[LKFEIYI]

Reminder

• I have read AGENTS.md and docs/developers_guide/agent_governance.md.
• I have linked an issue or explained why this PR does not need one.
• I have added adequate unit tests and/or case tests, or explained why not.
• I have listed the exact verification commands run and their results.
• I have described user-visible behavior changes, including INPUT parameter changes.
• I have explained core-module impact for ESolver, HSolver, ElecState, Hamilt, Operator, Psi, or other source/ changes.
• I have requested any needed governance exception below.

Linked Issue

Second part of #8052, split according to the review feedback.

Follow-up PRs, each stacked on the previous one: PCC coupling → lowpass and full cavity → SCF integration → solvent-aware cavity.

What's changed?

imp_sol 2 selects the self-consistent continuum solvation model (SCCS; Andreussi, Dabo and Marzari, J. Chem. Phys. 136, 064102 (2012)) for neutral, periodic cells, with calculation scf, energies and forces, PW and LCAO, nspin 1 and 2. The original ABACUS solvent model stays imp_sol 1, and true/false are still read as 1/0.

• Cavity and dielectric. ε(ρ) uses the Andreussi switching function between sccs_rho_min and sccs_rho_max. sccs_preset provides Environ's vacuum, water-neutral, water-cation and water-anion parameters; custom uses sccs_epsilon, sccs_rho_min/max, sccs_gamma and sccs_pressure.
• Electrostatics. The generalized Poisson equation is solved with Environ's square-root preconditioned CG. The solute charge is the electron density plus Gaussian ions with a fixed spread of 0.5 bohr (in qe-environ, this value is seldom changed, so I just fixed this value here).
• Non-electrostatic terms. Surface tension γS and pressure PV.

New code: source/source_hamilt/module_sccs/ (cavity, presets, periodic Coulomb operator, Gaussian ions and forces, sqrt-CG response, energy functional), the PotSccs potential component in source_estate/module_pot/, and read_inp_sccs.{h,cpp} for the INPUT keys.

Deliberate differences from Environ:

• The surface potential is the derivative of the regularized discrete surface, -div(grad s / |grad s|_eta), instead of Environ's unregularized curvature formula. With a density saddle on a grid point (He2), QE-Environ's surface potential reaches 100-3000 times its median and 4 of 6 runs do not converge in 200 steps; ABACUS converges in 5. So this approach seems better.
• The sqrt-CG stops when both the RMS (normalized by the grid size) and the maximum of the charge residual meet sccs_tol_rms and sccs_tol_max. Environ stops on the unnormalized sum(r^2), so its accuracy depends on the grid size.
• A previous sqrt-CG solution is used as the initial guess only if it lowers the initial residual.

Governance Notes

• INPUT/docs changes:
  • imp_sol becomes an Integer (0/1/2); true/false remain valid.
  • New parameters: sccs_preset, sccs_epsilon, sccs_rho_min, sccs_rho_max, sccs_gamma, sccs_pressure, sccs_maxiter, sccs_tol_rms, sccs_tol_max, sccs_surface_eta. They are listed right after the original solvent keys.
• Core module impact:
  • PotBase gains three virtual hooks with empty defaults: get_solvation_energy, add_solvation_force and solvent_electrostatic_potential. Potential forwards them to its components. No existing component overrides them, so other calculations are unchanged.
  • Potential::pot_register / get_pot_type take const Input_para& explicitly; the old one-argument pot_register forwards PARAM.inp. get_pot_type now reads gga_grad and dip_cor_flag from that argument instead of PARAM.inp.
  • ElecState::cal_energies takes imp_sol as int; imp_sol 2 reads the energies from the potential.
  • PW (force_pw.cpp) and LCAO (force_stress_lcao.cpp) add the solvation force through Potential::add_solvation_force; imp_sol 1 keeps its own path.

[mohanchen]

@ABACUS-PR-BOT Could you review the PR?

[ABACUS-PR-BOT]

Independently reviewed statically at 7cdc76650742af3d40120d8969465e46bce44172. The numerical finding is an analytic finite-grid limitation; molecular impact has not been measured. No builds, runtime/numerical jobs, or CLI/doc-generation checks were run; the required CI connector was unavailable. The trusted governance check exited 0 with warnings; git diff --check exited 2 for two added blank EOF lines.

Non-blocking (verification record): Please add the exact commands, configuration/executable identity and observed results for the SCCS unit/MPI/integration checks and any Environ comparison. The PR reminder is checked but the body has no verification section. Also record documentation-generation verification or why it was not run; those generated files are intentionally untracked.

[LKFEIYI]

The SCCS in the current PR is not fully ready, as it requires PCC for charged systems and a low-pass filter for cavity forces in the following PRs.

[mohanchen]

@ABACUS-PR-BOT do you have further comments?

[ABACUS-PR-BOT]

Follow-up static review at 7367e0ac1c960e679f557aab38249596ec247200: no additional implementation defects found. The scope-help, LCAO regression/force-gate and component-hook updates are present in the current source. I left one non-blocking regression-test follow-up in the existing nonfinite-input thread; the earlier finite-grid derivative limitation and verification-record request remain open.

git diff --check and the pinned governance check exited 0 (governance warnings remain). No builds, runtime tests or CI verification were performed.

[mohanchen]

LGTM. Thanks for your contributions.