Eve library

phase-128.md

Tracker

Revision 1

On this page

Historical task source. The Eve task board holds current task state.

Phase 128: Kalika Materials Science Research Platform — Crystal Structure Visualization, Property Dashboards, Autonomous Materials Discovery Agents & Collaborative Research Workbench#

The definitive materials science research environment. This phase builds everything a materials scientist needs from first login to published paper: interactive 3D crystal structure visualization (WebGPU-accelerated, matching VESTA/CrystalMaker/Materials Studio quality), electronic structure visualization (band structures, DOS, Fermi surfaces, charge densities), property dashboards (phase diagrams, phonons, transport, elasticity), materials-specific computational notebooks (extending Phase 105), autonomous discovery agents (generate-screen-compute-validate loops), collaborative research infrastructure (shared projects, provenance, publication output), experimental characterization integration, a materials knowledge graph, educational onboarding, and production-grade performance/scalability. Extends Phase 105 (Research Platform), Phase 107 (UI/UX), and the broader materials stack in Phases 116-127, especially Phases 119, 121, and 125. Technology split: TypeScript (~70%) for UI, notebooks, dashboards, agents, knowledge graph, and orchestration; Rust (~15%) for WebGPU 3D rendering kernels, isosurface extraction, and large-structure LOD; Python (~15%) for DFT/ML wrappers, database connectors, and validation scripts.


128.1 Scaffolding & Core Types (@kalika/materials-platform)#

128.1.1 Library Scaffolding#

  • 128.1.1.1 Create libs/kalika/materials-platform/ directory with package.json, project.json, tsconfig.json, and src/index.ts barrel export. Register path mapping @kalika/materials-platform in tsconfig.base.json
  • 128.1.1.2 Create Rust crate libs/kalika/materials-platform/crate/ with Cargo.toml (dependencies: wgpu, glam, bytemuck, naga, rayon, serde, napi-rs), src/lib.rs, and napi-rs bindings for exposing GPU rendering pipeline to TypeScript
  • 128.1.1.3 Configure Nx project.json with targets: build (tsup + cargo build), test (vitest + cargo test), lint (eslint + clippy), build:wasm (wasm-pack for browser-side 3D rendering)
  • 128.1.1.4 Create Python integration layer libs/kalika/materials-platform/python/ with pyproject.toml for Materials Project API, AFLOW REST, JARVIS-DFT connectors, and pymatgen interoperability utilities

128.1.2 Core Type Definitions#

  • 128.1.2.1 Define MaterialsProject type: top-level workspace containing name, description, list of CrystalStructure entries, list of Calculation records, PropertyDashboard configurations, DiscoveryAgent sessions, ResearchNotebook references, collaborator list with permission levels, creation/modification timestamps, and provenance metadata
  • 128.1.2.2 Define CrystalStructure type: lattice vectors (3x3 matrix), fractional coordinates per site, atomic species per site, site symmetry labels, space group (Hermann-Mauguin and Hall symbols, IT number), magnetic moments per site, oxidation states per site, occupancy factors (for disordered structures), anisotropic displacement parameters (6 independent U_ij per site), and metadata (source CIF, provenance, tags)
  • 128.1.2.3 Define Visualization3D type: scene graph containing list of AtomNode (position, element, radius, color, transparency, label, thermal ellipsoid), BondNode (start/end atom indices, bond order, length, color, cylinder radius), PolyhedronNode (central atom, vertex atoms, face list, color, transparency), UnitCellNode (edges, axis labels), IsosurfaceNode (mesh vertices/normals/colors, isovalue), camera state (position, target, up, projection type, FOV), and lighting configuration
  • 128.1.2.4 Define PropertyDashboard type: dashboard layout (grid of panels), each panel containing a DashboardWidget (plot type, data source, axis configuration, color map, interactive callbacks), with support for linked interactions (click point in phase diagram → highlight structure in 3D viewer) and responsive layout specifications
  • 128.1.2.5 Define DiscoveryAgent type: agent configuration (goal description, element palette, property constraints, compute budget), execution state (idle, running, paused, completed), candidate list with scores, workflow DAG (generate → screen → compute → validate stages), logs, and human-in-the-loop checkpoint definitions
  • 128.1.2.6 Define ResearchNotebook type: extending Phase 105 notebook model with materials-specific cell types (structure cell, band structure cell, phase diagram cell, reaction energy cell, screening cell, comparison cell), materials-specific autocomplete dictionary, and template references
  • 128.1.2.7 Define Calculation type: input structure, method (DFT functional, pseudopotential, k-mesh, cutoff energy, convergence criteria), output (total energy, forces, stress, electronic structure, phonons), computational cost (CPU-hours, wall-time, memory peak), status (queued, running, completed, failed), and full provenance (code version, commit hash, machine, submission time)
  • 128.1.2.8 Define MaterialsDatabase type: interface for querying external databases (Materials Project, AFLOW, OQMD, JARVIS, ICSD, COD) with unified query schema: composition, space group, band gap range, formation energy range, property filters, pagination, and result caching

