Historical task source. The Eve task board holds current task state.
Phase 127: Kalika Materials Code Interoperability & Databases — VASP, Quantum ESPRESSO, ABINIT, LAMMPS, Phonopy, WANNIER90, ASE, pymatgen Compatibility & Materials Database APIs#
Comprehensive interfaces to all major computational materials science codes and databases. This phase is critical for researcher adoption — users need to integrate Kalika with their existing DFT, MD, and post-processing workflows. Targets feature parity with ASE, pymatgen, atomate/atomate2, FireWorks/jobflow, and custodian. Covers input generation, output parsing, error handling, database APIs, and workflow orchestration for VASP, Quantum ESPRESSO, ABINIT, FHI-aims, LAMMPS, GROMACS, Phonopy, WANNIER90, LOBSTER, BoltzTraP2, and 10+ additional codes. TypeScript (~70%) for orchestration and I/O logic, Rust (~10%) for performance-critical parsing (large trajectory files, charge density grids), Python (~20%) for bridges to the ASE/pymatgen/AiiDA ecosystems.
127.1 Scaffolding & Core Types (@kalika/code-interfaces)#
127.1.1 Library Scaffolding#
- 127.1.1.1 Create directory structure for
libs/kalika/code-interfaces/withpackage.json,project.json,tsconfig.json, andsrc/index.tsbarrel export — register path mapping intsconfig.base.json - 127.1.1.2 Create directory structure for
libs/kalika/vasp-io/withpackage.json,project.json,tsconfig.json, andsrc/index.ts— add path mapping - 127.1.1.3 Create directory structure for
libs/kalika/qe-io/withpackage.json,project.json,tsconfig.json, andsrc/index.ts— add path mapping - 127.1.1.4 Create directory structure for
libs/kalika/lammps-io/withpackage.json,project.json,tsconfig.json, andsrc/index.ts— add path mapping - 127.1.1.5 Create directory structure for
libs/kalika/materials-databases/withpackage.json,project.json,tsconfig.json, andsrc/index.ts— add path mapping - 127.1.1.6 Create directory structure for
libs/kalika/workflows/withpackage.json,project.json,tsconfig.json, andsrc/index.ts— add path mapping - 127.1.1.7 Create directory structures for ancillary interface
libraries:
libs/kalika/abinit-io/,libs/kalika/fhi-aims-io/,libs/kalika/phonopy-io/,libs/kalika/wannier90-io/,libs/kalika/lobster-io/,libs/kalika/boltztrap-io/,libs/kalika/gromacs-io/— each withpackage.json,project.json,tsconfig.json,src/index.ts, and path mappings
127.1.2 Core Type Definitions#
- 127.1.2.1 Implement
CalculationInputtype hierarchy: base type withcode: CodeName,version?: string,structure: KalikaStructure,parameters: Record<string, unknown>,resources: ComputeResources(nprocs, memory, walltime, queue),workDir: string,inputFiles: Map<string, string>— plus code-specific subtypesVaspInput,QeInput,AbinitInput,LammpsInputeach with typed parameter sets - 127.1.2.2 Implement
CalculationOutputtype hierarchy: base type withcode: CodeName,status: JobStatus,walltime: number,exitCode: number,energy?: number,forces?: Float64Array[],stress?: Float64Array,structure?: KalikaStructure,rawFiles: Map<string, Buffer>,parseErrors: ParseError[]— plus code-specific subtypes with typed parsed results (eigenvalues, DOS, band structure, etc.) - 127.1.2.3 Implement
CodeInterfaceabstract class:generateInputs(calc: CalculationInput): Map<string, string | Buffer>,parseOutputs(files: Map<string, Buffer>): CalculationOutput,validateInputs(calc: CalculationInput): ValidationResult[],detectErrors(output: CalculationOutput): CodeError[],suggestFixes(errors: CodeError[]): Fix[]— the contract every code adapter must implement - 127.1.2.4 Implement
WorkflowSteptype:id: string,name: string,code: CodeName,inputTemplate: CalculationInput,dependencies: string[](step IDs),outputMapping: Map<string, string>(how outputs from dependencies map to this step's inputs),condition?: (outputs: Map<string, CalculationOutput>) => boolean(conditional execution),retryPolicy: RetryPolicy - 127.1.2.5 Implement
WorkflowDAGclass: topological sort ofWorkflowStepnodes, cycle detection, dependency resolution,addStep(),removeStep(),connect(from, to, mapping),validate(): ValidationResult[],getExecutionOrder(): string[][](returns layers of parallelizable steps), serialization to/from JSON - 127.1.2.6 Implement
JobStatusenum and state machine:CREATED → SUBMITTED → QUEUED → RUNNING → COMPLETED | FAILED | CANCELLED, with transition validation, timestamp tracking for each state change, andJobStatusHistoryrecording all transitions
127.1.3 Structure Conversion Utilities#
- 127.1.3.1 Implement
KalikaStructurecanonical type: lattice vectors (3x3 matrix), species list, fractional coordinates, Cartesian coordinates, site properties (magnetic moment, charge, selective dynamics flags),toCartesian(),toFractional(),getVolume(),getLatticeParameters()(a, b, c, alpha, beta, gamma),getReciprocalLattice(),getSuperCell(nx, ny, nz),applySymmetry(spaceGroup) - 127.1.3.2 Implement CIF ↔ KalikaStructure converter: parse CIF files (IUCr CIF 1.1 and 2.0 format) including symmetry operations, site occupancies, anisotropic displacement parameters — and write CIF from KalikaStructure with space group, Wyckoff positions, and all standard CIF data blocks
- 127.1.3.3 Implement POSCAR/CONTCAR ↔ KalikaStructure converter: read/write VASP 5.x format (species line, direct/Cartesian, selective dynamics, velocities) — exact formatting to match VASP expectations
- 127.1.3.4 Implement XYZ/extXYZ ↔ KalikaStructure converter: standard XYZ and extended XYZ (with lattice, properties columns) — supporting multi-frame trajectories
- 127.1.3.5 Implement PDB ↔ KalikaStructure converter: Protein Data Bank format for biomolecular structures — ATOM/HETATM records, CRYST1 (unit cell), CONECT (bonds)
127.1.4 Unit System & Physical Constants#
- 127.1.4.1 Implement unit conversion registry: energy (eV, Ry, Ha, J,
kcal/mol, kJ/mol, cm⁻¹, K, meV), length (Å, Bohr, nm, pm), force (eV/Å,
Ha/Bohr, N), pressure (GPa, kbar, eV/ų, Ha/Bohr³, Pa), time (fs, ps, au),
mass (amu, kg, electron mass) — with composable unit arithmetic:
convert(value, 'eV/Ų', 'Ha/Bohr²') - 127.1.4.2 Implement physical constants module: CODATA 2018 values for Bohr radius, Hartree energy, Rydberg constant, electron mass, Boltzmann constant, Planck constant, speed of light, fine structure constant, Avogadro number — all with uncertainties, and derived constants (eV_to_Ha, Bohr_to_Ang, etc.)
