nspin=2 + out_mat_hs2 writes empty hrs1/hrs2 CSR files while srs1 and dmrs are valid

[ReyMGil]

Describe the bug

For a collinear spin-polarized LCAO calculation (nspin = 2), out_mat_hs2 generates hrs1_nao.csr and hrs2_nao.csr, but both Hamiltonian files contain only empty CSR blocks (data_size = 0). Consequently, the Hamiltonian cannot be used by post-processing codes such as TB2J.

In contrast:

• srs1_nao.csr is correctly populated.
• dmrs1_nao.csr and dmrs2_nao.csr are correctly generated and differ from each other.
• Mulliken analysis shows finite magnetic moments, indicating that the SCF converged to a magnetic solution.

This suggests that the issue is specific to writing the real-space Hamiltonian for nspin = 2.

Expected behavior

hrs1_nao.csr and hrs2_nao.csr should contain the spin-up and spin-down real-space Hamiltonians, respectively, similarly to how srs1_nao.csr contains the overlap matrix.

To Reproduce

Use a collinear spin-polarized LCAO calculation with

basis_type lcao
nspin 2
out_mat_hs2 1 16
out_dmr 1
gamma_only 0
vl_in_h 1

After the SCF finishes, inspect

OUT./hrs1_nao.csr
OUT./hrs2_nao.csr

Both files contain only empty CSR blocks.

Environment

No response

Additional Context

• ABACUS version: v3.11.0-beta.3
• Basis: LCAO
• Spin mode: nspin = 2 (collinear AFM)
• out_mat_hs2 precision: 16 digits
• The same material calculated with noncollinear (nspin = 4) correctly produces a populated hrs1_nao.csr.
• dmrs1_nao.csr, dmrs2_nao.csr, and mulliken.txt confirm that the magnetic solution is not trivial.

I also inspected the implementation in ctrl_scf_lcao.cpp, where ModuleIO::write_hsr() receives

std::vector<hamilt::HContainer*> hr_vec = p_hamilt->getHR_vector();

I tested a local patch that explicitly called refresh(true) and updateHk() before getHR_vector(), but this did not change the output: hrs1_nao.csr and hrs2_nao.csr remained empty.

Questions

1. Is out_mat_hs2 currently expected to work for collinear (nspin = 2) LCAO calculations?
2. Are there additional input parameters required to populate hrs1_nao.csr and hrs2_nao.csr?
3. Could this be a regression introduced in recent versions of ABACUS?

Task list for Issue attackers (only for developers)

• Verify the issue is not a duplicate.
• Describe the bug.
• Steps to reproduce.
• Expected behavior.
• Error message.
• Environment details.
• Additional context.
• Assign a priority level (low, medium, high, urgent).
• Assign the issue to a team member.
• Label the issue with relevant tags.
• Identify possible related issues.
• Create a unit test or automated test to reproduce the bug (if applicable).
• Fix the bug.
• Test the fix.
• Update documentation (if necessary).
• Close the issue and inform the reporter (if applicable).