128.2 Crystal Structure Visualization (@kalika/crystal-viewer)#

128.2.1 Rendering Pipeline (WebGPU via wgpu-rs)#

  • 128.2.1.1 Implement WebGPU rendering context initialization: adapter selection (high-performance GPU preferred), device/queue creation, swap chain configuration (sRGB, MSAA 4x), depth buffer, and fallback to WebGL2 when WebGPU is unavailable — in Rust compiled to WASM via wasm-pack
  • 128.2.1.2 Implement instanced sphere rendering for atoms: single sphere mesh (icosphere subdivision level 3, ~640 triangles) rendered via GPU instancing with per-instance position, radius, and RGBA color — achieving 60fps for 10,000+ atoms on mid-range GPUs
  • 128.2.1.3 Implement instanced cylinder rendering for bonds: single cylinder mesh rendered with per-instance start/end positions, radius, and color, with hemispherical caps for smooth bond termination at atom surfaces
  • 128.2.1.4 Implement polygon mesh rendering for coordination polyhedra: convex hull computation from vertex atom positions, face normal calculation, per-face or per-polyhedron coloring with controllable transparency (order- independent transparency via weighted blended OIT)
  • 128.2.1.5 Implement Phong/PBR shading pipeline: ambient, diffuse, specular lighting with configurable light positions, material roughness/ metalness for PBR mode, environment mapping for reflections, and shadow mapping (cascaded shadow maps for large structures)
  • 128.2.1.6 Implement atom picking via GPU ray casting: shoot ray from mouse position through view frustum, intersect with atom spheres (analytic ray-sphere intersection, not bounding box), return nearest hit atom with < 1ms latency for structures up to 100,000 atoms using BVH acceleration
  • 128.2.1.7 Implement anti-aliasing: MSAA 4x as default, optional FXAA post-process for performance, and supersampling (render at 2x resolution) for publication-quality screenshots

128.2.2 Representation Modes#

  • 128.2.2.1 Implement ball-and-stick representation: atoms as spheres with radii proportional to covalent/ionic/van der Waals radius (user selectable, default covalent × 0.4), bonds as cylinders connecting bonded atoms, with automatic bond detection (sum of covalent radii + tolerance, default tolerance 0.45 angstrom) and manual bond addition/removal
  • 128.2.2.2 Implement space-filling representation: atoms as spheres at full van der Waals radii, no bonds displayed, overlap between atoms visible (useful for packing visualization), with optional slicing plane to show interior
  • 128.2.2.3 Implement polyhedral representation: identify coordination polyhedra (octahedra, tetrahedra, square planar, trigonal bipyramidal, etc.) by detecting nearest neighbors of selected central atoms (e.g., all Ti in TiO2), render as semi-transparent colored polyhedra with shared faces/edges highlighted, support for distortion indicators (polyhedral volume, quadratic elongation, bond angle variance)
  • 128.2.2.4 Implement wireframe representation: unit cell edges as lines, bonds as thin lines, atoms as small dots or not shown — minimal rendering for structural overview and performance with very large systems
  • 128.2.2.5 Implement mixed representation mode: different representation per atom selection (e.g., polyhedra for octahedral sites, ball-and-stick for framework atoms, space-filling for intercalant ions) — essential for complex structures like perovskites and zeolites
  • 128.2.2.6 Implement thermal ellipsoid rendering: anisotropic displacement parameters (ADPs, U_ij tensor) rendered as 3D ellipsoids (ORTEP style) at configurable probability level (50%, 90%, 99%), with principal axes shown as lines through the ellipsoid