127.2 VASP Interface (@kalika/vasp-io)#
127.2.1 INCAR Generation#
- 127.2.1.1 Implement INCAR writer: serialize a typed
IncarParamsobject to VASP INCAR format — proper formatting of scalars, arrays (MAGMOM), booleans (.TRUE./.FALSE.), string values, comment lines, tag ordering by category (general → electronic → ionic → output) - 127.2.1.2 Implement INCAR tag validation engine: validate all major tags against allowed values (ALGO: Normal/Fast/VeryFast/All/Damped/..., PREC: Accurate/Normal/Low/..., ISPIN: 1/2, LREAL: Auto/.TRUE./.FALSE., ISMEAR: -5/-4/-1/0/1/2/..., IBRION: -1/0/1/2/3/5/6/7/8/44), range checks (ENCUT > 0, EDIFF > 0, SIGMA > 0, NSW >= 0), mutual exclusion checks (IBRION with NSW=0 is static), and dependency checks (LORBIT requires appropriate RWIGS or PAW)
- 127.2.1.3 Implement INCAR tag database: comprehensive metadata for every supported VASP tag — name, type (int/float/bool/string/array), default value, allowed values/range, description, VASP version introduced, dependencies on other tags, category (electronic/ionic/output/symmetry/ parallel/hybrid/GW/response) — at least 150 tags covering all VASP functionality
- 127.2.1.4 Implement INCAR preset:
static— IBRION=-1, NSW=0, ISMEAR=-5 (tetrahedron with Blöchl corrections for insulators) or ISMEAR=1/SIGMA=0.2 (metals), LREAL=Auto, PREC=Accurate, EDIFF=1e-6, LORBIT=11, LWAVE=.FALSE., LCHARG=.TRUE. - 127.2.1.5 Implement INCAR preset:
relaxation— IBRION=2 (CG), ISIF=3 (full cell+ions), NSW=100, EDIFF=1e-6, EDIFFG=-0.02 (force convergence), LREAL=Auto, PREC=Accurate, ISMEAR=1/SIGMA=0.2 (metals) or ISMEAR=0 (insulators), LWAVE=.FALSE., LCHARG=.FALSE. - 127.2.1.6 Implement INCAR preset:
band_structure— non-self-consistent calculation: ICHARG=11, ISMEAR=0/SIGMA=0.05, LORBIT=11, LWAVE=.FALSE., LCHARG=.FALSE., NEDOS=2001 — designed to be chained after a static run using the CHGCAR - 127.2.1.7 Implement INCAR preset:
dos— ISMEAR=-5 (tetrahedron), NEDOS=3001, LORBIT=11 (projected DOS), EMIN/EMAX auto-set from band edges, ICHARG=11 (non-SCF from CHGCAR) - 127.2.1.8 Implement INCAR preset:
dielectric— LEPSILON=.TRUE. (DFPT dielectric), IBRION=8 (DFPT), EDIFF=1e-8, LREAL=.FALSE., PREC=Accurate, plus variant for LOPTICS=.TRUE. (frequency-dependent dielectric from RPA) - 127.2.1.9 Implement INCAR preset:
neb— IBRION=3 (damped MD for NEB), POTIM=0, ICHAIN=0, IMAGES=N (configurable), SPRING=-5, LCLIMB=.TRUE. (CI-NEB), EDIFF=1e-5, NSW=200 - 127.2.1.10 Implement INCAR preset:
hybrid— LHFCALC=.TRUE., AEXX=0.25 (HSE06), HFSCREEN=0.2, ALGO=All (or Damped), TIME=0.4, PRECFOCK=Fast, NKRED if applicable — with variant for PBE0 (AEXX=0.25, HFSCREEN=0) - 127.2.1.11 Implement INCAR preset:
gw— ALGO=GW, NOMEGA=100, NBANDS=large, ENCUTGW=auto, LSPECTRAL=.TRUE., plus variants for G0W0, GW0, scGW, and BSE (ALGO=BSE, NBANDSO/NBANDSV) - 127.2.1.12 Implement INCAR preset:
phonon_finite_diff— IBRION=5 (finite differences) or IBRION=6 (central differences), NFREE=2, EDIFF=1e-8, PREC=Accurate, LREAL=.FALSE., ADDGRID=.TRUE., ISYM=0 - 127.2.1.13 Implement INCAR preset:
elastic— IBRION=6, ISIF=3, NFREE=4, EDIFF=1e-8, PREC=Accurate, LREAL=.FALSE. - 127.2.1.14 Implement INCAR preset:
nmr— LNMR_SYM_RED=.TRUE., NSLPLINE=.TRUE., LCHIMAG=.TRUE., DQ=0.001, ICHIBARE=1-3, EDIFF=1e-8, LREAL=.FALSE. - 127.2.1.15 Implement INCAR preset:
aimd— IBRION=0, ISIF=2 (NVT) or ISIF=3 (NPT), SMASS=0 (NVT via Nosé-Hoover) or SMASS=-1 (NVE), TEBEG/TEEND (temperatures), POTIM=1.0 (1 fs timestep), NSW=10000, PREC=Normal, ALGO=VeryFast, LREAL=Auto, LWAVE=.FALSE. - 127.2.1.16 Implement INCAR preset:
soc— LSORBIT=.TRUE., LNONCOLLINEAR=.TRUE., SAXIS specification, NBANDS typically doubled, GGA_COMPAT=.FALSE., LORBMOM=.TRUE.
127.2.2 POSCAR & KPOINTS Generation#
- 127.2.2.1 Implement POSCAR writer: write VASP 5+ POSCAR from KalikaStructure — comment line, scaling factor, lattice vectors (16-digit precision), element symbols line, atom counts line, Selective dynamics (optional T/F flags), Direct/Cartesian toggle, fractional/Cartesian coordinates (16-digit precision), optional velocities block
- 127.2.2.2 Implement POSCAR reader: parse VASP 5+ POSCAR/CONTCAR — handle scaling factor (including negative = volume), lattice vectors, species (VASP 5 element line), counts, selective dynamics flags, direct/Cartesian coordinates, velocities — return KalikaStructure
- 127.2.2.3 Implement KPOINTS writer: automatic Monkhorst-Pack mesh (subdivisions + shift), automatic Gamma-centered mesh, and explicit k-point list mode — with auto-generation of k-mesh density from target k-point density (e.g., 50 k-points/Å⁻¹) and lattice parameters
- 127.2.2.4 Implement KPOINTS line-mode writer: generate k-point path for band structure calculations — input: list of high-symmetry points (Γ, X, M, K, L, W, U, etc.) with labels and coordinates, number of points per segment; include automatic high-symmetry path detection from space group (using Setyawan-Curtarolo conventions)
- 127.2.2.5 Implement high-symmetry k-path database: Brillouin zone high-symmetry points and standard paths for all 14 Bravais lattices (cF, cI, cP, tI, tP, oF, oI, oC, oP, hR, hP, mC, mP, aP) following Setyawan-Curtarolo (Comp. Mat. Sci. 49, 2010) conventions
127.2.3 POTCAR Management#
- 127.2.3.1 Implement POTCAR library manager: configure path to local
VASP POTCAR library (e.g.,
$VASP_PP_DIR/), scan for available functionals (PBE, PBE_52, PBE_54, LDA, PW91), list available potentials per element per functional, and concatenate POTCARs in the correct element order - 127.2.3.2 Implement recommended POTCAR selection: maintain a mapping of recommended PAW potentials per element (e.g., Li_sv, Na_pv, Ca_sv, Ti_pv, etc.) following pymatgen/Materials Project conventions — with fallbacks when recommended potential is unavailable
- 127.2.3.3 Implement POTCAR validation: verify POTCAR integrity via checksum (SHA-256 or TITEL hash), ensure functional consistency (all potentials from same functional family), verify element order matches POSCAR, extract ENMAX values for automatic ENCUT determination (ENCUT = max(ENMAX) × 1.3)
127.2.4 Output Parsing#
- 127.2.4.1 Implement OUTCAR parser: extract total energy (E0, energy without entropy, energy(sigma→0)), ionic forces (eV/Å), stress tensor (kbar), magnetization (per atom and total), timing breakdown, number of SCF iterations per ionic step, convergence history (energy, force, stress per step), warnings and errors
- 127.2.4.2 Implement OUTCAR advanced property parser: Born effective charges (LEPSILON), macroscopic dielectric tensor (ionic + electronic contributions), elastic tensor C_ij (6×6), piezoelectric tensor e_ij, NMR chemical shielding tensors (σ_iso, σ_aniso, η), electric field gradients (V_zz, η), internal strain tensor
- 127.2.4.3 Implement vasprun.xml parser: full XML parsing of VASP's structured output — extract: calculation parameters (INCAR as parsed), k-points, basis set info, initial/final structure, electronic structure (eigenvalues at each k-point and band, occupations), DOS (total + projected onto atoms/orbitals), forces, stress, energy per ionic step, dielectric function (real + imaginary parts vs. energy), born charges, timing — handle both collinear and non-collinear spin
- 127.2.4.4 Implement DOSCAR parser: read total DOS (energy, DOS-up,
DOS-down, integrated-DOS-up, integrated-DOS-down) and projected DOS (per
atom, per orbital: s, p_y, p_z, p_x, d_xy, d_yz, d_z2, d_xz, d_x2-y2, and
f orbitals if LORBIT=11) — return structured
DensityOfStatesobject with energy grid, total, and projected components - 127.2.4.5 Implement EIGENVAL parser: read k-point coordinates,
weights, and eigenvalues (per k-point, per band, per spin) — return
BandStructureobject with k-path, eigenvalues, Fermi energy, band gap detection (direct/indirect, VBM/CBM k-points) - 127.2.4.6 Implement PROCAR parser: read projected band structure — orbital weights (s, p, d, f decomposition) and atom decomposition at each k-point and band — for fat band and orbital-character analysis
- 127.2.4.7 Implement CHGCAR/AECCAR reader: parse charge density on FFT grid — header (structure, grid dimensions), volumetric data (spin-up, spin-down for spin-polarized), augmentation charges — support CHGCAR, AECCAR0 (core), AECCAR2 (valence) for Bader analysis preparation
- 127.2.4.8 Implement LOCPOT parser: read local potential on FFT grid — same format as CHGCAR but containing the electrostatic potential — for work function calculations (vacuum level extraction) and dipole corrections
- 127.2.4.9 Implement WAVECAR reader (Rust + WASM): parse VASP WAVECAR binary format — extract header (ISPIN, NKPTS, NBANDS, ENCUT, lattice), plane-wave coefficients for each k-point and band — for Wannier90 input (.amn/.mmn computation) and wavefunction visualization. Performance-critical: use Rust for binary parsing of potentially multi-GB files
- 127.2.4.10 Implement XDATCAR parser: read MD trajectory — sequence of structures at each ionic step in direct coordinates — return array of KalikaStructure with step indices
- 127.2.4.11 Implement OSZICAR parser: read SCF convergence per ionic step — electronic iteration energies (dE, mag), ionic step energies and forces — for real-time convergence monitoring
