18.0-305.3.1.el8.x86_64', version='#1 SMP Tue Jun 1 16:14:33 UTC 2021', machine='x86_64') Threads 24 [ENV] PYSCF_EXT_PATH /home/jangidbhavnesh/bin/pyscf:/home/jangidbhavnesh/bin/pyscf-forge:/home/jangidbhavnesh/bin:/home/jangidbhavnesh/bin/pyscf:/home/jangidbhavnesh/bin/pyscf-forge:/home/jangidbhavnesh/bin:/project/lgagliardi/jangidbhavnesh/Apps/anaconda3/envs/block2env/lib/python3.11/site-packages/pyscf-forge:/project/lgagliardi/jangidbhavnesh/Apps/anaconda3/envs/block2env/lib/python3.11/site-packages/mrh:/project/lgagliardi/jangidbhavnesh/Apps/anaconda3/envs/block2env/lib/python3.11/site-packages/dmrgscf: nuclear repulsion = 728.388422725618 ******** <class 'pyscf.x2c.sfx2c1e.sfX2C1eROHF'> ******** ******** <class 'pyscf.x2c.sfx2c1e.SpinFreeX2CHelper'> ******** ******** AVAS flags ******** ** AVAS ** of alpha electrons 17of beta electrons 17Entering product-state fixed-point CI iteration WARN: CI vectors for charge-separated rootspace 2 not converged Electron hop space 3: |
I will compile the system where it happened, and will send you. |
This convergence issue is also happening for the peroxo Cu2O2]2+, where i am using similar fragmentation and active spaces, I am including here the root spaces for your reference to debug it. Electron hop space 8: |
In all of these cases the individual FCI solves are failing, so I'm not sure what I can do about it. Have you tried playing with pspace_size? |
Increasing pspace_size did converge few of the above non-convergence behavior. Why don't we have default pspace_size of 500 or higher for LAS? I guess it will require large memory!! |
Yeah, issue #48 |
The branch lassis_ncharge_norb has solved all non-convergence cases of LASSIS. But now it is not-converging for the Cu2O2 system
I think the error is in the initial CI vector guess, and thus it is oscillating
Cycle 0: max grad = 1.303990e+00 ; sigma = 0.000000e+00 ; 0/2 fragment CI solvers converged
Cycle 1: max grad = 9.742448e-01 ; sigma = 6.188922e-02 ; 1/2 fragment CI solvers converged
Cycle 2: max grad = 9.792066e-01 ; sigma = 4.230953e-02 ; 1/2 fragment CI solvers converged
Cycle 3: max grad = 9.792031e-01 ; sigma = 4.545490e-04 ; 1/2 fragment CI solvers converged
Cycle 4: max grad = 9.792048e-01 ; sigma = 1.062201e-05 ; 1/2 fragment CI solvers converged
Cycle 5: max grad = 9.792054e-01 ; sigma = 1.975213e-07 ; 1/2 fragment CI solvers converged
Cycle 6: max grad = 9.792050e-01 ; sigma = 1.399439e-08 ; 1/2 fragment CI solvers converged
Cycle 7: max grad = 9.792051e-01 ; sigma = 1.501803e-08 ; 1/2 fragment CI solvers converged
Cycle 8: max grad = 9.792052e-01 ; sigma = 1.800345e-09 ; 1/2 fragment CI solvers converged
Cycle 9: max grad = 9.792055e-01 ; sigma = 2.839442e-09 ; 1/2 fragment CI solvers converged
Cycle 10: max grad = 9.792054e-01 ; sigma = 6.354458e-08 ; 1/2 fragment CI solvers converged