128.2.3 Atom & Bond Appearance#

  • 128.2.3.1 Implement CPK coloring schemes: Jmol color table (118 elements), VESTA color table, RasMol color table, and custom per-element color assignment with color picker — stored as named color schemes that can be saved and shared
  • 128.2.3.2 Implement element-based atom sizing: lookup tables for atomic radius, ionic radius (by oxidation state and coordination number), covalent radius, and van der Waals radius from published crystallographic data (e.g., Shannon ionic radii, Cordero covalent radii)
  • 128.2.3.3 Implement atom transparency control: per-atom or per-element alpha, useful for highlighting subsurface atoms or de-emphasizing framework atoms while focusing on guest molecules
  • 128.2.3.4 Implement selection highlighting: selected atoms rendered with glowing outline (bloom post-process on selected atoms), different selection colors for primary/secondary/tertiary selections, and selection info panel showing element, position, distances to neighbors
  • 128.2.3.5 Implement atom labels: element symbol, site index, Wyckoff position, charge state (e.g., Fe3+), custom text — rendered as billboarded SDF text in 3D space with configurable font size, offset, and visibility threshold (hide labels when zoomed out past a distance)
  • 128.2.3.6 Implement bond order visualization: single bonds as single cylinder, double bonds as two parallel cylinders, triple bonds as three cylinders in triangular arrangement, aromatic bonds as dashed cylinder — bond order inferred from bond length tables or specified manually
  • 128.2.3.7 Implement bond length labels: display bond length in angstroms as 3D text at bond midpoint, togglable per-bond or globally, formatted to 3 decimal places, with lines connecting label to bond

128.2.4 Unit Cell & Supercell#

  • 128.2.4.1 Implement unit cell display: parallelepiped edges drawn as lines with configurable thickness and color, lattice parameter labels (a, b, c, alpha, beta, gamma) positioned along edges, origin marker, and colored axis vectors (a=red, b=green, c=blue) with arrowheads
  • 128.2.4.2 Implement supercell visualization: generate NxMxL supercell (user-specified repetitions along a, b, c), render with periodic boundary handling (atoms at cell faces appear on both sides), selective display (show only atoms within a distance cutoff from a selected point), and supercell boundary edges
  • 128.2.4.3 Implement conventional vs. primitive cell toggle: convert between primitive and conventional cell representations, display both simultaneously (primitive cell highlighted within conventional), and support for non-standard cell choices

128.2.5 Symmetry & Planes#

  • 128.2.5.1 Implement symmetry element visualization: rotation axes (C2, C3, C4, C6) as colored rods through the structure, mirror planes as semi- transparent rectangles, inversion centers as special markers, screw axes with helical indicator, glide planes with directional arrow — all derived from the space group operations
  • 128.2.5.2 Implement Miller plane visualization: given (hkl) indices, compute and render the crystallographic plane as a semi-transparent colored surface intersecting the unit cell, with plane normal vector shown, d- spacing labeled, and support for displaying families of parallel planes {hkl}
  • 128.2.5.3 Implement Brillouin zone overlay: compute first Brillouin zone from reciprocal lattice vectors (Wigner-Seitz cell in reciprocal space), render as wireframe overlaid on crystal structure (with appropriate scaling), label high-symmetry points (Gamma, X, M, K, L, etc.)