- 127.2.4.12 Implement ELFCAR parser: read electron localization function on FFT grid — same volumetric format as CHGCAR but values in [0,1] range — for chemical bonding analysis
- 127.2.4.13 Implement IBZKPT parser: read irreducible k-points with weights after symmetry reduction — for verifying k-point meshes and symmetry-related debugging
- 127.2.4.14 Implement ML_LOGFILE parser: parse VASP 6.3+ ML force field training log — extract training errors (energy, force, stress RMSE), Bayesian error estimates, ab initio vs. ML calls, training set size growth — for monitoring on-the-fly ML potential quality
- 127.2.4.15 Implement vaspout.h5 reader: parse VASP 6.4+ HDF5 output format using an HDF5 reading library — extract all quantities (energies, forces, eigenvalues, DOS, charge density) from the structured HDF5 hierarchy — the future-proof VASP output format
127.2.5 Error Handling & Custodian-Matching Fixes#
- 127.2.5.1 Implement VASP error detector: scan OUTCAR and stdout for known error signatures — ZBRENT (bracketing failure in line minimization), EDDDAV (sub-space rotation error), ZPOTRF (Cholesky decomposition failure), SGRCON (symmetry error), PRICEL (primitive cell issue), BRMIX (Kerker mixing failure), DENTET (tetrahedron integration with too few k-points), RHOSYG (charge density symmetrization error), POSMAP (atom mapping error), TOO_FEW_BANDS, EDDRMM (RMM-DIIS not converging)
- 127.2.5.2 Implement automatic fix suggestions matching custodian VaspErrorHandler: ZBRENT → reduce POTIM by 50%, EDDDAV → switch ALGO to All, ZPOTRF → reduce POTIM or switch to CG (IBRION=2), BRMIX → set IMIX=1/AMIX=0.1/BMIX=0.001, DENTET → switch to ISMEAR=0, TOO_FEW_BANDS → increase NBANDS by 25%, EDDRMM → switch ALGO to Normal, PRICEL → set SYMPREC=1e-8
- 127.2.5.3 Implement SCF convergence monitor: track energy difference per electronic step, detect oscillation (sign alternation in dE), detect divergence (|dE| increasing), detect slow convergence (|dE| decreasing too slowly) — suggest ALGO/AMIX/AMIN/BMIX adjustments
- 127.2.5.4 Implement ionic convergence monitor: track max force, stress components, energy per ionic step — detect: not converging (forces increasing), oscillating, converged-but-not-global-minimum (energy plateau with large forces) — suggest IBRION/POTIM adjustments
- 127.2.5.5 Implement VASP version compatibility layer: detect VASP version from OUTCAR/vasprun.xml header, apply version-specific tag aliases and defaults (e.g., ML_LMLFF in 6.3+, LHARTREE in 6.4+), warn about deprecated tags, select appropriate output parser variant
127.3 Quantum ESPRESSO Interface (@kalika/qe-io)#
127.3.1 pw.x Input Generation#
- 127.3.1.1 Implement QE input file writer: generate pw.x input from
typed
QeInputobject — format Fortran namelist blocks (&CONTROL, &SYSTEM, &ELECTRONS, &IONS, &CELL) with proper quoting and comma separation, followed by card blocks (ATOMIC_SPECIES, ATOMIC_POSITIONS, K_POINTS, CELL_PARAMETERS, CONSTRAINTS, OCCUPATIONS, ATOMIC_FORCES) - 127.3.1.2 Implement &CONTROL block generation:
calculationtype (scf/relax/vc-relax/md/vc-md),prefix,outdir,pseudo_dir,restart_mode,tprnfor,tstress,etot_conv_thr,forc_conv_thr,nstep,disk_io,verbosity— with validation of allowed values - 127.3.1.3 Implement &SYSTEM block generation:
ibrav(0-14 Bravais lattice index) with celldm(1-6) or A/B/C/cosAB/cosAC/cosBC,nat,ntyp,ecutwfc,ecutrho(auto-set from ecutwfc × 4 for NC, ×8-12 for US/PAW),occupations(smearing/tetrahedra/fixed),smearing(gaussian/mv/mp/cold),degauss,nspin(1/2/4),starting_magnetization,noncolin,lspinorb,nbnd,tot_charge,input_dft(functional override),vdw_corr(grimme-d2/d3/ts/mbd),hubbard_u(DFT+U) - 127.3.1.4 Implement &ELECTRONS block generation:
conv_thr,mixing_mode,mixing_beta,mixing_ndim,diagonalization(david/cg/ppcg),electron_maxstep,startingpot,startingwfc - 127.3.1.5 Implement &IONS block generation:
ion_dynamics(bfgs/damp/verlet/langevin/...),upscale,trust_radius_min,trust_radius_max,bfgs_ndim - 127.3.1.6 Implement &CELL block generation:
cell_dynamics(bfgs/damp-w/...),press,press_conv_thr,cell_dofree(all/x/y/z/xy/xz/yz/xyz/shape/volume/2Dxy/2Dshape) - 127.3.1.7 Implement ATOMIC_SPECIES card writer: element symbol, atomic mass, pseudopotential filename — with automatic pseudopotential selection from configured library
- 127.3.1.8 Implement ATOMIC_POSITIONS card writer: support crystal (fractional), angstrom, bohr, alat, crystal_sg (space group Wyckoff positions) — with optional per-atom if_pos constraints (0/1 for each direction, QE equivalent of VASP selective dynamics)
- 127.3.1.9 Implement K_POINTS card writer: automatic (nk1 nk2 nk3 sk1 sk2 sk3), gamma, tpiba, crystal, tpiba_b (band structure), and crystal_b (band structure in crystal coords) — with auto-generation from target density
- 127.3.1.10 Implement CELL_PARAMETERS card writer: when ibrav=0, write lattice vectors in angstrom, bohr, or alat units
127.3.2 QE Presets & Pseudopotential Management#
- 127.3.2.1 Implement QE presets:
scf(standard SCF),relax(ionic relaxation),vc-relax(full cell relaxation),band_structure(NSCF + bands.x),dos(NSCF + dos.x + projwfc.x),phonon(SCF → ph.x DFPT),md_nvt(Car-Parrinello or Born-Oppenheimer MD) — each preset sets all required namelists and cards with sensible defaults - 127.3.2.2 Implement pseudopotential library manager: support SSSP (Standard Solid-State Pseudopotentials) efficiency and precision libraries, PseudoDojo NC/US/PAW — for each element, store recommended pseudopotential filename, ecutwfc, ecutrho, and dual (rho/wfc ratio) — auto-select based on target accuracy
- 127.3.2.3 Implement pseudopotential validation: verify UPF file header (PP_HEADER), extract Z_valence, mesh_size, functional, relativistic flag, is_ultrasoft/is_paw — ensure consistency with &SYSTEM settings
127.3.3 QE Output Parsing#
- 127.3.3.1 Implement pw.x output parser: extract total energy (Ry), forces (Ry/Bohr), stress (kbar), eigenvalues, Fermi energy, magnetization, SCF convergence history, timing, number of k-points, number of plane waves, symmetry operations detected — handle both terminated and crashed runs
- 127.3.3.2 Implement QE XML parser (data-file-schema.xml): parse QE 6.x+ XML output — extract: input parameters, atomic structure (initial + final), band structure (eigenvalues at all k-points), total energy, forces, stress, timing, convergence info — this is the robust machine-readable output format
- 127.3.3.3 Implement ph.x output parser: extract dynamical matrices at each q-point, phonon frequencies (cm⁻¹ and THz), Born effective charges (Z*), macroscopic dielectric constant (ε∞), mode symmetry labels, Raman tensors (if calculated), IR intensities
- 127.3.3.4 Implement matdyn.x output parser: parse phonon dispersion
(frequencies along q-path) and phonon DOS from matdyn.x output — return
PhononBandStructureandPhononDOSobjects - 127.3.3.5 Implement pp.x output parser: read charge density, potential, ELF in cube format and XSF (XCrySDen) format — volumetric data on real-space grid for visualization
127.3.4 Additional QE Tool Interfaces#
- 127.3.4.1 Implement bands.x interface: generate input for band structure post-processing (filband, lsym), parse output (bands.dat, bands.dat.gnu) for plotting
- 127.3.4.2 Implement dos.x interface: generate input (fildos, Emin, Emax, DeltaE, ngauss, degauss), parse output DOS file — total DOS
- 127.3.4.3 Implement projwfc.x interface: generate input (filpdos, Emin, Emax, DeltaE, ngauss, degauss), parse projected DOS output — per-atom, per-orbital PDOS files (.pdos_atm#_wfc#*)
- 127.3.4.4 Implement epsilon.x interface: generate input for optical properties (interband transitions), parse dielectric function output (real and imaginary parts vs. energy)
- 127.3.4.5 Implement hp.x interface: generate input for linear-response Hubbard U/V calculation (nq1-3, perturb_only_atom, compute_hp), parse output U and V values — for self-consistent DFT+U
127.4 ABINIT Interface (@kalika/abinit-io)#
127.4.1 Input Generation#
- 127.4.1.1 Implement ABINIT input file writer: generate ABINIT input format (variable = value lines, multi-dataset syntax with jdtset/ndtset and variable? suffixes) — handle all value types: scalars, arrays, 2D arrays (rprim, xred), string-valued variables
- 127.4.1.2 Implement core variable generation:
acell,rprim(lattice),ntypat,znucl,natom,typat,xred/xcart/xangst(coordinates),ecut(PW cutoff),ngkpt+nshiftk+shiftk(k-points),tolvrs/toldfe/toldff(convergence),nstep,occopt+tsmear,nsppol/nspinor/nspden(spin),ixc(XC functional),pawecutdg(PAW augmentation cutoff) - 127.4.1.3 Implement multi-dataset generation: create multi-dataset
ABINIT inputs for chained calculations (e.g., dataset 1: SCF, dataset 2:
NSCF band structure, dataset 3: DOS) — with proper
getden/irdden/irdwfkchaining between datasets - 127.4.1.4 Implement ABINIT presets:
ground_state(standard SCF),relaxation(ionmov=2/22, optcell=0/2),band_structure(iscf=-2, kptopt=-N for line mode),dos(prtdos=1/2/3),dfpt_phonon(rfphon=1, rfatpol, rfdir),gw(optdriver=3/4, gwcalctyp),bse(optdriver=99, bs_loband, bs_nstates)
127.4.2 Output Parsing#
