623 enhance safety

[johanneshofmann87]

Adds a centralized safety/validation layer to ensure Hamiltonian-global variables and core arrays are initialized and correctly set by each Ham_Set, preventing undefined/uninitialized state from propagating into the Monte Carlo core.

Changes Made

• Initialize key Hamiltonian globals (Ndim, N_FL, N_SUN, Ltrot, Thtrot, Projector, Symm) to sentinel/safe defaults in Alloc_Ham.
• Validate required Hamiltonian globals and allocations right after ham%Ham_set() (including Op_V, Op_T, and projector trial wavefunctions).
• Ensure Thtrot is explicitly initialized in relevant Ham_Set implementations.
• Add Hermiticity check in Op_set to validate operator matrices, merging Hermiticity and diagonal-detection into a single O(N²) upper-triangle pass.
• Add Op_equal function in Operator_mod to compare two Operator instances on their defining properties (used for structural checks).
• Add Op_V_is_symmetric function in Operator_mod that verifies the product (I+g₁V₁)···(I+gₙVₙ) equals its reverse — satisfied by palindromic arrangement or mutually commuting operators.
• Validate that all projector indices P(i) in Op_V and Op_T are within [1, Ndim] in Validate_Ham_Variables, preventing out-of-bounds memory access before any matrix indexing occurs.
• Validate that the Op_V operator sequence is symmetric when Symm = .true., terminating with a clear error if not.
• Add iso_fortran_env import for error_unit in Operator_mod.
• Add test suite coverage for Op_set validation (tests 28–32), including Hermiticity failure, type failure, projector index failure, and 1×1 matrix failure.
• Add test suite coverage for Op_equal and Op_V_is_symmetric (tests 33–35): operator equality comparison, symmetric Op_V pass, and asymmetric Op_V fail.

[Copilot]

Pull request overview

Adds a centralized safety/validation layer to ensure Hamiltonian-global variables and core arrays are initialized and correctly set by each Ham_Set, preventing undefined/uninitialized state from propagating into the Monte Carlo core.

Changes:

• Initialize key Hamiltonian globals (Ndim, N_FL, N_SUN, Ltrot, Thtrot, Projector, Symm) to sentinel/safe defaults in Alloc_Ham.
• Validate required Hamiltonian globals and allocations right after ham%Ham_set() (including Op_V, Op_T, and projector trial wavefunctions).
• Ensure Thtrot is explicitly initialized in several Hamiltonian Ham_Set implementations.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
Prog/main.F90	Calls Validate_Ham_Variables() immediately after ham%Ham_set(); removes redundant projector WF allocation pre-check now covered by validation.
Prog/Hamiltonian_main_mod.F90	Adds sentinel initialization (Init_Ham_Variables) and post-Ham_Set validation (Validate_Ham_Variables); exports the validator.
Prog/Hamiltonians/Hamiltonian_Hubbard_smod.F90	Explicitly initializes Thtrot = 0 before computing Ltrot.
Prog/Hamiltonians/Hamiltonian_Kondo_smod.F90	Explicitly initializes Thtrot = 0 before computing Ltrot.
Prog/Hamiltonians/Hamiltonian_Spin_Peierls_smod.F90	Explicitly initializes Thtrot = 0 before computing Ltrot.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[johanneshofmann87]

Thanks, @jonasschwab , for already taking a look at this. We run into this today where we noticed that we ran ALF with N_SUN=0, as it was not set explicitly anywhere. surprisingly, the code ran rather well and the results (mostly) made sense. but the fermion determinate did NOT influence the acceptance/rejection of the Metropolis step.

Since we are at enhancing the 'security' of the model definition: Are you aware of other implicit assumptions that we are relying on without checking them explicitly?
The hermitian matrix check (and tests in the test suite) are added now.

[Copilot]

Pull request overview

Copilot reviewed 14 out of 14 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[johanneshofmann87]

@copilot During review on GitHub, copilot provided the following suggestion:
"Op_set now performs an explicit O(N^2) Hermiticity scan over the upper triangle, and later in the routine it scans the upper triangle again to determine whether the operator is diagonal. For large-N diagonal operators this duplicates work in the setup path. Consider merging the Hermiticity and diagonal-detection logic into a single pass over (i,j) to avoid the second traversal and keep the validation cost minimal."

[Copilot]

@copilot During review on GitHub, copilot provided the following suggestion:
"Op_set now performs an explicit O(N^2) Hermiticity scan over the upper triangle, and later in the routine it scans the upp...

Done in commit b952e94. The Hermiticity scan and diagonal-detection now share a single O(N²) upper-triangle pass: Op%diag is initialized to .true. before the loop, and inside each iteration we first validate Hermiticity, then (guarded by if (Op%diag)) check whether the upper-triangle element is non-zero — only a single check per element needed since a passed Hermiticity check implies Op%O(I,J)==0 ↔ Op%O(J,I)==0. The redundant second loop is removed.

[johanneshofmann87]

@jonasschwab @fassaad I think this update covers some important consistency checks for models implemented by users that I am aware of. Do you know of additional implicit assumptions that we are not enforcing in the code?

[jonasschwab]

One could also add a test if the arrangement of operators is indeed symmetric if Symm is true.

[johanneshofmann87]

One could also add a test if the arrangement of operators is indeed symmetric if Symm is true.

I implemented such a test in the last commit.

[Copilot]

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 3 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[johanneshofmann87]

@copilot update the description of this PR to reflect the actual changes implemented here.

[Copilot]

@copilot update the description of this PR to reflect the actual changes implemented here.

Updated the PR description to accurately list all changes implemented: sentinel initialization, Validate_Ham_Variables checks (globals, allocations, projector index bounds, Symm symmetry), the Op_equal and Op_V_is_symmetric functions, merged Hermiticity+diagonal detection, error_unit import fix, and the full test coverage (tests 28–35).