128.2.6 Defect & Surface Visualization#

  • 128.2.6.1 Implement vacancy visualization: missing atom site shown as transparent sphere with dashed outline or X marker, with surrounding atoms showing relaxation vectors (arrows indicating displacement from ideal positions)
  • 128.2.6.2 Implement interstitial visualization: extra atom highlighted with distinct color/glow, with distance measurements to nearest lattice sites automatically displayed
  • 128.2.6.3 Implement substitutional defect visualization: substituted atom rendered with different color and size from host atom, with optional side-by-side comparison showing original vs. defective structure
  • 128.2.6.4 Implement surface slab visualization: slab model with vacuum region, surface termination atoms colored distinctly, surface normal indicator, layer-by-layer coloring (surface, subsurface, bulk-like), and adsorbed molecule rendering with adsorption site geometry highlighted

128.2.7 Camera & Interaction#

  • 128.2.7.1 Implement orbital camera controls: arcball rotation (rotate around structure center), pan (translate camera), zoom (dolly in/out with scroll wheel, pinch on touch), with smooth inertial damping and configurable sensitivity
  • 128.2.7.2 Implement perspective/orthographic toggle: perspective projection for natural 3D appearance, orthographic for accurate structural comparison and measurement, with smooth animated transition between modes
  • 128.2.7.3 Implement preset crystal orientations: one-click views along [100], [010], [001], [110], [111], [1-10] directions, with smooth animated camera transition (spherical linear interpolation of quaternion orientation)
  • 128.2.7.4 Implement measurement tools: click two atoms to measure distance (angstroms), click three atoms to measure angle (degrees), click four atoms to measure dihedral angle, with persistent measurement annotations displayed in 3D scene

128.2.8 Animation & Export#

  • 128.2.8.1 Implement phonon mode animation: displace atoms according to phonon eigenvector at selected q-point and branch, animate sinusoidal oscillation with configurable amplitude and speed, support for superposition of multiple modes, and freeze-frame export at specific phase
  • 128.2.8.2 Implement MD trajectory playback: load trajectory frames (positions per timestep), play/pause/scrub with frame slider, adjustable playback speed, trail rendering (show atom paths over N frames), and on-the-fly RDF/MSD computation during playback
  • 128.2.8.3 Implement NEB path animation: render minimum energy path between initial and final structures, interpolate intermediate images, animate transition with energy profile overlay, highlight breaking/forming bonds during transition
  • 128.2.8.4 Implement PNG export: render current view at configurable resolution (up to 8K for publication), with transparent background option, embedded scale bar, and crystal axis indicator
  • 128.2.8.5 Implement SVG export: convert 3D scene to 2D vector graphics via painter's algorithm (depth-sorted rendering), with atom circles, bond lines, and labels as editable SVG elements — for journal figure preparation in Illustrator/Inkscape
  • 128.2.8.6 Implement glTF 3D model export: export full scene as glTF 2.0 binary (.glb) for import into Blender, PowerPoint, or web 3D viewers, with PBR materials and animation data preserved
  • 128.2.8.7 Implement POV-Ray scene export: generate POV-Ray scene description file for photorealistic ray-traced rendering, with material definitions, lighting setup, and camera matching the current viewport

128.2.9 Notebook Integration#

  • 128.2.9.1 Implement inline 3D viewer in Phase 105 notebooks: embed interactive crystal structure viewer as a notebook cell output, with full rotation/zoom/pick capabilities within the notebook, synchronized with structure data in code cells
  • 128.2.9.2 Implement structure editing in notebook cells: click-to-add atoms, drag-to-move atoms, delete atoms, modify species — edits reflected back to the code cell's structure variable, enabling interactive structure manipulation within the reactive notebook paradigm