- 127.4.2.1 Implement ABINIT main output parser: extract total energy (Ha), forces (Ha/Bohr), stress (Ha/Bohr³), eigenvalues, Fermi energy, magnetization, SCF convergence, timing, symmetry info, warnings/errors — handling multi-dataset outputs
- 127.4.2.2 Implement ABINIT NetCDF output reader: parse _DEN.nc (density), _WFK.nc (wavefunctions), _DDB (derivative database for phonons), _EIG.nc (eigenvalues) — using a NetCDF reading library
- 127.4.2.3 Implement ABINIT DDB file parser: parse derivative database (dynamical matrices, Born effective charges, dielectric tensor) — for Anaddb post-processing (phonon band structure, thermodynamic properties, Grüneisen parameters)
- 127.4.2.4 Implement ABINIT GW output parser: extract quasiparticle energies (E_KS, E_QP, Z renormalization factors), self-energy Σ(ω) on real/imaginary frequency axis, spectral function
127.5 FHI-aims Interface (@kalika/fhi-aims-io)#
127.5.1 Input Generation#
- 127.5.1.1 Implement geometry.in writer: write FHI-aims geometry file
from KalikaStructure —
lattice_vectorlines (3 lines for periodic),atomoratom_fraclines (Cartesian or fractional), optionalvelocity,initial_moment,constrain_relaxationper atom - 127.5.1.2 Implement control.in writer: generate FHI-aims control file
—
xc(pbe/hse06/pbe0/lda/...),relativistic(atomic_zora/zora/none),spin(collinear/none),k_grid(n1 n2 n3),sc_accuracy_rho,sc_accuracy_eev,sc_accuracy_etot,sc_accuracy_forces,relax_geometry(bfgs/trm),relax_unit_cell(full/shape/none),compute_forces,output bandlines,output dossettings — followed by species-specific basis set blocks - 127.5.1.3 Implement FHI-aims basis set selection: for each element, append the appropriate species_defaults block (light, intermediate, tight, really_tight) from the FHI-aims species_defaults directory — NAO (numeric atomic orbital) basis functions with radial functions, angular momentum channels, and basis set superposition
- 127.5.1.4 Implement FHI-aims presets:
scf(standard ground state),relax(geometry optimization with BFGS),band_structure(with output band directives along high-symmetry path),dos(with output dos settings),hybrid(HSE06/PBE0 with RI-V acceleration)
127.5.2 Output Parsing#
- 127.5.2.1 Implement aims.out parser: extract total energy (eV), forces, stress, eigenvalues, Fermi energy, band gap, timing, SCF convergence history, Mulliken charges, Hirshfeld charges — handle periodic and molecular outputs
- 127.5.2.2 Implement FHI-aims band structure parser: read band*.out
files (k-point, eigenvalues per band, occupation) — construct
BandStructureobject with k-path labels
127.6 Other DFT Code Interfaces (@kalika/code-interfaces)#
127.6.1 SIESTA Interface#
- 127.6.1.1 Implement SIESTA .fdf input writer: generate Flexible Data Format input — SystemName, SystemLabel, NumberOfAtoms, NumberOfSpecies, PAO.BasisSize (SZ/SZP/DZ/DZP/TZP), MeshCutoff, DM.MixingWeight, DM.Tolerance, MaxSCFIterations, MD.TypeOfRun, MD.NumCGsteps — plus block entries: %block LatticeVectors, %block AtomicCoordinatesAndAtomicSpecies, %block kgrid_Monkhorst_Pack, %block ChemicalSpeciesLabel
- 127.6.1.2 Implement SIESTA output parser: parse .out file for total energy, forces, stress, Mulliken populations, eigenvalues, band structure (SystemLabel.bands), DOS (SystemLabel.DOS) — plus .XV (optimized coordinates), .ANI (trajectory)
- 127.6.1.3 Implement SIESTA pseudopotential handling: .psf (Troullier- Martins) and .psml (PSML standard) format detection and validation
127.6.2 CP2K Interface#
- 127.6.2.1 Implement CP2K .inp writer: generate CP2K input in hierarchical section format (&GLOBAL, &FORCE_EVAL{&DFT{&QS, &SCF, &XC, &MGRID, &POISSON}, &SUBSYS{&CELL, &COORD, &KIND}}, &MOTION{&GEO_OPT, &MD, &CELL_OPT}) — proper indentation and section nesting
- 127.6.2.2 Implement CP2K output parser: extract total energy, forces, stress, SCF convergence, eigenvalues, Mulliken charges, band structure, MD trajectory — handle GPW (Gaussian and Plane Waves) and GAPW outputs
- 127.6.2.3 Implement CP2K basis set and GTH pseudopotential selection: map elements to appropriate MOLOPT/ccGRB basis sets and GTH-PBE/GTH-BLYP pseudopotentials
127.6.3 GPAW Interface#
- 127.6.3.1 Implement GPAW Python script generator: generate a Python
script that imports
gpaw, createsGPAW(mode=PW(ecut), xc='PBE', kpts=..., txt='output.txt'), attaches to ASE Atoms, runscalc.get_potential_energy()— for SCF, relaxation, band structure, DOS workflows - 127.6.3.2 Implement GPAW output parser: parse gpaw .txt output and .gpw binary restart file — extract energy, forces, eigenvalues, DOS
127.6.4 WIEN2k Interface#
- 127.6.4.1 Implement WIEN2k struct file writer: generate case.struct with lattice type, lattice parameters, atom positions (in internal coordinates), RMT values (muffin-tin radii), R0 (first radial mesh point), NPT (number of radial mesh points), ISPLIT — the master input file for all-electron LAPW calculations
- 127.6.4.2 Implement WIEN2k case.in* file generators: case.in1 (LAPW1: K-VECTORS, Emin/Emax), case.in2 (LAPW2: Fermi energy, partial charges), case.inm (MIXER: mixing parameters)
- 127.6.4.3 Implement WIEN2k output parser: parse case.scf (SCF summary with energy, charge, moment per iteration), case.energy (eigenvalues), case.dos* (DOS)
127.6.5 Other Code Stubs#
- 127.6.5.1 Implement Elk elk.in writer: key blocks — tasks, avec (lattice), atoms (species + positions), ngridk, rgkmax, swidth — and output parser for TOTENERGY.OUT, EIGVAL.OUT, TDOS.OUT
- 127.6.5.2 Implement CASTEP .cell/.param writer: .cell (lattice_cart, positions_frac, kpoint_mp_grid, symmetry_ops, species_pot), .param (task, xc_functional, cut_off_energy, elec_energy_tol, max_scf_cycles) — and .castep output parser
- 127.6.5.3 Implement FLEUR inp.xml writer: generate XML-based input (lattice, atomGroups, xcFunctional, kPointMesh, cutoffs) — and output parser for out.xml
- 127.6.5.4 Implement CRYSTAL .d12 writer: geometry (dimensionality, space group, lattice, atoms), basis set (all-electron or ECP), DFT parameters — and output parser for total energy, forces, band structure
- 127.6.5.5 Implement Gaussian .gjf/.com writer: route section (#p method/basis), charge/multiplicity, geometry, additional keywords — and .log + .fchk output parser (energy, forces, frequencies, molecular orbitals)
- 127.6.5.6 Implement ORCA .inp writer: method block (! DFT PBE def2-SVP), %pal (parallelism), %scf (convergence), coordinate block — and output parser for energy, forces, frequencies, NMR, EPR parameters
127.7 LAMMPS Interface (@kalika/lammps-io)#
127.7.1 Data File I/O#
- 127.7.1.1 Implement LAMMPS data file writer: generate atom data in full/atomic/charge/molecular atom styles — header (atom count, bond count, atom types, box bounds), Masses section, Atoms section (id, mol-id, type, charge, x, y, z), Velocities section, plus optional Bonds/Angles/Dihedrals/Impropers sections for molecular systems
- 127.7.1.2 Implement LAMMPS data file reader: parse all sections of a LAMMPS data file — atoms, velocities, bonds, angles, masses, box dimensions (orthogonal and triclinic with tilt factors xy, xz, yz) — return KalikaStructure with topology information
127.7.2 Input Script Generation#
- 127.7.2.1 Implement LAMMPS input script builder: fluent API for
constructing LAMMPS input scripts —
setUnits('metal'),setAtomStyle('atomic'),setBoundary('p p p'),readData('data.lmp'),setPairStyle('eam/alloy'),setPairCoeff('* * pot.eam.alloy Al Cu'),addFix('1 all nvt temp 300 300 0.1'),addCompute(...),setTimestep(0.001),addDump(...),setThermoStyle(...),setThermoFreq(100),run(10000)— serialize to valid LAMMPS script - 127.7.2.2 Implement pair style configuration helpers: generate
pair_style and pair_coeff commands for common potentials —
eam/alloy(single file, all elements),eam/fs(Finnis-Sinclair),tersoff(Si, C, BN),reaxff(with control file),lj/cut(with per-type ε, σ),buck(Buckingham),snap(SNAP ML potential),pace(ACE ML potential),meam— with automatic unit conversion - 127.7.2.3 Implement fix command helpers: generate fix commands for
standard ensembles —
nvt(Nosé-Hoover thermostat),npt(barostat + thermostat with iso/aniso/tri options),nve(microcanonical),langevin(Langevin thermostat),neb(nudged elastic band),shake(bond constraints),deform(box deformation for strain),wall/reflect(walls) - 127.7.2.4 Implement compute command helpers:
pe/atom,stress/atom,rdf(radial distribution function with cutoff and bins),msd(mean square displacement),centro/atom(centrosymmetry parameter),cna/atom(common neighbor analysis),voronoi/atom,heat/flux(for Green-Kubo thermal conductivity) - 127.7.2.5 Implement dump command helpers:
atom(id, type, x, y, z),custom(configurable columns: id, type, x, y, z, vx, vy, vz, fx, fy, fz, c_pe, c_stress[1-6]),cfg(AtomEye),netcdf(NetCDF format) — with dump_modify (element names, sort, pad) - 127.7.2.6 Implement LAMMPS presets:
energy_minimization(min_style cg, minimize etol ftol maxiter maxeval),nvt_equilibration(fix nvt + thermo + dump),npt_equilibration(fix npt with iso/aniso barostat),tensile_test(fix deform + compute stress/atom + dump),thermal_conductivity(NEMD or Green-Kubo setup),neb(neb command + replica setup)
127.7.3 Output Parsing#