Cycle 11: max grad = 9.792054e-01 ; sigma = 3.990772e-08 ; 1/2 fragment CI solvers converged
Cycle 12: max grad = 9.792052e-01 ; sigma = 4.644789e-09 ; 1/2 fragment CI solvers converged
Cycle 13: max grad = 9.792058e-01 ; sigma = 1.801436e-08 ; 1/2 fragment CI solvers converged
Cycle 14: max grad = 9.792054e-01 ; sigma = 1.346507e-09 ; 1/2 fragment CI solvers converged
Cycle 15: max grad = 9.792051e-01 ; sigma = 3.278728e-10 ; 1/2 fragment CI solvers converged
Cycle 16: max grad = 9.792059e-01 ; sigma = 8.958523e-09 ; 1/2 fragment CI solvers converged
Cycle 17: max grad = 9.792051e-01 ; sigma = 5.277934e-08 ; 1/2 fragment CI solvers converged
Cycle 18: max grad = 9.792058e-01 ; sigma = 3.797504e-08 ; 1/2 fragment CI solvers converged
Cycle 19: max grad = 9.792054e-01 ; sigma = 1.282569e-08 ; 1/2 fragment CI solvers converged
Cycle 20: max grad = 9.792057e-01 ; sigma = 1.194849e-08 ; 1/2 fragment CI solvers converged
Cycle 21: max grad = 9.792059e-01 ; sigma = 2.576917e-08 ; 1/2 fragment CI solvers converged
Cycle 22: max grad = 9.792060e-01 ; sigma = 4.345384e-08 ; 1/2 fragment CI solvers converged
Cycle 23: max grad = 9.792058e-01 ; sigma = 3.733658e-08 ; 1/2 fragment CI solvers converged
Cycle 24: max grad = 9.792058e-01 ; sigma = 4.252797e-09 ; 1/2 fragment CI solvers converged
Cycle 25: max grad = 9.792060e-01 ; sigma = 5.333186e-08 ; 1/2 fragment CI solvers converged
Cycle 26: max grad = 9.792054e-01 ; sigma = 6.186610e-08 ; 1/2 fragment CI solvers converged
Cycle 27: max grad = 9.792056e-01 ; sigma = 5.591119e-09 ; 1/2 fragment CI solvers converged
Cycle 28: max grad = 9.792055e-01 ; sigma = 5.536913e-08 ; 1/2 fragment CI solvers converged
Cycle 29: max grad = 9.792055e-01 ; sigma = 2.399429e-08 ; 1/2 fragment CI solvers converged
Cycle 30: max grad = 9.792063e-01 ; sigma = 2.225988e-09 ; 1/2 fragment CI solvers converged
Cycle 31: max grad = 9.792060e-01 ; sigma = 2.205525e-09 ; 1/2 fragment CI solvers converged
Cycle 32: max grad = 9.792058e-01 ; sigma = 4.471667e-08 ; 1/2 fragment CI solvers converged
Cycle 33: max grad = 9.792057e-01 ; sigma = 6.477512e-08 ; 1/2 fragment CI solvers converged
Cycle 34: max grad = 9.792056e-01 ; sigma = 2.736670e-09 ; 1/2 fragment CI solvers converged
Cycle 35: max grad = 9.792053e-01 ; sigma = 4.129106e-10 ; 1/2 fragment CI solvers converged
Cycle 36: max grad = 9.792058e-01 ; sigma = 7.185008e-11 ; 1/2 fragment CI solvers converged
Cycle 37: max grad = 9.792060e-01 ; sigma = 2.335582e-08 ; 1/2 fragment CI solvers converged
Cycle 38: max grad = 9.792053e-01 ; sigma = 4.656385e-08 ; 1/2 fragment CI solvers converged
Cycle 39: max grad = 9.792061e-01 ; sigma = 7.513609e-08 ; 1/2 fragment CI solvers converged
Cycle 40: max grad = 9.792055e-01 ; sigma = 2.608431e-09 ; 1/2 fragment CI solvers converged
