<div dir="ltr"> Dear Vladimir, It worked after I set GRID_OPT to be TRUE.! Thank your Vladimir. I still need some your help since I has not much experiences in the EMBED calculations. I found that the embedding DFTB and PDB calculations were very hard to converged even after 100 optimize steps for H2O-H2 system. For example, after 50 iterations, I got following error massage, -------- Optimize embedding potential info at step = 50 ------------ Functional value = NaN Real energy change = NaN Step size = 0.5000000000 Trust radius = 0.5000000000 --------------------------------------------------- Convergence check : Maximum density difference = ******************** Convergence limit for max. density diff. = 0.0250000000 Convergence in max. density diff. = NO Integrated density difference = NaN Conv. limit for integrated density diff. = 0.5000000000 Convergence in integrated density diff. = NO Integrated squared density difference = NaN Maximum density change in: subsystem 1, spin 1: 0.0000000000 subsystem 2, spin 1: 0.0000000000 ******************************************************************************* * ___ * * / \ * * [ABORT] * * \___/ Embedding potential optimization not converged. * * | * * O/| * * /| | * * / \ force_env_methods.F:1483 * ******************************************************************************* As you can see, the “Integrated density difference” value was NaN and the embedding potential optimization was not converged. And I found that the embedding potential was also NaN too which wrote in the h2o_h2_dftb_pbe_DFET-embed_pot_*.cube files. Is these normal output for the embed calculation? Or how can I fix this? Quickstep- EMBEDDING POTENTIAL at optimization step 1 5 0.000000 0.000000 0.000000 75 0.125982 0.000000 0.000000 75 0.000000 0.125982 0.000000 75 0.000000 0.000000 0.125982 8 0.000000 3.307021 2.834589 0.000000 1 0.000000 1.889726 1.889726 0.000000 1 0.000000 4.724315 1.889726 0.000000 1 0.000000 3.307021 5.196747 0.000000 1 0.000000 3.307021 6.614041 0.000000 NaN NaN 0.00000E+00 NaN NaN NaN 0.00000E+00 NaN 0.00000E+00 0.00000E+00 0.00000E+00 0.00000E+00 NaN 0.00000E+00 0.00000E+00 NaN 0.00000E+00 NaN 0.00000E+00 NaN NaN NaN NaN 0.00000E+00 NaN 0.00000E+00 0.00000E+00 0.00000E+00 NaN NaN 0.00000E+00 0.00000E+00 NaN NaN NaN 0.00000E+00 0.00000E+00 NaN NaN 0.00000E+00 NaN 0.00000E+00 0.00000E+00 0.00000E+00 NaN NaN 0.00000E+00 0.00000E+00 NaN 0.00000E+00 0.00000E+00 NaN NaN 0.00000E+00 NaN NaN NaN 0.00000E+00 NaN 0.00000E+00 NaN 0.00000E+00 0.00000E+00 NaN NaN 0.00000E+00 NaN NaN NaN NaN 0.00000E+00 0.00000E+00 0.00000E+00 NaN NaN NaN 0.00000E+00 0.00000E+00 NaN 0.00000E+00 NaN Yours, Linfeng  在 2019年5月15日星期三 UTC+8下午7:51:27，Vladimir Rybkin写道：<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div dir="ltr">Dear Linfeng<div> </div><div>1) the first force_eval is the &EMBED one, the others are actual subsystems.</div><div> </div><div>2) >> The first FORCE_EVA section should be descripting how the other 4 FORCE_EVAL embed and how to mapping the fragments. The Am I correct?</div><div>Yes</div><div> </div><div>3) to 6) Yes</div><div> </div><div>The error is as follows: for DFTB there is no basis set. However, in OPT_EMBED you ask for GRID_OPT .FALSE. This is not obvious but: the embedding potential optimization is requested in the finite RI basis of the total system. This basis, however, is not specified in DFTB. So, to make a long story short, set GRID_OPT .TRUE. This is general, safe and fast (and the default).