128.3 Electronic Structure Visualization (@kalika/electronic-viz)#

128.3.1 Band Structure Plots#

  • 128.3.1.1 Implement band structure plot renderer: plot E(k) along high- symmetry path with automatic k-path detection from space group (using Setyawan-Curtarolo conventions), vertical lines at high-symmetry points with labels (Gamma, X, M, etc.), Fermi level as horizontal dashed line, energy axis with configurable range and units (eV, Ry, Ha)
  • 128.3.1.2 Implement fat band rendering: color each band segment by orbital character (s=red, p=green, d=blue, f=orange or user-defined), with line width proportional to projection weight (0-1), and legend showing orbital-color mapping — essential for understanding electronic character
  • 128.3.1.3 Implement spin-resolved band structure: separate panels for spin-up and spin-down channels, or overlaid with different colors/line styles, with exchange splitting visualization and spin-polarization energy labeled
  • 128.3.1.4 Implement non-collinear spin texture visualization: band structure colored by spin expectation value component (S_x, S_y, S_z), with spin arrows on band points for Rashba/Dresselhaus systems
  • 128.3.1.5 Implement band structure comparison overlay: plot band structures from multiple calculations (different functionals, different structures) on the same axes with different colors/line styles, automatic alignment by Fermi level or vacuum level, and difference highlighting
  • 128.3.1.6 Implement interactive band structure: hover over any point to show tooltip with eigenvalue (eV), k-point (fractional coordinates and label), orbital decomposition (% s, p, d, f), and band index; click to highlight that band across the entire path

128.3.2 Density of States#

  • 128.3.2.1 Implement total DOS plot: g(E) vs. E with configurable Gaussian/Lorentzian broadening (sigma parameter), Fermi level marker, fill below curve (occupied states shaded), and configurable energy range
  • 128.3.2.2 Implement atom-projected DOS (PDOS): stacked area plot or overlaid line plots of DOS contribution from each atom or atom group, with color per element type, and interactive toggle to show/hide individual atom contributions
  • 128.3.2.3 Implement orbital-projected DOS: decompose PDOS into orbital contributions (s, p_x, p_y, p_z, d_xy, d_yz, d_xz, d_x2-y2, d_z2, and f orbitals), with stacked area or individual line plots per orbital
  • 128.3.2.4 Implement spin-resolved DOS: spin-up plotted positive, spin- down plotted negative (mirror plot), with net spin polarization highlighted and magnetic moment integration displayed
  • 128.3.2.5 Implement interactive DOS-structure linking: click on energy point in DOS → highlight atoms in 3D crystal viewer that contribute most at that energy, with opacity proportional to contribution weight

128.3.3 Fermi Surface Visualization#

  • 128.3.3.1 Implement Fermi surface extraction: from eigenvalues on a regular k-grid, extract isosurface at E_F using marching cubes algorithm (Rust implementation for performance), with configurable grid density and interpolation (trilinear or Fourier)
  • 128.3.3.2 Implement Fermi surface rendering: 3D isosurface colored by scalar field (Fermi velocity magnitude, effective mass, spin component), with Brillouin zone wireframe overlay, high-symmetry point labels, and multiple sheets (one per band crossing E_F) rendered with different colors
  • 128.3.3.3 Implement Fermi surface interaction: orbital camera rotation, interactive slicing plane (drag plane through BZ, show 2D cross-section), sheet visibility toggles, and color map selection
  • 128.3.3.4 Implement Fermi surface property mapping: color surface by group velocity (gradient of E(k)), effective mass tensor eigenvalues, electron-phonon coupling strength lambda_nk, or superconducting gap Delta_nk, with continuous color scale and histogram of mapped values