- 127.7.3.1 Implement LAMMPS log file parser: extract thermodynamic data columns (Step, Temp, Press, PotEng, KinEng, TotEng, Volume, Density, Lx/Ly/Lz, custom computes) from each run — handle multiple run blocks, minimization output, and LAMMPS-style column headers
- 127.7.3.2 Implement LAMMPS dump file parser (Rust + WASM): read custom dump format (ITEM: TIMESTEP, ITEM: NUMBER OF ATOMS, ITEM: BOX BOUNDS, ITEM: ATOMS) — multi-frame trajectory support, lazy loading for large files, support both orthogonal and triclinic boxes. Rust for performance on multi-GB trajectory files
- 127.7.3.3 Implement LAMMPS restart file reader: read binary restart format — extract atom positions, velocities, box dimensions, force field parameters — for analysis and restart continuation
127.7.4 Potential File Management#
- 127.7.4.1 Implement LAMMPS potential file registry: manage local potential file library (EAM: .eam.alloy/.eam.fs, Tersoff: .tersoff, ReaxFF: ffield.reax, SNAP: .snapparam+.snapcoeff, ACE: .yace) — index by element combination, potential type, source publication
- 127.7.4.2 Implement OpenKIM query interface: query the OpenKIM API
(query.openkim.org) for available interatomic potentials by element
combination, model type, property tested — retrieve KIM model IDs for use
with
kim_interactionscommand in LAMMPS
127.8 GROMACS Interface (@kalika/gromacs-io)#
127.8.1 File I/O#
- 127.8.1.1 Implement .gro coordinate file reader/writer: GROMACS format (title, atom count, residue number + name + atom name + atom number + coordinates in nm + optional velocities in nm/ps, box vectors) — convert to/from KalikaStructure
- 127.8.1.2 Implement .top topology file writer: generate GROMACS
topology —
[defaults](nbfunc, comb-rule, gen-pairs, fudgeLJ, fudgeQQ),[atomtypes],[moleculetype],[atoms](nr, type, resnr, residue, atom, cgnr, charge, mass),[bonds],[pairs],[angles],[dihedrals](proper + improper),[system],[molecules]— with #include for force field itp files - 127.8.1.3 Implement .mdp parameter file generator: generate run
parameters for:
em(energy minimization: integrator=steep/cg, emtol, emstep, nsteps),nvt(integrator=md, dt, nsteps, tcoupl= v-rescale, ref_t),npt(pcoupl=parrinello-rahman/berendsen, ref_p, tau_p),production(long MD run parameters) — with proper neighbor searching, cutoff, PME, constraint settings - 127.8.1.4 Implement .ndx index group file reader/writer: parse/generate GROMACS index files with named atom groups (System, Protein, DNA, Water, SOL, custom groups)
127.8.2 Trajectory & Energy Parsing#
- 127.8.2.1 Implement .xtc trajectory reader (Rust + WASM): read compressed GROMACS trajectory format (XDR-encoded, lossy compression) — extract frame timestamps, box vectors, atom coordinates — lazy frame iteration for multi-GB trajectories. Rust for XDR decoding performance
- 127.8.2.2 Implement .trr trajectory reader: read full-precision GROMACS trajectory (coordinates, velocities, forces, box, lambda) — XDR format parsing
- 127.8.2.3 Implement .edr energy file reader: parse GROMACS portable energy format — extract all energy terms (Potential, Kinetic, Total, Temperature, Pressure, Volume, Density, individual LJ/Coulomb contributions) vs. time step
- 127.8.2.4 Implement force field selection helper: map force field names (amber99sb-ildn, amber14sb, charmm36m, charmm27, oplsaa, gromos54a7) to appropriate GROMACS include paths and default settings
127.9 Phonopy / Phono3py Interface (@kalika/phonopy-io)#
127.9.1 Phonopy File I/O#
- 127.9.1.1 Implement phonopy.yaml reader/writer: parse the master Phonopy YAML configuration (supercell_matrix, primitive_matrix, atom types, positions, displacements) — including phonopy_disp.yaml (supercells with displacements)
- 127.9.1.2 Implement FORCE_SETS file reader/writer: read/write Phonopy force sets format (number of atoms, number of displacements, for each: atom index, displacement vector, forces on all atoms) — the bridge between DFT force calculations and phonon analysis
- 127.9.1.3 Implement BORN file reader/writer: read/write Born effective charges and dielectric tensor — format: dielectric tensor (3×3), then Born charges (3×3 per atom) — needed for LO-TO splitting correction in polar materials
- 127.9.1.4 Implement band.yaml reader: parse phonon band structure
output — q-points along path, frequencies (THz), eigenvectors (complex,
per atom per mode), labels at high-symmetry points — return
PhononBandStructureobject - 127.9.1.5 Implement mesh.yaml reader: parse phonon DOS mesh output — q-points on mesh, frequencies, weights — for phonon density of states calculation
- 127.9.1.6 Implement thermal_properties.yaml reader: parse thermodynamic properties — temperature array, free energy (kJ/mol), entropy (J/K/mol), heat capacity Cv (J/K/mol) — from the quasiharmonic phonon partition function
127.9.2 Phonopy Workflow Integration#
- 127.9.2.1 Implement supercell displacement generator: given a KalikaStructure and supercell_matrix, generate all symmetrically distinct displaced supercells (reproducing Phonopy's displacement generation algorithm — apply symmetry to reduce displacement count) — output as array of KalikaStructure ready for DFT calculations
- 127.9.2.2 Implement force constant calculator: given FORCE_SETS (displacements + DFT forces), compute harmonic force constants Φ(R) using the finite displacement method — matrix equation relating displacements to forces via crystal symmetry constraints
- 127.9.2.3 Implement phonon property calculator: given force constants, compute phonon frequencies and eigenvectors at arbitrary q-points via dynamical matrix D(q) = Σ_R Φ(R) exp(iq·R) / √(M_i M_j) — including non-analytic correction for polar materials (Born charges + dielectric tensor → LO-TO splitting)
- 127.9.2.4 Implement phonon DOS calculator: compute phonon density of states by sampling frequencies on a dense q-mesh with tetrahedron integration — total DOS and projected DOS (per atom contribution)
- 127.9.2.5 Implement phonon thermodynamics: from phonon DOS, compute temperature-dependent free energy F(T), entropy S(T), heat capacity C_v(T), zero-point energy via the phonon partition function Z = Π_qν [2 sinh(ℏω/2k_BT)]⁻¹
127.9.3 Phono3py Interface#
- 127.9.3.1 Implement Phono3py FORCE_CONSTANTS_3RD reader/writer: read/write third-order force constants fc3(R1,R2) — the anharmonic force constants needed for lattice thermal conductivity
- 127.9.3.2 Implement Phono3py thermal conductivity output parser: read kappa-m*.hdf5 or kappa-m*.dat — lattice thermal conductivity tensor κ_L(T) vs. temperature, mode-resolved κ contributions, phonon lifetimes τ(q,ν), group velocities v_g(q,ν), Grüneisen parameters
127.10 WANNIER90 Interface (@kalika/wannier90-io)#
127.10.1 Input Generation#
- 127.10.1.1 Implement wannier90.win writer: generate Wannier90 input
file —
num_wann,num_bands,unit_cell_cart(lattice),atoms_frac/atoms_cart,mp_grid(k-mesh),begin kpointsblock,begin projectionsblock — with all major keywords:dis_win_min/max,dis_froz_min/max(disentanglement windows),num_iter,conv_tol,conv_window,guiding_centres,write_hr,bands_plot,bands_num_points,begin kpoint_pathblock - 127.10.1.2 Implement projection block generator: generate
begin projectionsfrom high-level specification — atom-centered projections (e.g.,Fe: d,O: p,C: sp3), support all angular momentum channels (s, p, d, f), hybrid orbitals (sp, sp2, sp3, sp3d, sp3d2), custom radial functions, random projections — with automaticl,mr,rquantum number assignment - 127.10.1.3 Implement disentanglement window auto-setter: given
eigenvalues from DFT, automatically determine optimal
dis_win_min,dis_win_max(outer window encompassing target bands),dis_froz_min,dis_froz_max(frozen window around Fermi level or band gap) — heuristic based on band character and energy range
127.10.2 File I/O#
- 127.10.2.1 Implement .amn file reader/writer: Wannier90 projection matrix A_{mn}(k) — for each k-point, the num_bands × num_wann complex matrix of initial projections — format: header (num_bands, num_kpts, num_wann), then (band, wann, kpt, Re, Im) entries
- 127.10.2.2 Implement .mmn file reader/writer: overlap matrix M_{mn}(k,b) — for each k-point and each nearest-neighbor b-vector, the num_bands × num_bands complex overlap matrix — format: header, then (kpt, kpt+b, bx, by, bz) followed by matrix entries
- 127.10.2.3 Implement .eig file reader/writer: eigenvalues ε_{nk} — for each k-point and band, the DFT eigenvalue (eV) — simple three-column format: (band, kpt, eigenvalue)
- 127.10.2.4 Implement .wout output parser: parse Wannier90 output — extract: initial/final spread (Ω, Ω_I, Ω_D, Ω_OD), Wannier function centers and spreads (per WF), convergence history, disentanglement convergence (if applicable), symmetry-adapted WF info
127.10.3 Wannier Output & Applications#
- 127.10.3.1 Implement _hr.dat reader: parse Wannier tight-binding Hamiltonian H(R) — degeneracy weights, then (Rx, Ry, Rz, i, j, Re_H, Im_H) entries — for Wannier-interpolated band structure and transport
- 127.10.3.2 Implement Wannier-interpolated band structure calculator: given H(R) from _hr.dat, compute eigenvalues at arbitrary k-points via H(k) = Σ_R H(R) exp(ik·R) — much denser k-mesh than DFT for smooth band structure and accurate Fermi surface
