add omp_sz_hacker_ForLargeSystems

[tmisawa]

#204

[Copilot]

Pull Request Overview

This PR enhances the existing hacker mode by adding a specialized path for large systems (mode 2) and implements the supporting functions.

• Broadened the hacker check to >=1 and introduced a new branch for hacker==2 calling omp_sz_hacker_ForLargeSystems.
• Added implementation of omp_sz_hacker_ForLargeSystems, make_true_spin, update_lists, and compare_ulong in src/sz.c.
• Updated sz.h with prototypes for the new functions and added error handling for invalid hacker values.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
src/sz.c	Modified branching on hacker value, added new large-system support mode, introduced helper functions and error checks.
src/include/sz.h	Declared prototypes for omp_sz_hacker_ForLargeSystems, make_true_spin, update_lists, and compare_ulong.

Comments suppressed due to low confidence (5)

src/sz.c:1711

• [nitpick] Function name 'omp_sz_hacker_ForLargeSystems' mixes case and underscores; consider using a consistent snake_case style (e.g. 'omp_sz_hacker_for_large_systems') to match existing conventions.

unsigned long int omp_sz_hacker_ForLargeSystems(

src/sz.c:397

• [nitpick] The error message refers to 'read_hacker in ModPara file' which may confuse readers; consider simplifying to 'Invalid hacker value: must be 0, 1, or 2.' and reference the actual parameter name.

          fprintf(stderr, "Error: read_hacker in ModPara file must be 0, 1 or 2 for Hubbard model.");

src/sz.c:1698

• The doc comment for 'omp_sz_hacker_ForLargeSystems' should more precisely describe the return value semantics, valid input ranges, and behavior on edge cases (e.g., when ihfbit is zero).

/** 

src/include/sz.h:118

• The new API 'omp_sz_hacker_ForLargeSystems' (and its helpers) lack dedicated unit tests; consider adding tests to verify correct behavior across different system sizes and edge conditions.

unsigned long int omp_sz_hacker_ForLargeSystems(

src/sz.c:1801

• [nitpick] Sorting the full list for very large systems may be a bottleneck. Consider more efficient on-the-fly or streaming approaches to avoid high memory and CPU usage.

      qsort(tmp, len_n, sizeof(long unsigned int), compare_ulong);

[Copilot]

The result of malloc is not checked; if allocation fails, 'tmp' will be NULL, leading to undefined behavior. Add a NULL check and handle allocation failures.

[k-yoshimi]

@tmisawa
Please correspond to the following comments:

Comments suppressed due to low confidence (5)

src/sz.c:1711

• [nitpick] Function name 'omp_sz_hacker_ForLargeSystems' mixes case and underscores; consider using a consistent snake_case style (e.g. 'omp_sz_hacker_for_large_systems') to match existing conventions.

unsigned long int omp_sz_hacker_ForLargeSystems(

src/sz.c:397

• [nitpick] The error message refers to 'read_hacker in ModPara file' which may confuse readers; consider simplifying to 'Invalid hacker value: must be 0, 1, or 2.' and reference the actual parameter name.

          fprintf(stderr, "Error: read_hacker in ModPara file must be 0, 1 or 2 for Hubbard model.");

src/sz.c:1698

• The doc comment for 'omp_sz_hacker_ForLargeSystems' should more precisely describe the return value semantics, valid input ranges, and behavior on edge cases (e.g., when ihfbit is zero).

/** 

src/include/sz.h:118

• The new API 'omp_sz_hacker_ForLargeSystems' (and its helpers) lack dedicated unit tests; consider adding tests to verify correct behavior across different system sizes and edge conditions.

unsigned long int omp_sz_hacker_ForLargeSystems(

src/sz.c:1801

• [nitpick] Sorting the full list for very large systems may be a bottleneck. Consider more efficient on-the-fly or streaming approaches to avoid high memory and CPU usage.

      qsort(tmp, len_n, sizeof(long unsigned int), compare_ulong);

[tmisawa]

@k-yoshimi
I have resolved the conflict.
I have also checked the comments by Copilot and made the necessary changes where possible.