128.3.4 Charge Density & Orbital Visualization#

  • 128.3.4.1 Implement 3D charge density isosurface: load volumetric data (CHGCAR, cube file format), extract isosurface at user-selected isovalue via marching cubes (Rust), render as semi-transparent surface with color mapped to local value, isovalue slider for real-time isosurface update
  • 128.3.4.2 Implement 2D charge density cross-section: slice volumetric data along any crystallographic plane (hkl) or arbitrary plane defined by three points, render as color-mapped 2D image overlaid on the crystal structure, with contour lines at configurable intervals
  • 128.3.4.3 Implement charge density difference visualization: compute delta_rho = rho_compound - rho_A - rho_B (bonding charge redistribution), display positive (charge accumulation, blue) and negative (charge depletion, red) isosurfaces simultaneously for chemical bonding analysis
  • 128.3.4.4 Implement spin density visualization: render magnetization density m(r) = rho_up(r) - rho_down(r) isosurface, positive (majority spin) and negative (minority spin) in different colors, for magnetic structure analysis
  • 128.3.4.5 Implement ELF (Electron Localization Function) visualization: ELF isosurface rendering with values ranging 0-1, highlight covalent bonds (ELF ~ 0.8-1.0), lone pairs, metallic bonding (ELF ~ 0.5), with isovalue slider
  • 128.3.4.6 Implement Wannier function visualization: load Wannier90 output, render real-space Wannier function isosurface (positive/negative lobes in different colors), display Wannier center positions on crystal structure, and spread (Omega) values per Wannier function

128.4 Property Visualization Dashboards (@kalika/materials-dashboard)#

128.4.1 Phase Diagram Explorer#

  • 128.4.1.1 Implement binary phase diagram renderer: composition axis (x=0 to x=1 for A_{1-x}B_x), formation energy axis, convex hull as solid line connecting stable phases, unstable phases as points above hull with energy-above-hull labeled, and clickable phase points showing structure details
  • 128.4.1.2 Implement ternary phase diagram renderer: equilateral triangle with composition axes (A, B, C corners), convex hull projected as colored regions, tie lines connecting coexisting phases, and interactive click-on-composition to show stable phase assemblage and reaction energies
  • 128.4.1.3 Implement convex hull visualization: 3D convex hull (composition + energy) for ternary systems with interactive rotation, facet coloring by stability region, and mouse-over showing hull distance for any composition
  • 128.4.1.4 Implement energy-above-hull heatmap: for ternary systems, render a 2D heatmap of E_above_hull(x_A, x_B) over the composition triangle, with color scale from green (stable, E=0) to red (unstable, E > threshold), and contour lines
  • 128.4.1.5 Implement temperature-dependent phase diagram integration: connect to Phase 125 CALPHAD data, render binary/ternary phase diagrams with temperature as vertical axis (binary) or slider (ternary), showing liquidus, solidus, solvus lines, invariant reactions (eutectic, peritectic, eutectoid), and two-phase regions labeled
  • 128.4.1.6 Implement Pourbaix diagram renderer: plot stability regions in pH vs. electrode potential (E) space for aqueous electrochemistry, with regions labeled by stable phase (metal, oxide, dissolved ion), phase boundaries as lines, and water stability window overlay (H2/O2 evolution lines)

128.4.2 Phonon Visualization Dashboard#

  • 128.4.2.1 Implement phonon dispersion dashboard widget: interactive phonon band structure plot (extending Phase 119 data), with fat bands colored by atom contribution (which atoms move most in each mode), branch highlighting on hover, and linked animation (click mode → animate in 3D crystal viewer)
  • 128.4.2.2 Implement phonon DOS dashboard widget: total + atom-projected phonon DOS, with energy axis linked to dispersion plot (horizontal line synced between panels), and thermodynamic integral readouts (zero-point energy, thermal energy at selected T)
  • 128.4.2.3 Implement animated phonon mode viewer: select mode from dispersion plot, animate atomic displacements in 3D crystal viewer with amplitude and speed controls, display mode symmetry label and frequency
  • 128.4.2.4 Implement thermal property plots: C_v(T), entropy S(T), Helmholtz free energy F(T), mean square displacement <u^2>(T) computed from phonon DOS via standard statistical mechanics integrals, with multiple materials overlaid for comparison
  • 128.4.2.5 Implement Grueneisen parameter dispersion: plot mode Grueneisen parameter gamma_i(q) along high-symmetry path, color-coded by sign (positive=normal, negative=NTE), with volume-averaged Grueneisen parameter and thermal expansion coefficient displayed

