DFT Production-Core Milestone

The DFT package contains two intentionally different surfaces. The legacy DFTSystem/run_scf surface supplies tiny Γ-point teaching, dense-reference, spin, occupation, finite-difference stress, and restart diagnostics. The periodic PeriodicDFTSystem/run_periodic_scf surface supplies the materials-workload path: PBE-PW92, reciprocal-space GTH operators, Monkhorst-Pack integration, block-Davidson/Rayleigh-Ritz solves, frozen-density band paths, and periodic forces.

The periodic implementation has verified results for specific workloads, but is not broadly chemically certified. Capability claims are tied to the material-validation summary.

Nonlocal Pseudopotentials

UPF and GTH nonlocal metadata is converted into normalized real-space separable projectors. The operator applies:

V̂_NL ψ = Σᵢ |βᵢ⟩ Dᵢ ⟨βᵢ|ψ⟩

For UPF, diagonal PP_DIJ values are used as projector couplings after Ry-to-Hartree conversion. For GTH, parsed projector coefficients seed the coupling. This gives a Hermitian validation path, but it is not yet a full chemically faithful reproduction of every format convention.

SCF applies nonlocal projectors by default when available. SCFConfig(apply_nonlocal=False) keeps the old local-only path available for debugging and comparison.

Solvers

Dense diagonalization remains the tiny-grid reference for the legacy path. Periodic SCF and bands use the MLX-native block-Davidson/Rayleigh-Ritz solver without building the full plane-wave Hamiltonian. Diagnostics expose residuals, orthonormality error, subspace work, and convergence metadata.

Spin, Occupations, k-Points, And Bands

The new spin layer is collinear only:

• unpolarized: one total density ρ(r).
• polarized: separate ρ↑(r) and ρ↓(r) diagnostics.

The legacy layer exposes fixed and Fermi-Dirac occupation diagnostics. The periodic layer uses reduced-coordinate Monkhorst-Pack meshes and 0.5|G + k|², including Bloch-phase local and nonlocal GTH evaluation. run_periodic_band_structure reuses a converged SCF density and solves non-self-consistently along a high-symmetry path.

Stress, Relaxation, And Restart

Finite-difference stress estimates diagonal orthorhombic stress by changing cell lengths and rerunning SCF. Geometry optimization remains ion-position-first by default, with config fields now prepared for cell and coupled relaxation modes.

Dense SCF restart files store density, orbitals, ion positions, cell lengths, spin metadata, and Γ k-point metadata for small-system continuation workflows.

Reference Validation

Reference comparison is intentionally static and lightweight. Fixtures are JSON summaries; QE/CP2K are not imported, built, or required in CI. The comparison helper records observed energy, expected energy, error, and pass/fail against a documented tolerance.

DFT/QM Platform Scope

get_dft_qm_scope_report() classifies local DFT/QM capability against CP2K and Quantum ESPRESSO reference families without changing the runtime dependency boundary.

Feature	Local Status	Reference Family
Plane-wave SCF core	verified (bulk-Si EOS)	CP2K Quickstep, QE PWscf
UPF/GTH pseudopotentials and nonlocal projectors	proof-level	QE UPF, CP2K GTH
Geometry relaxation and finite-difference stress	proof-level	CP2K MOTION/GEO_OPT, QE relax
Static reference comparison	supported	static CP2K/QE fixture summaries
QM/MM force-environment orchestration	deferred	CP2K FORCE_EVAL/QMMM
PH/EPW/NEB/TDDFT/MPI/offload suite breadth	deferred	QE and CP2K production suites
Importing, wrapping, building, or running CP2K/QE	anti-goal	external executables

Plane-wave SCF core is verified only for the bulk-silicon PBE equation of state against an all-electron (FLEUR/WIEN2k) reference (Lejaeghere Δ factor 1.942 meV/atom); broader chemistry stays proof-level, and the separate MLX-versus-QE PWscf cross-engine parity is still diagnostic, not closed.

dft_qm_scope_readiness_report() returns a shared readiness payload for these features. Deferred, anti-goal, and unknown features report blockers before any production-suite claim can be emitted.

Hot-Path Recommendation

The first future custom Metal kernel should target Hamiltonian application, specifically the combined kinetic + local + nonlocal application path used by Davidson and band calculations.

Reason:

• Dense diagonalization is a reference path and should not be optimized first.
• SCF and band workloads repeatedly apply Hψ.
• Nonlocal projector application adds many grid reductions and scatter-like projector accumulations.
• A fused Metal path can reduce Python-loop overhead before deeper eigensolver work.

Second-tier targets are projector construction/interpolation and orthonormalization. FFT/Hartree should be measured carefully before replacing MLX/Accelerate-backed paths.