- 127.10.3.3 Implement _band.dat / _band.kpt reader: parse Wannier90 band structure output (k-point path coordinates + eigenvalues) for comparison with DFT bands
- 127.10.3.4 Implement DFT→Wannier90 workflow connector: for VASP, generate INCAR tags (LWANNIER90=.TRUE. or manual .amn/.mmn via LWRITE_MMN_AMN); for QE, generate pw2wannier90.x input (outdir, prefix, seedname, write_amn, write_mmn, write_unk) — the glue between DFT and Wannier90
- 127.10.3.5 Implement BoltzWann transport interface: generate BoltzWann input block in wannier90.win (boltzwann=.TRUE., boltz_calc_also_dos, boltz_mu_min/max, boltz_temp_min/max/step, boltz_relax_time) — parse BoltzWann output (σ, S, κ_e, PF vs. T, μ)
127.11 LOBSTER Interface (@kalika/lobster-io)#
127.11.1 Input Generation#
- 127.11.1.1 Implement lobsterin writer: generate LOBSTER input file —
COHPstartEnergy,COHPendEnergy,basisSet(pbeVaspFit2015 / basisSetMappingFile),gaussianSmearingWidth,userecommendedBasisSet,cohpGenerator(from/to atom lists),saveProjectionToFile,loadProjectionFromFile,basisfunctions(per-element orbital specifications: e.g.,Fe 3d 4s 4p) - 127.11.1.2 Implement automatic orbital selection: given a structure and list of atom pairs of interest, automatically select appropriate basis functions for LOBSTER analysis (valence orbitals, bonding-relevant orbitals) following LOBSTER recommendations per element
127.11.2 Output Parsing#
- 127.11.2.1 Implement COHPCAR.lobster parser: read Crystal Orbital
Hamilton Population data — energy grid, COHP(E) and integrated COHP
(ICOHP) for each atom pair — per-spin if spin-polarized, per-orbital
decomposition if requested. Return structured
CohpDatawith atom pair labels, distances, and COHP/ICOHP arrays - 127.11.2.2 Implement COOPCAR.lobster parser: read Crystal Orbital
Overlap Population — same format as COHPCAR but for overlap population —
return
CoopData - 127.11.2.3 Implement ICOHPLIST.lobster parser: extract integrated COHP values (bond strengths) at Fermi level for all atom pairs — sorted by ICOHP value, with distances, labels, and translation vectors
- 127.11.2.4 Implement CHARGE.lobster parser: read Löwdin and Mulliken charges per atom — return structured charge data
- 127.11.2.5 Implement GROSSPOP.lobster parser: read gross orbital populations per atom per orbital — for detailed electronic structure analysis
- 127.11.2.6 Implement COBICAR.lobster parser: read Crystal Orbital Bond Index (COBI) data — the bond-order analogue of COHP/COOP — parse energy-resolved and integrated COBI per atom pair
127.12 BoltzTraP2 Interface (@kalika/boltztrap-io)#
127.12.1 Input Preparation#
- 127.12.1.1 Implement BoltzTraP2 input generator: from DFT band structure (eigenvalues on dense k-grid from VASP EIGENVAL or QE XML), generate BoltzTraP2 input — structure (lattice, atoms), eigenvalues ε_{nk} at all irreducible k-points, k-point weights, Fermi energy — in BoltzTraP2's expected format (or via Python bridge to btp2 API)
127.12.2 Output Parsing#
- 127.12.2.1 Implement BoltzTraP2 transport coefficient parser: extract
σ/τ (electrical conductivity / relaxation time), S (Seebeck coefficient),
κ_e/τ (electronic thermal conductivity / relaxation time), R_H (Hall
coefficient) — all as functions of temperature T and chemical potential μ
— return structured
TransportDatawith interpolation capabilities - 127.12.2.2 Implement thermoelectric figure of merit calculator: given BoltzTraP2 σ/τ, S, κ_e/τ, plus lattice thermal conductivity κ_L from Phono3py, and an assumed relaxation time τ — compute power factor PF = S²σ, and ZT = S²σT/(κ_e + κ_L) as function of T and carrier concentration — the key materials screening metric for thermoelectrics
127.13 Ecosystem Compatibility (@kalika/code-interfaces)#
127.13.1 ASE Compatibility#
- 127.13.1.1 Implement ASE Atoms ↔ KalikaStructure bidirectional converter: map ASE Atoms (cell, positions, numbers, pbc, tags, momenta, initial_magnetic_moments, initial_charges, constraints) to/from KalikaStructure — preserve all site properties, handle FixAtoms and FixedLine/FixedPlane constraints ↔ selective dynamics
- 127.13.1.2 Implement ASE Calculator adapter: wrap Kalika's code
interfaces as an ASE-compatible Calculator (Python class via bridge) —
implement
calculate(),get_potential_energy(),get_forces(),get_stress()— enabling ASE workflows (optimizers, MD, NEB) to use Kalika-managed DFT codes as the energy/force engine - 127.13.1.3 Implement ASE trajectory reader/writer: read/write ASE's .traj format (Python pickle-based) and ASE's JSON trajectory format — for trajectory exchange between ASE and Kalika workflows
- 127.13.1.4 Implement ASE database adapter: read/write ASE's SQLite database format (.db) — structure, energy, forces, stress, key-value pairs, data blobs — enabling Kalika to query and contribute to ASE databases
127.13.2 pymatgen Compatibility#
- 127.13.2.1 Implement pymatgen Structure/Molecule ↔ KalikaStructure converter: map pymatgen Structure (lattice, species, frac_coords, site_properties) and Molecule (species, cart_coords) to/from KalikaStructure — handle oxidation states, spin, selective dynamics, Magmom objects
- 127.13.2.2 Implement pymatgen DOS/BandStructure ↔ Kalika converter: convert pymatgen CompleteDos, Dos, BandStructureSymmLine objects to/from Kalika's DOS and BandStructure types — preserving orbital projections, spin channels, k-point labels
- 127.13.2.3 Implement pymatgen PhaseDiagram compatibility: generate
pymatgen-compatible
PDEntryobjects from Kalika calculation results (composition + energy) for phase stability analysis — or consume pymatgen PhaseDiagram results (stable entries, e_above_hull) - 127.13.2.4 Implement pymatgen InputSet/OutputParser adapter: bridge Kalika's VASP/QE input generators to pymatgen's VaspInputSet / QEInputSet interface, and Kalika's output parsers to pymatgen's Vasprun / Outcar classes — for code that expects pymatgen objects
127.13.3 Workflow Framework Compatibility#
- 127.13.3.1 Implement AiiDA compatibility layer: convert Kalika WorkflowDAG to AiiDA WorkChain/CalcJob representation — map CalculationInput to AiiDA CalcJob inputs, CalculationOutput to AiiDA CalcJob outputs, WorkflowStep dependencies to AiiDA WorkChain logic — enable Kalika workflows to run on AiiDA infrastructure
- 127.13.3.2 Implement FireWorks/Jobflow compatibility: convert Kalika WorkflowDAG to Jobflow Flow/Job representation (or FireWorks Workflow/ Firework/FireTask) — mapping step dependencies to job dependencies, output passing to data stores — for submission to FireWorks LaunchPad
- 127.13.3.3 Implement custodian error handler registry: register
Kalika's VASP/QE error handlers in a format compatible with custodian's
ErrorHandler interface —
check()(detect error) andcorrect()(apply fix) methods — so Kalika workflows get custodian-level robustness - 127.13.3.4 Implement atomate2 workflow recipe matching: for key atomate2 workflows (StaticMaker, RelaxMaker, BandStructureMaker, ElasticMaker, PhononMaker, DefectMaker), create equivalent Kalika WorkflowDAG templates that produce compatible output schemas
127.13.4 Additional Ecosystem Tools#
- 127.13.4.1 Implement AFLOW prototype library interface: query AFLOW prototype encyclopedia (aflow.org/prototype-encyclopedia/) — given a composition and space group, retrieve matching prototypes (Strukturbericht designation, Pearson symbol, Wyckoff positions) — generate KalikaStructure from prototype
- 127.13.4.2 Implement Open Babel molecular format bridge: via Python bridge to openbabel, convert between molecular formats (SMILES, InChI, MOL, SDF, PDB, XYZ, CIF, VASP) — enabling Kalika to ingest molecular structures from any standard chemistry format
- 127.13.4.3 Implement CIF symmetry operations handler: parse and apply CIF symmetry operations (space group generators in x,y,z notation) to generate all symmetry-equivalent positions from asymmetric unit — essential for building full unit cells from CIF data
127.14 Materials Databases (@kalika/materials-databases)#
127.14.1 Materials Project API#
- 127.14.1.1 Implement Materials Project REST client: authenticated HTTP client for MP API v2023+ (api.materialsproject.org) — API key management, rate limiting (configurable RPS), request retry with exponential backoff, pagination handling for large result sets
- 127.14.1.2 Implement MP structure search: query materials by elements (contains/excludes), formula (exact/anonymous/reduced), crystal system, space group, number of sites, band gap range, energy above hull range, is_stable, magnetic ordering, has_props filter — return MaterialDoc list with material_id, formula, structure
- 127.14.1.3 Implement MP property retrieval: fetch detailed properties for a material_id — total energy (per atom, per formula unit), formation energy, energy above hull, band gap (PBE, HSE, GW if available), DOS (total + projected), band structure, elastic tensor (C_ij, K_VRH, G_VRH, Poisson ratio), magnetic moment, dielectric constant, piezo tensor — return typed property objects
- 127.14.1.4 Implement MP phase diagram client: fetch all entries in a
chemical system (e.g., Li-Fe-O), construct