</div><div> </div><div>Yours,</div><div> </div><div>Vladimir среда, 15 мая 2019 г., 12:18:52 UTC+2 пользователь linfeng gan написал:<blockquote class="gmail_quote" style="margin:0;margin-left:0.8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"> Hi Vladimir, Thanks for your feedback. But there are still something confusing me. The input which I used was simply modified form “tests/QS/regtest-embed/H2O_H2_pbe_mp2.inp” in the CP2K test dir. Correct me if I understand the input keyword wrong. In the input system, there are a water molecular and a hydrogen molecular. (1) Question one The MULTIPLE_FORCE_EVALS section, I set FORCE_EVAL_ORDER 2 3 4 5 So, there should be 5 FORCE_EVAL sections in my input file. Am I correct? (2) Question two !******************************* !FORCE_EVA section 1 ! !mapping the fragments !******************************* The first FORCE_EVA section should be descripting how the other 4 FORCE_EVAL embed and how to mapping the fragments. The Am I correct? (3) Question three !******************************* !FORCE_EVA section 2 ! !Subsys 1 the H2O molecular !******************************* The second FORCE_EVA section should be descripting how to calculate the H2O molecular energy with DFTB. (4) Question 4 !******************************* !FORCE_EVA section 3 ! !Subsys 2 the H2 molecular !******************************* &DFT &QS CLUSTER_EMBED_SUBSYS .TRUE. &END QS &END DFT The third FORCE_EVA section should be descripting how to calculate the H2 molecular energy with DFTB. This is the embedded cluster and the CLUSTER_EMBED_SUBSYS .TRUE. is set. (5) Question five !******************************* !FORCE_EVA section 4 ! !<a name="6543435d-9023-4476-8117-c758c773616d@googlegroups.com_380ddddd-fc80-4b12-b63c-7227d3c5874d@googlegroups.com__Hlk8834912">Total system including H2O and H2</a> !******************************* &DFT &QS REF_EMBED_SUBSYS .TRUE. &OPT_EMBED REG_LAMBDA 0.00001 N_ITER 3 DENS_CONV_INT 0.5 DENS_CONV_MAX 0.025 OPTIMIZER QUASI_NEWTON GRID_OPT .FALSE. &END OPT_EMBED &END QS &END DFT The fourth FORCE_EVA section descripting how to calculate the total system, including H2O and H2 molecular, energy with DFTB. Because this is the total system, the REF_EMBED_SUBSYS .TRUE. is set. (6) Question six !******************************* !FORCE_EVA section 5 ! !Higher level calculation on subsys 2, the H2O !******************************* &DFT &QS HIGH_LEVEL_EMBED_SUBSYS .TRUE. &END QS &END DFT The fifth FORCE_EVA section descripting how to calculate the higher level system, in this case H2, energy with PBE. The HIGH_LEVEL_EMBED_SUBSYS .TRUE. should be set in this section. To avoid the grid influence I set the same grids for all the subsystem as follow, &MGRID NGRIDS 5 CUTOFF 600 REL_CUTOFF 60 &END MGRID But I still got the same error, ******************************************************************************* * ___ * * / \ * * [ABORT] * * \___/ The pointer gto_basis_set is not associated * * | * * O/| * * /| | * * / \ aobasis/basis_set_types.F:791 * ******************************************************************************* What should I do next? I post my input file in the end of this, Regards, Linfeng &GLOBAL PROJECT h2o_h2_dftb_pbe_DFET PRINT_LEVEL HIGH RUN_TYPE ENERGY &END GLOBAL &MULTIPLE_FORCE_EVALS FORCE_EVAL_ORDER 2 3 4 5 MULTIPLE_SUBSYS T &END MULTIPLE_FORCE_EVALS !******************************* !FORCE_EVA section 1 ! !mapping the fragments !******************************* &FORCE_EVAL METHOD EMBED !Select DFET or DMFET. Default value: DFET &EMBED DFET NGROUPS 1 &MAPPING &FORCE_EVAL_EMBED &FRAGMENT 1 1 3 &END &FRAGMENT 2 4 5 &END &FRAGMENT 3 1 5 &END &END &FORCE_EVAL 1 &FRAGMENT 1 1 3 MAP 1 &END &END &FORCE_EVAL 2 &FRAGMENT 1 1 2 MAP 2 &END &END &FORCE_EVAL 3 &FRAGMENT 1 1 5 MAP 3 &END &END &FORCE_EVAL 4 &FRAGMENT 1 1 2 MAP 2 &END &END &END &END EMBED &SUBSYS &CELL ABC [angstrom] 5.000 5.000 5.000 &END CELL &COORD O 1.75 1.5 0.0 H 1.0 1.0 0.0 H 2.5 1.0 0.0 H 1.75 2.75 0.0 H 1.75 3.50 0.0 &END &END SUBSYS &END FORCE_EVAL !