Split vnl_pw.cpp into several smaller files (deepmodeling#7869)

* Refactor vnl_pw, step 1: extract getvnl into vnl_pw_getvnl.cpp

- Move getvnl<FPTYPE, Device> template body + 4 explicit
  instantiations (CPU/GPU float/double) from vnl_pw.cpp
  to new file vnl_pw_getvnl.cpp (~165 lines)
- Update CMakeLists.txt and Makefile.Objects accordingly
- vnl_pw.cpp reduced from 1797 to 1641 lines

Verification:
  cd build_max_para_test && make -j 30
  => build passed, no undefined reference

* Style: split multi-variable declarations in vnl_pw_getvnl.cpp

- One variable per line for better readability and maintainability
- Rename pointer variables: _tab -> tab_ptr, _indv -> indv_ptr,
  _nhtol -> nhtol_ptr, _nhtolm -> nhtolm_ptr
  (avoid leading underscore, use _ptr suffix)

Verification:
  cd build_max_para_test && make -j 30
  => build passed

* Doc: add one-variable-per-line rule to AGENTS.md

Rule deepmodeling#8 in Required Baseline: declare one variable per line;
do not use comma-separated declarations.

This improves readability, makes diffs cleaner, and helps
debugging with single-step execution.

* Refactor: replace new/delete with std::vector in vnl_pw_getvnl.cpp

- Replace int* h_atom_nh/na/nb with std::vector<int>
- Replace ModuleBase::Vector3<double>* _gk with std::vector<Vector3>
- Use .data() to get raw pointer for GPU sync operations
- Remove all delete[] calls (RAII handles cleanup automatically)

Benefits:
- Exception safety (no memory leak if exception thrown)
- Cleaner code (no manual memory management)
- Follows modern C++ best practices

Verification:
  cd build_max_para_test && make -j 30
  => build passed

* Doc: add doxygen comments to vnl_pw_getvnl.cpp

- Add file-level @file/@brief documentation
- Add getvnl() function documentation with workflow steps
- Add comment explaining explicit template instantiations

* Step 2: extract qrad/radial_fft_q into vnl_pw_qrad.cpp

- Move compute_qrad(), radial_fft_q() (CPU matrix version),
  radial_fft_q<FPTYPE,Device>() (template version), and
  4 explicit template instantiations to vnl_pw_qrad.cpp
- Replace new[]/delete[] with std::vector in compute_qrad()
- Add doxygen comments for all functions
- Update CMakeLists.txt and Makefile.Objects

Verification:
  cd build_max_para_test && make -j 30
  => build passed

* Step 3: extract cal_effective_D/newq/newd_* into vnl_pw_deeq.cpp

- Move cal_effective_D(), newq(), newd_so(), newd_nc() to vnl_pw_deeq.cpp
- Replace new[]/delete[] qnorm with std::vector in newq()
- Add doxygen comments for all functions
- Update CMakeLists.txt and Makefile.Objects
- Fix blas_connector.h include path (source_base/module_external/)

Verification:
  cd build_max_para_test && make -j 30
  => build passed

* refactor(pwdft): move LCAO alpha-channel VNL helpers to vnl_pw_alpha.cpp

Step 4 of vnl_pw.cpp split:
- Move Cal_C(), CG(), init_vnl_alpha() into new vnl_pw_alpha.cpp
- Wrap whole file with #ifdef __LCAO
- Drop dead commented-out getvnl_alpha block
- Replace new[]/delete[] with std::vector for jl/aux in init_vnl_alpha
- Sync CMakeLists.txt and Makefile.Objects

Verified: make -j 30 in build_max_para_test passes.

* refactor(pwdft): move init_vnl into vnl_pw_init_vnl.cpp

Step 5 of vnl_pw.cpp split:
- Move init_vnl() into new vnl_pw_init_vnl.cpp
- Replace new[]/delete[] with std::vector for jl/aux inside the tab fill loop
- Sync CMakeLists.txt and Makefile.Objects

Verified: make -j 30 in build_max_para_test passes.

* refactor(pwdft): clean up vnl_pw.cpp after split, drop unused includes

Step 6 of vnl_pw.cpp split (final):
- Keep only ctor/dtor, release_memory(), init(), print_vnl(),
  rescale_vnl() and the get_*_data<T>() template specializations
- Drop now-unused includes (clebsch_gordan, math_integral, math_polyint,
  math_sphbes, math_ylmreal, parallel_reduce, vnl_op, parallel_comm)
- Add file-level doxygen comment describing the new module layout

Verified: make -j 30 in build_max_para_test passes.

* Remove accidentally added empty CONNECT file

The empty CONNECT file was unintentionally added in commit 3b53f58
during the vnl_pw_qrad split. It is not referenced by any code, build
file, or script, so it is safe to delete.

* update format

* refactor(pwdft): replace raw new/delete arrays with std::vector in vl/vnl pseudopotential code

- convert pseudopot_cell_vl local buffers (vloc1d/aux1/aux) and member zp to std::vector
- convert pseudopot_cell_vnl::indv_ijkb0 to std::vector and drop manual delete
- use std::fill/std::copy instead of ZEROS/COPYARRAY helpers on vectors

* update AGENTS.md

* docs: add std::vector conversion guideline to AGENTS.md baseline

* style(pwdft): replace tab indentation with 4 spaces in vl_pw files

* refactor(pw): decouple print_vloc from PARAM and move it out of init_vloc

- print_vloc takes out_element_info and global_out_dir explicitly instead
  of reading PARAM internally
- callers invoke print_vloc right after init_vloc at all four call sites

* refactor(ofdft): cache PARAM.inp.nspin into local const in KEDF_XWM::get_energy

Cache PARAM.inp.nspin into a local const to reduce repeated global
parameter accesses, and add the missing newline at end of file.

---------

Co-authored-by: abacus_fixer <mohanchen@pku.eud.cn>

Author: mohanchen

AGENTS.md

@@ -8,7 +8,7 @@ rules. Read the complete governance document before making or reviewing changes:
 
 ## Required Baseline
 
-- Follow the seven ABACUS coding rules summarized from the project governance:
+- Follow the nine ABACUS coding rules summarized from the project governance:
   1. Do not increase cross-layer control through `GlobalV`, `GlobalC`, or
      `PARAM`; pass dependencies explicitly where practical. Migration-neutral
      moves must keep the PR-level global dependency budget non-increasing and
@@ -23,6 +23,9 @@ rules. Read the complete governance document before making or reviewing changes:
   6. Add focused tests for key features, bug fixes, INPUT behavior changes,
      heterogeneous kernels, and core-module refactors.
   7. Keep code compatible with the repository C++11 baseline.
+  8. Declare one variable per line; do not use comma-separated declarations.
+  9. Do not call MPI routines directly; use the internally-guarded wrappers
+     (e.g., `Parallel_Reduce::reduce_*`, `Parallel_Common::bcast_*`) instead.
 - Use LF line endings for text files. Only `.bat` and `.cmd` files may use CRLF.
 - Keep source file additions deterministic: update the relevant `CMakeLists.txt`
   or explain why the file is generated or included indirectly.
@@ -31,6 +34,10 @@ rules. Read the complete governance document before making or reviewing changes:
   is required.
 - Report the exact verification performed. Do not claim completion without
   fresh test or check output.
+- Prefer `std::vector` over raw `new`/`delete` for dynamic arrays; before
+  converting class members, confirm no external code consumes them as raw
+  pointers (e.g., `std::vector<bool>` has no `.data()`), and use
+  `std::fill`/`std::copy` instead of `ZEROS`/`COPYARRAY` on vector buffers.
 
 ## Repository Map
 
@@ -83,6 +90,9 @@ rules. Read the complete governance document before making or reviewing changes:
   test sufficiency, and exception approval require human review.
 - Exceptions must be recorded in the PR with reason, scope, risk, and a follow-up
   cleanup plan.
+- After a refactor, propose brief lessons worth recording in this file, then
+  ask the developer whether to write them in; be cautious and skip unclear
+  or unverified lessons.
 
 ## Local Commands
 

source/Makefile.Objects

@@ -723,7 +723,12 @@ OBJS_SRCPW=h_ewald_pw.o\
     dnrm2.o\
     vl_pw.o\
     vnl_pw.o\
+    vnl_pw_alpha.o\
+    vnl_pw_deeq.o\
+    vnl_pw_getvnl.o\
     vnl_pw_grad.o\
+    vnl_pw_init_vnl.o\
+    vnl_pw_qrad.o\
     chgmixing.o\
     charge.o\
     charge_init.o\

source/source_esolver/esolver_fp.cpp

@@ -168,6 +168,8 @@ void ESolver_FP::before_scf(UnitCell& ucell, const int istep)
 
         // reset local pseudopotentials
         this->locpp.init_vloc(ucell, this->pw_rhod);
+        this->locpp.print_vloc(ucell, this->pw_rhod,
+                               this->inp_->out_element_info, PARAM.globalv.global_out_dir);
         ModuleBase::GlobalFunc::DONE(GlobalV::ofs_running, "LOCAL POTENTIAL");
 