PhaseDiagramfrom MP entries — compute stability (e_above_hull), decomposition products, equilibrium reaction, chemical potential ranges — for thermodynamic stability screening - 127.14.1.5 Implement MP thermo data access: fetch formation energies, decomposition energies, equilibrium reactions, energy above hull for all entries in a composition space — for large-scale stability analysis and Pourbaix diagram construction
- 127.14.1.6 Implement MP task document access: fetch raw task documents (VASP calculation metadata) — INCAR settings used, k-point density, pseudopotentials, energy corrections applied (GGA mixing, anion corrections) — for reproducing or understanding MP calculations
127.14.2 AFLOW API#
- 127.14.2.1 Implement AFLOW REST client: HTTP client for aflowlib.org REST API — construct AFLOW search URLs with filters (species, nspecies, Pearson, space_group, Egap, enthalpy_formation_atom), parse JSON responses, handle pagination (catalog, aurl)
- 127.14.2.2 Implement AFLOW property retrieval: fetch structure (geometry, species, positions), formation enthalpy (enthalpy_formation_atom, enthalpy_formation_cell), band gap (Egap, Egap_type), elastic properties (ael_bulk_modulus_vrh, ael_shear_modulus_vrh, ael_poisson_ratio), magnetic properties — return typed objects
- 127.14.2.3 Implement AFLOW prototype matching: given a KalikaStructure, query AFLOW prototype encyclopedia to identify the structure prototype (label, Strukturbericht, mineral name, Pearson symbol) — match by space group and Wyckoff positions
127.14.3 JARVIS API#
- 127.14.3.1 Implement JARVIS REST client: HTTP client for jarvis.nist.gov API — query JARVIS-DFT, JARVIS-FF, JARVIS-ML datasets — filters by elements, formula, space group, dimensionality (3D/2D/1D), band gap range, formation energy range
- 127.14.3.2 Implement JARVIS property retrieval: fetch DFT properties (optb88vdw formation energy, band gap, elastic constants, dielectric, piezoelectric, Born charges, phonon frequencies), 2D-specific properties (exfoliation energy, work function), ALIGNN ML predictions, solar cell efficiency metrics (SLME, SQ limit)
- 127.14.3.3 Implement JARVIS 2D materials search: specialized query for 2D materials database — filter by number of layers, exfoliation energy, band gap, magnetic ordering — return structures with layer-specific properties
127.14.4 OQMD & NOMAD APIs#
- 127.14.4.1 Implement OQMD REST client: query oqmd.org/api/v1 — search by composition, filter by stability (delta_e), formation energy (delta_e), band gap, space group — fetch structure (sites, lattice) and calculated properties
- 127.14.4.2 Implement NOMAD API client: query nomad-lab.eu API v1 — search by elements, formula, code (VASP/QE/etc.), upload date, dataset — fetch structures, energies, band structure, DOS from the FAIR (Findable Accessible Interoperable Reusable) materials data repository
127.14.5 Crystallographic Databases#
- 127.14.5.1 Implement COD (Crystallography Open Database) client: query cod.iimcb.gov.pl REST API — search by formula, elements, cell parameters, space group, journal reference — download CIF files, parse to KalikaStructure — completely free and open access
- 127.14.5.2 Implement ICSD query interface: generate ICSD-compatible search queries (composition, space group, Pearson, Wyckoff, cell parameters, structure type) — parse ICSD CIF exports — note: ICSD requires institutional subscription, so implement query/parse but not direct API access
- 127.14.5.3 Implement Materials Cloud API client: query materialscloud.org — browse AiiDA-based datasets (curated entries with full provenance), fetch structures and properties — especially useful for 2D materials, phonons, and GW datasets
127.14.6 Standardized APIs#
- 127.14.6.1 Implement OPTIMADE client: query any OPTIMADE-compliant
database (optimade.org) using standardized filter language —
filter=elements HAS "Li" AND nelements=3 AND nsites<20— parse OPTIMADE JSON:API responses, extract structures (species, coordinates, lattice_vectors) — works with MP, AFLOW, OQMD, COD, NOMAD, MC3D, and 30+ other databases through a single query interface - 127.14.6.2 Implement OPTIMADE provider registry: fetch and cache the list of OPTIMADE providers from providers.optimade.org — enable federated queries across all providers, aggregate results, de-duplicate by composition/space group
- 127.14.6.3 Implement multi-database search aggregator: given search criteria (composition, property ranges), query multiple databases (MP, AFLOW, OQMD, JARVIS, NOMAD, COD) in parallel via OPTIMADE where available and native APIs otherwise — merge results, flag duplicates, compare reported properties across sources
127.14.7 Molecular & Chemical Databases#
- 127.14.7.1 Implement PubChem REST client: query pubchem.ncbi.nlm.nih.gov PUG REST API — search by name, SMILES, InChI, molecular formula — retrieve 3D conformers, properties (molecular weight, charge, LogP, TPSA), CID — convert PubChem 3D structures to KalikaStructure (molecular, non-periodic)
- 127.14.7.2 Implement Citrination/Citrine API client: query Citrine Informatics platform — search by material system, property (band gap, thermal conductivity, yield strength), data source — retrieve curated experimental and computational data — useful for ML training data
127.15 Workflow Engine (@kalika/workflows)#
127.15.1 DAG Execution Engine#
- 127.15.1.1 Implement workflow execution engine: execute a
WorkflowDAG— topological traversal, parallel execution of independent steps, data passing between steps (output file mapping → input file placement), status tracking per step, event emission (step started/completed/ failed), timeout handling, overall workflow status aggregation - 127.15.1.2 Implement job submission abstraction:
JobSubmitterinterface with implementations for local execution (child process), SSH remote (paramiko-style), PBS/Torque (qsub), SLURM (sbatch), SGE (qsubwith SGE flags), LSF (bsub) — each generating the appropriate job script with resource requests (nodes, cores, memory, walltime, queue) - 127.15.1.3 Implement job status polling: for each scheduler backend
(local/PBS/SLURM/SGE/LSF), implement
getStatus(jobId): JobStatus— map scheduler-specific states (SLURM: PENDING/RUNNING/COMPLETING/ COMPLETED/FAILED/TIMEOUT/CANCELLED) to Kalika JobStatus enum — with configurable polling interval and exponential backoff - 127.15.1.4 Implement error recovery engine: when a workflow step fails, invoke the code-specific error detector (127.2.5.1 for VASP, etc.), apply automatic fixes if available, and restart the step — with configurable retry policy (max retries, fix escalation sequence) and dead-letter queue for unrecoverable failures
127.15.2 Provenance & Results#
- 127.15.2.1 Implement provenance tracker: for each workflow execution, record the full provenance chain — input structure, calculation parameters (every INCAR/KPOINTS/POTCAR tag), code version, output files, parsed results, wall time, node/core count — stored as a DAG of ProvenanceNode objects with immutable content hashing
- 127.15.2.2 Implement results database: store parsed calculation results (structure, energy, forces, band gap, DOS, elastic constants, etc.) in a queryable format — SQLite backend with typed columns, compound indices, and full-text search on notes/tags — for local result management and comparison
- 127.15.2.3 Implement results comparison engine: given two or more calculation results for the same structure (e.g., PBE vs. HSE, or different k-point densities), compute differences in energy, forces, band gap, DOS overlap, elastic constants — for convergence testing and functional benchmarking
- 127.15.2.4 Implement provenance serialization: export/import full provenance records as JSON/YAML — including structure snapshots, input parameters, output summaries, timing — for reproducibility documentation and sharing
127.15.3 Built-in Workflow Templates#
- 127.15.3.1 Implement
relaxation_staticworkflow: relax structure (ISIF=3, IBRION=2) → static calculation (ISMEAR=-5) → optional DOS/band structure — with automatic CONTCAR→POSCAR chaining, CHGCAR passing for NSCF, convergence checking between relax and static - 127.15.3.2 Implement
convergence_encutworkflow: run series of static calculations with increasing ENCUT (e.g., 200-800 eV in 50 eV steps) — extract energy vs. ENCUT, determine converged ENCUT (energy change < threshold, default 1 meV/atom) - 127.15.3.3 Implement
convergence_kpointsworkflow: run series of static calculations with increasing k-point density (e.g., 2×2×2 to 12×12×12) — extract energy vs. k-points, determine converged mesh - 127.15.3.4 Implement
elastic_constantsworkflow: apply 6 independent deformations (3 normal + 3 shear, with ±δ for central differences) to relaxed structure → static calculation for each → extract stress tensors → compute full 6×6 elastic tensor C_ij via stress-strain relation → derive bulk modulus K, shear modulus G (Voigt-Reuss-Hill averages), Young's modulus E, Poisson ratio ν - 127.15.3.5 Implement
phonon_supercellworkflow: generate supercell with Phonopy-compatible displacements → DFT force calculation for each displaced supercell → collect forces into FORCE_SETS → compute force constants → phonon band structure + DOS + thermodynamic properties — with automatic supercell size selection based on target accuracy - 127.15.3.6 Implement
defect_formation_energyworkflow: relax bulk supercell → create defect supercell (vacancy, substitution, interstitial) → relax defect supercell → compute raw defect formation energy E_f = E_defect - E_bulk + Σ n_i μ_i + q(E_VBM + E_Fermi) + E_correction — apply finite-size corrections (Freysoldt/Kumagai for charged defects) - 127.15.3.7 Implement
surface_energyworkflow: cleave slab from bulk (Miller indices hkl, slab thickness, vacuum thickness) → optimize slab → compute surface energy γ = (E_slab - n·E_bulk) / (2A) — with slab thickness convergence testing (increasing layers until γ converges) - 127.15.3.8 Implement
neb_barrierworkflow: interpolate images between initial and final structures (linear interpolation or IDPP — image dependent pair potential) → run CI-NEB calculation → extract minimum energy path and activation barrier — with automatic image count selection - 127.15.3.9 Implement
gw_band_gapworkflow: DFT relaxation → DFT static (dense k-mesh, many empty bands) → G0W0 calculation → extract quasiparticle band gap — with automatic NBANDS, ENCUTGW, NOMEGA convergence checks - 127.15.3.10 Implement
optical_bseworkflow: DFT → GW (quasiparticle energies) → BSE (Bethe-Salpeter equation for excitons) → absorption spectrum with excitonic effects — extract optical gap, exciton binding energy, dielectric function ε(ω) - 127.15.3.11 Implement
thermoelectric_ztworkflow: DFT relaxation → static (dense k-mesh) → Wannier90 interpolation → BoltzTraP2/BoltzWann transport (σ/τ, S, κ_e/τ) + phonon workflow (κ_L from Phono3py) → combine for ZT(T, n) map — the complete thermoelectric screening pipeline - 127.15.3.12 Implement
band_structure_hybridworkflow: PBE relaxation → PBE static → HSE06 or PBE0 band structure (NSCF with hybrid functional on PBE charge density) — for accurate band gap prediction beyond semilocal DFT
127.15.4 Workflow Utilities#
- 127.15.4.1 Implement workflow serialization: save/load WorkflowDAG definitions as YAML/JSON — including all step parameters, dependencies, output mappings, retry policies — for workflow sharing and version control
- 127.15.4.2 Implement workflow dry-run mode: validate a WorkflowDAG without executing — check all input file dependencies are satisfied, all referenced codes are available, all parameter values are valid, estimated compute cost (core-hours) from similar past runs
- 127.15.4.3 Implement workflow progress dashboard data: emit structured status updates for each workflow — overall progress (steps completed/ total), per-step status (waiting/running/done/failed), estimated time remaining, resource utilization — for consumption by a UI layer
- 127.15.4.4 Implement workflow template registry: register, discover,
and instantiate built-in and user-defined workflow templates by name —
registry.get('elastic_constants').instantiate({ structure, code: 'vasp', functional: 'PBE' })— with parameter validation against template schema
127.16 Integration Tests & Cross-Code Validation#
127.16.1 Round-Trip Tests#
- 127.16.1.1 Implement VASP round-trip tests: generate INCAR/POSCAR/ KPOINTS for Si diamond (FCC, 2 atoms) with each preset → write to string → parse back → verify all parameters preserved — for each preset (static, relaxation, band_structure, dos, dielectric, hybrid, phonon, elastic, aimd, soc)
- 127.16.1.2 Implement QE round-trip tests: generate pw.x input for Si, GaAs, Fe (magnetic) with each preset → write → parse → verify all namelists and cards preserved — test ibrav=0 and ibrav>0 modes
- 127.16.1.3 Implement LAMMPS round-trip tests: generate data file + input script for Cu FCC (EAM), SiO2 (Tersoff), polymer (full atom style) → write → parse → verify atom coordinates, box dimensions, pair coefficients preserved
- 127.16.1.4 Implement structure format round-trip tests: convert Si diamond through all format converters (CIF → KalikaStructure → POSCAR → KalikaStructure → QE input → KalikaStructure → XYZ → KalikaStructure → CIF) — verify lattice parameters and fractional coordinates preserved within tolerance (1e-10)
127.16.2 Cross-Code Consistency Tests#
- 127.16.2.1 Implement cross-code input consistency tests: for the same structure (Si diamond) and equivalent parameters (PBE, 500 eV, 8×8×8 k-mesh), generate inputs for VASP, QE, ABINIT, FHI-aims — verify that equivalent physical parameters are set (same XC functional, same effective PW cutoff in Ry vs eV, same k-mesh)
- 127.16.2.2 Implement unit conversion tests: verify all unit conversions against NIST reference values — eV↔Ha, Å↔Bohr, GPa↔kbar, etc. — with tolerances matching published CODATA uncertainties
- 127.16.2.3 Implement database API response tests: mock HTTP responses from MP, AFLOW, JARVIS, OQMD APIs with real response payloads → parse → verify extracted structures and properties match known values (e.g., Si mp-149 band gap ≈ 0.6 eV PBE, ≈ 1.17 eV HSE)
127.16.3 Output Parser Validation#
- 127.16.3.1 Implement VASP output parser tests: include fixture files (truncated real OUTCAR, vasprun.xml, DOSCAR, EIGENVAL, PROCAR for Si, Fe, GaAs) → parse → verify extracted energies, forces, band gap, DOS shape against reference values from the actual VASP runs
- 127.16.3.2 Implement QE output parser tests: include fixture files (truncated pw.x output, XML data-file-schema for Si, MgO) → parse → verify energies (Ry), forces, eigenvalues against known values
- 127.16.3.3 Implement Phonopy parser tests: include fixture files (band.yaml, mesh.yaml, FORCE_SETS for Si 2×2×2 supercell) → parse → verify phonon frequencies at Γ (3 acoustic = 0, 3 optical ≈ 15.5 THz for Si), thermodynamic properties at 300 K against published values
- 127.16.3.4 Implement Wannier90 parser tests: include fixture files (.wout for Si 4-band sp3 Wannier functions, _hr.dat) → parse → verify spread convergence, Wannier centers at bond-center positions, Hamiltonian matrix elements against reference
- 127.16.3.5 Implement LOBSTER parser tests: include fixture COHPCAR for Fe-Fe bond → parse → verify ICOHP value, energy range, orbital decomposition against published LOBSTER tutorial values
127.16.4 Workflow Template Tests#
- 127.16.4.1 Implement workflow DAG validation tests: construct each built-in workflow template → validate DAG (no cycles, all dependencies satisfied, all output mappings valid) → verify execution order layers — test with different codes (VASP, QE) and parameter variations
- 127.16.4.2 Implement workflow serialization round-trip tests: for each built-in workflow template, serialize to YAML → deserialize → verify all steps, dependencies, and parameters preserved
- 127.16.4.3 Implement error handler tests: simulate common VASP errors (inject ZBRENT, EDDDAV, ZPOTRF signatures into mock OUTCAR) → verify error detection triggers → verify fix suggestions are correct (POTIM reduction, ALGO change, etc.)
- 127.16.4.4 Implement provenance tracking tests: execute a mock 3-step workflow → verify provenance DAG records all inputs, outputs, parameters, timing for each step — verify immutable content hashes match expected values
127.17 Documentation & Type Exports#
127.17.1 Public API Surface#
- 127.17.1.1 Implement barrel exports for
@kalika/code-interfaces: export all core types (CalculationInput, CalculationOutput, CodeInterface, WorkflowStep, WorkflowDAG, JobStatus, KalikaStructure), all code-specific interfaces (VaspInterface, QeInterface, AbinitInterface, etc.), all converters (CIF, POSCAR, XYZ, PDB), unit conversion utilities, physical constants - 127.17.1.2 Implement barrel exports for
@kalika/vasp-io: export INCAR writer/reader/validator/presets, POSCAR reader/writer, KPOINTS writer (automatic + line-mode), POTCAR manager, all output parsers (OUTCAR, vasprun.xml, DOSCAR, EIGENVAL, PROCAR, CHGCAR, etc.), error handlers, version compatibility layer - 127.17.1.3 Implement barrel exports for
@kalika/qe-io: export pw.x input writer/presets, pseudopotential manager, all output parsers (pw.x, XML, ph.x, matdyn.x, pp.x), post-processing tool interfaces (bands.x, dos.x, projwfc.x, epsilon.x, hp.x) - 127.17.1.4 Implement barrel exports for
@kalika/lammps-io: export data file reader/writer, input script builder, pair style helpers, fix/compute/dump helpers, presets, log parser, dump parser, potential registry, OpenKIM interface - 127.17.1.5 Implement barrel exports for
@kalika/materials-databases: export all database clients (MP, AFLOW, JARVIS, OQMD, NOMAD, COD, ICSD, Materials Cloud, OPTIMADE, PubChem, Citrination), multi-database aggregator, OPTIMADE provider registry - 127.17.1.6 Implement barrel exports for
@kalika/workflows: export WorkflowDAG, execution engine, job submitters (local/SLURM/PBS/SGE/ LSF), all built-in workflow templates, provenance tracker, results database, template registry - 127.17.1.7 Implement barrel exports for ancillary libraries:
@kalika/abinit-io,@kalika/fhi-aims-io,@kalika/phonopy-io,@kalika/wannier90-io,@kalika/lobster-io,@kalika/boltztrap-io,@kalika/gromacs-io— each exporting all public types, generators, parsers, and utilities