Cycle 41: max grad = 9.792060e-01 ; sigma = 1.083663e-09 ; 1/2 fragment CI solvers converged
Cycle 42: max grad = 9.792055e-01 ; sigma = 5.857692e-08 ; 1/2 fragment CI solvers converged
Cycle 43: max grad = 9.792057e-01 ; sigma = 9.714040e-08 ; 1/2 fragment CI solvers converged
Cycle 44: max grad = 9.792054e-01 ; sigma = 1.164703e-07 ; 1/2 fragment CI solvers converged
Cycle 45: max grad = 9.792055e-01 ; sigma = 4.130015e-09 ; 1/2 fragment CI solvers converged
Cycle 46: max grad = 9.792049e-01 ; sigma = 1.829358e-08 ; 1/2 fragment CI solvers converged
Cycle 47: max grad = 9.792057e-01 ; sigma = 1.263970e-08 ; 1/2 fragment CI solvers converged
Cycle 48: max grad = 9.792056e-01 ; sigma = 3.470041e-08 ; 1/2 fragment CI solvers converged
Cycle 49: max grad = 9.792060e-01 ; sigma = 6.678783e-08 ; 1/2 fragment CI solvers converged
Input
import numpy
from pyscf import gto
from pyscf import scf, lib
from pyscf import mcscf
from pyscf.tools import molden
from mrh.my_pyscf.fci import csf_solver
from mrh.my_pyscf.gto import ANO_RCC_VDZP
mol = gto.Mole()
mol.max_memory = 100000
mol.atom = 'bisoxo.xyz'
mol.basis = ANO_RCC_VDZP
mol.symmetry = False
mol.verbose = 4
mol.spin = 0
mol.charge = 2
mol.build()
mf = scf.ROHF(mol).x2c()
mf.max_cycle = 100
mf.conv_tol = 1e-12
mf.init_guess='chk'
mf.chkfile='bisoxo.chk'
mf.kernel()
from pyscf.mcscf import avas
ncas, nelecas, mo_coeff = avas.kernel(mf, ['Cu 3d', 'Cu 4d', 'O 2p', 'O 3p'], minao=mol.basis)
molden.from_mo(mol, 'bisoxo.molden', mo_coeff)
from mrh.my_pyscf.mcscf.lasscf_o0 import LASSCF
from mrh.my_pyscf.mcscf import lasscf_async as asyn
from mrh.my_pyscf import lassi
from mrh.my_pyscf.fci import csf_solver
las = asyn.LASSCF(mf, (10,12,10),((5,4),(5, 5), (4, 5)),spin_sub=(2,1,2))
'''
mo_list = [33,34,35,36,37,38,39,40,41,47] + [42,43,44,45,46,48] + [x for x in range(49, 65)]
mo_coeff = las.sort_mo (mo_list, mo_coeff)
mo0 = las.set_fragments_(([0],[2, 3], [1]), mo_coeff)
las = lassi.states.spin_shuffle(las)
las.weights = [1.0/las.nroots,]*las.nroots
las.max_cycle_macro = 200
las.kernel(mo0)
from mrh.my_pyscf.tools import molden
molden.from_lasscf(las, 'bisoxo.las.molden')
'''
from mrh.my_pyscf.mcscf.lasscf_o0 import LASSCF
las1 = LASSCF(mf, (10,12,10),((5,4),(5, 5), (4, 5)),spin_sub=(2,1,2))
mo_coeff = lib.chkfile.load (las.chkfile, 'las')['mo_coeff']
las1.lasci_(mo_coeff)
lsi = lassi.LASSIS(las1)
lsi.kernel()
XYZ
12
Cu 0.000000 1.400000 0.000000
Cu 0.000000 -1.400000 0.000000
O 0.000000 0.000000 1.150000
O 0.000000 0.000000 -1.150000
N 0.000000 3.400000 0.000000
N 0.000000 -3.400000 0.000000
H -0.939693 3.742020 0.000000
H 0.939693 -3.742020 0.000000
H 0.469846 3.742020 0.813798
H -0.469846 -3.742020 -0.813798
H 0.469846 3.742020 -0.813798
H -0.469846 -3.742020 0.813798
The text was updated successfully, but these errors were encountered: