Formatting files in /source/source_hamilt/module_xc

[mohanchen]

Formatting files in /source/source_hamilt/module_xc

[zzlinpku]

Code Review Summary

This is a well-executed code style refactoring PR. The new file naming scheme (xc_*.cpp for native implementations, libxc_*.cpp for Libxc wrappers) is clear and consistent. A few items to address:

🔴 Must fix: Header guard in libxc.h still uses the old name.
🟡 Consider: Update comment references to old file names, and note the reference member risk in exx_info.h.

[zzlinpku]

Code Review Summary

This is a well-structured refactoring PR. The file renames are consistent and improve readability, the indentation cleanup is thorough, and the exx_info.h lifetime warning is valuable.

However, there are two issues that should be addressed before merging — one potential name collision risk and one concern about mixing behavioral changes with formatting.

See inline comments for details.

[zzlinpku]

🔴 Header guard LIBXC_H is too generic — risk of collision with the actual libxc library.

The guard was renamed from XC_FUNCTIONAL_LIBXC_H to LIBXC_H. Since this project links against the external libxc library, a guard this generic could collide with libxc's own internal headers (or any future header that uses this common name). If a collision occurs, one header would silently be skipped, causing hard-to-diagnose compilation errors.

Suggestion: use a project-namespaced guard, e.g.:

#ifndef ABACUS_MODULE_XC_LIBXC_H
#define ABACUS_MODULE_XC_LIBXC_H

[zzlinpku]

🔴 Behavioral change: tau_xc signature now takes hybrid_alpha as a parameter.

The old implementation read GlobalC::exx_info.info_global.hybrid_alpha directly inside the function body. Passing it as a parameter is a good decoupling improvement, but this is a functional change hidden inside a formatting PR.

1. Please confirm all call sites have been updated (not just test_xc4.cpp). The new xc_grad.cpp must also pass the correct hybrid_alpha value wherever tau_xc / tau_xc_spin are called.
2. Consider noting this interface change explicitly in the PR description so reviewers don't miss it among ~3000 lines of formatting changes.

[zzlinpku]

🟡 Redundant zero-initialization — all these variables are assigned real values on the very next lines.

For example, p1 is initialized to 0.0 here but immediately set to 0.0232660 two lines below. This pattern repeats throughout xc_gga_corr.cpp, xc_gga_exch.cpp, and xc_gga_wrap.cpp (~200+ lines total). While it suppresses compiler warnings, it significantly inflates the diff and makes it harder to spot the actual logic changes (like the tau_xc signature change).

Suggestion: either combine declaration and assignment (double p1 = 0.0232660;), or keep the original uninitialized declarations since they're assigned before any use.

[zzlinpku]

🟢 Good addition. This lifetime dependency warning is valuable — the dangling-reference risk with const& members is a subtle trap, and documenting it here will prevent future bugs if Exx_Info is ever copied or moved.

[zzlinpku]

🟡 New #include dependency introduced.

The old xc_functional_gradcorr.cpp included xc_functional_libxc.h under USE_LIBXC, but did NOT include exx_info.h. This new include under __EXX is presumably needed because the old tau_xc read GlobalC::exx_info internally, while the new version requires the caller (gradcorr) to pass hybrid_alpha. This is consistent with the interface change — just flagging it for awareness since it's a new compile-time dependency.