******************************* !FORCE_EVA section 2 ! !Subsys 1 the H2O molecular !******************************* &FORCE_EVAL METHOD Quickstep &SUBSYS &CELL ABC [angstrom] 5.000 5.000 5.000 &END CELL &COORD O 1.75 1.5 0.0 H 1.0 1.0 0.0 H 2.5 1.0 0.0 &END COORD &END SUBSYS &DFT &QS METHOD DFTB &DFTB SELF_CONSISTENT T &PARAMETER PARAM_FILE_PATH DFTB/scc PARAM_FILE_NAME scc_parameter UFF_FORCE_FIELD ../uff_table &END PARAMETER &END DFTB &END QS &POISSON &EWALD EWALD_TYPE SPME GMAX 25 &END EWALD POISSON_SOLVER ANALYTIC &END POISSON &MGRID NGRIDS 5 CUTOFF 600 REL_CUTOFF 60 &END MGRID &SCF MAX_SCF 100 !Maximum number of SCF iteration &OT PRECONDITIONER FULL_SINGLE_INVERSE MINIMIZER DIIS &END OT &END SCF &END DFT STRESS_TENSOR ANALYTICAL &END FORCE_EVAL !******************************* !FORCE_EVA section 3 ! !Subsys 2 the H2 molecular !******************************* &FORCE_EVAL METHOD Quickstep &SUBSYS &CELL ABC [angstrom] 5.000 5.000 5.000 &END CELL &COORD H 1.75 2.75 0.0 H 1.75 3.50 0.0 &END COORD &END SUBSYS &DFT &QS CLUSTER_EMBED_SUBSYS .TRUE. METHOD DFTB &DFTB SELF_CONSISTENT T &PARAMETER PARAM_FILE_PATH DFTB/scc PARAM_FILE_NAME scc_parameter UFF_FORCE_FIELD ../uff_table &END PARAMETER &END DFTB &END QS &POISSON &EWALD EWALD_TYPE SPME GMAX 25 &END EWALD POISSON_SOLVER ANALYTIC &END POISSON &MGRID NGRIDS 5 CUTOFF 600 REL_CUTOFF 60 &END MGRID &SCF MAX_SCF 100 &OT PRECONDITIONER FULL_SINGLE_INVERSE MINIMIZER DIIS &END OT &END SCF &END DFT STRESS_TENSOR ANALYTICAL &END FORCE_EVAL !******************************* !FORCE_EVA section 4 ! !Total system including H2O and H2 !******************************* &FORCE_EVAL METHOD Quickstep &SUBSYS &CELL ABC [angstrom] 5.000 5.000 5.000 &END CELL &COORD O 1.75 1.5 0.0 H 1.0 1.0 0.0 H 2.5 1.0 0.0 H 1.75 2.75 0.0 H 1.75 3.50 0.0 &END COORD &END SUBSYS &DFT &QS REF_EMBED_SUBSYS .TRUE. &OPT_EMBED REG_LAMBDA 0.00001 N_ITER 3 DENS_CONV_INT 0.5 DENS_CONV_MAX 0.025 OPTIMIZER QUASI_NEWTON GRID_OPT .FALSE. &END OPT_EMBED !&OPT_DMFET !&END EPS_DEFAULT 1.0E-10 !Default value: 1.00000000E-010 METHOD DFTB &DFTB SELF_CONSISTENT T &PARAMETER PARAM_FILE_PATH DFTB/scc PARAM_FILE_NAME scc_parameter UFF_FORCE_FIELD ../uff_table &END PARAMETER &END DFTB &END QS &POISSON &EWALD EWALD_TYPE SPME GMAX 25 &END EWALD POISSON_SOLVER ANALYTIC &END POISSON &MGRID NGRIDS 5 CUTOFF 600 REL_CUTOFF 60 &END MGRID &SCF MAX_SCF 100 &OT PRECONDITIONER FULL_SINGLE_INVERSE MINIMIZER DIIS &END OT &END SCF &END DFT STRESS_TENSOR ANALYTICAL &END FORCE_EVAL !******************************* !FORCE_EVA section 5 ! !Higher level calculation on subsys 2, the H2O !******************************* &FORCE_EVAL METHOD Quickstep &DFT BASIS_SET_FILE_NAME BASIS_RI_cc-TZ POTENTIAL_FILE_NAME HF_POTENTIALS &QS HIGH_LEVEL_EMBED_SUBSYS .TRUE. METHOD GPW EPS_DEFAULT 1.0E-15 EPS_PGF_ORB 1.0E-30 &END QS &SCF MAX_SCF 100 &OT PRECONDITIONER FULL_SINGLE_INVERSE MINIMIZER DIIS &END OT &END SCF &XC &XC_FUNCTIONAL PBE &END XC_FUNCTIONAL &END XC &POISSON &EWALD EWALD_TYPE SPME GMAX 25 &END EWALD POISSON_SOLVER ANALYTIC &END POISSON &MGRID NGRIDS 5 CUTOFF 600 REL_CUTOFF 60 &END MGRID &END DFT &SUBSYS &CELL ABC [angstrom] 5.000 5.000 5.000 &END CELL &KIND H BASIS_SET cc-TZ BASIS_SET RI_AUX RI_TZ POTENTIAL GTH-HF-q1 &END KIND &KIND O BASIS_SET cc-TZ BASIS_SET RI_AUX RI_TZ POTENTIAL GTH-HF-q6 &END KIND &COORD H 1.75 2.75 0.0 H 1.75 3.50 0.0 &END &END SUBSYS &END FORCE_EVAL </div></blockquote></div></div></blockquote></div>