         // perform symmetry analysis

source/source_esolver/esolver_of.cpp

@@ -86,6 +86,7 @@ void ESolver_OF::before_all_runners(BaseCell& basecell, const Input_para& inp)
 
     // initialize local pseudopotential
     this->locpp.init_vloc(ucell,pw_rho);
+    this->locpp.print_vloc(ucell, pw_rho, inp.out_element_info, PARAM.globalv.global_out_dir);
     ModuleBase::GlobalFunc::DONE(GlobalV::ofs_running, "LOCAL POTENTIAL");
 
 

source/source_estate/setup_estate_pw.cpp

@@ -107,6 +107,7 @@ void setup_estate_pw_impl(
     }
 
     locpp.init_vloc(ucell, pw_rhod);
+    locpp.print_vloc(ucell, pw_rhod, inp.out_element_info, PARAM.globalv.global_out_dir);
     ModuleBase::GlobalFunc::DONE(GlobalV::ofs_running, "LOCAL POTENTIAL");
 
     ppcell.init(ucell, &sf, pw_wfc);

source/source_lcao/lcao_set.cpp

@@ -69,6 +69,7 @@ void LCAO_domain::set_pot(
 {
     //! 1) init local pseudopotentials
     locpp.init_vloc(ucell, &pw_rho);
+    locpp.print_vloc(ucell, &pw_rho, inp.out_element_info, PARAM.globalv.global_out_dir);
 
     //! 2) init potentials
     if (pelec->pot == nullptr)

source/source_pw/module_ofdft/kedf_xwm.cpp

@@ -64,22 +64,23 @@ void KEDF_XWM::set_para(double dV,
  */
 double KEDF_XWM::get_energy(const double* const* prho, ModulePW::PW_Basis* pw_rho)
 {
-    double** w1Rho5_6 = new double*[PARAM.inp.nspin];
-    for (int is = 0; is < PARAM.inp.nspin; ++is)
+    const int nspin = PARAM.inp.nspin;
+    double** w1Rho5_6 = new double*[nspin];
+    for (int is = 0; is < nspin; ++is)
     {
         w1Rho5_6[is] = new double[pw_rho->nrxx];
     }
     this->multi_kernel(prho, this->kernel1_.data(), w1Rho5_6, this->kappa_5_6, pw_rho);
 
-    double** w2Rho5_6 = new double*[PARAM.inp.nspin];
-    for (int is = 0; is < PARAM.inp.nspin; ++is)
+    double** w2Rho5_6 = new double*[nspin];
+    for (int is = 0; is < nspin; ++is)
     {
         w2Rho5_6[is] = new double[pw_rho->nrxx];
     }
     this->multi_kernel(prho, this->kernel2_.data(), w2Rho5_6, this->kappa_5_6, pw_rho);
 
     double energy = 0.; // in Ry
-    if (PARAM.inp.nspin == 1)
+    if (nspin == 1)
     {
         for (int ir = 0; ir < pw_rho->nrxx; ++ir)
         {
@@ -88,14 +89,14 @@ double KEDF_XWM::get_energy(const double* const* prho, ModulePW::PW_Basis* pw_rh
         }
         energy += this->dV_;
     }
-    else if (PARAM.inp.nspin == 2)
+    else if (nspin == 2)
     {
         // TODO: spin polarized
     }
     this->xwm_energy = energy;
     Parallel_Reduce::reduce_all(this->xwm_energy);
 
-    for (int is = 0; is < PARAM.inp.nspin; ++is)
+    for (int is = 0; is < nspin; ++is)
     {
         delete[] w1Rho5_6[is];
         delete[] w2Rho5_6[is];
@@ -379,4 +380,4 @@ void KEDF_XWM::fill_kernel(double tf_weight, double vw_weight, ModulePW::PW_Basi
         this->kernel1_[ig] = (this->c_0 * lindhard + this->c_2 * diff_lindhard) * this->c_kernel;
         this->kernel2_[ig] = this->c_1 * diff_lindhard * this->c_kernel;
     }
-}
+}

source/source_pw/module_pwdft/CMakeLists.txt

@@ -43,7 +43,12 @@ list(APPEND objects
     stress_pw.cpp
     vl_pw.cpp
     vnl_pw.cpp
+    vnl_pw_alpha.cpp
+    vnl_pw_deeq.cpp
+    vnl_pw_getvnl.cpp
     vnl_pw_grad.cpp
+    vnl_pw_init_vnl.cpp
+    vnl_pw_qrad.cpp
     stru_fac.cpp
     stru_fac_k.cpp
     soc.cpp