128.4.3 Transport Property Dashboard#

  • 128.4.3.1 Implement thermoelectric property plots: electrical conductivity sigma(T), Seebeck coefficient S(T), electronic thermal conductivity kappa_e(T), and figure of merit ZT(T) as function of temperature, computed from Boltzmann transport and thermodynamic data (Phases 123/125), with configurable carrier concentration (n-type/p-type doping slider)
  • 128.4.3.2 Implement carrier concentration optimization view: plot ZT vs. carrier concentration at fixed temperature, identify optimal doping level with peak marker, and show trade-off between sigma and S (Pisarenko plot)
  • 128.4.3.3 Implement multi-material transport comparison: overlay transport properties of multiple materials on same axes (e.g., compare Bi2Te3, PbTe, SnSe, SiGe), with material legend and property ranking table
  • 128.4.3.4 Implement thermal conductivity breakdown: plot total kappa(T) decomposed into electronic (kappa_e) and lattice (kappa_L) contributions, with individual phonon branch contributions to kappa_L (acoustic vs. optical), and cumulative kappa_L vs. mean free path for nanostructuring guidance

128.4.4 Spectroscopy Comparison Dashboard#

  • 128.4.4.1 Implement theory-experiment overlay: upload experimental spectrum (CSV: energy/frequency vs. intensity), overlay with computed spectrum, with alignment tools (shift, scale, broadening adjustment), residual plot, and R-factor (agreement metric) computation
  • 128.4.4.2 Implement multi-method comparison: plot spectra from DFT-PBE, DFT-HSE, GW, BSE, or ML predictions on same axes, with method labels and color coding, to assess theoretical method accuracy
  • 128.4.4.3 Implement peak annotation: automatic peak detection (local maxima above threshold), assignment to electronic transitions or phonon modes, clickable annotations linking to underlying electronic structure or phonon mode in other dashboard panels

128.4.5 Elastic Property Visualization#

  • 128.4.5.1 Implement directional Young's modulus surface: compute E(n) = 1 / (s_{ijkl} n_i n_j n_k n_l) for all directions n on the unit sphere, render as 3D representation surface (radius = E(n)) colored by magnitude, with crystal axes overlay — revealing elastic anisotropy at a glance
  • 128.4.5.2 Implement elastic anisotropy indicators: compute and display Zener ratio A = 2C44/(C11-C12), universal anisotropy index A^U, and Ranganathan-Ostoja-Starzewski log-Euclidean anisotropy — with visual comparison to isotropic reference
  • 128.4.5.3 Implement elastic wave velocity surfaces: compute phase velocity v(n) and group velocity for longitudinal and transverse acoustic waves in all directions from Christoffel equation eigenvalues, render as 3D velocity surfaces with polarization indicators

128.4.6 Magnetic Structure Dashboard#

  • 128.4.6.1 Implement spin arrow rendering: display magnetic moment vectors as 3D arrows on atomic sites, with arrow length proportional to moment magnitude, color by component (collinear: up=red, down=blue), and configurable arrow size/style
  • 128.4.6.2 Implement non-collinear spin texture visualization: render spin vectors with arbitrary orientations, color by polar/azimuthal angle, support for spin spirals (show pitch, chirality), and skyrmion textures (topological charge computation)
  • 128.4.6.3 Implement magnon dispersion dashboard: plot magnon (spin wave) frequency vs. q-point along high-symmetry path, with intensity coloring from spin-spin structure factor, linked to real-space spin wave animation

128.5 Materials Science Notebooks (@kalika/materials-notebooks)#

128.5.1 Materials-Specific Cell Types#

  • 128.5.1.1 Implement structure cell: notebook cell containing an interactive 3D crystal structure viewer (using @kalika/crystal-viewer), with inline editing capabilities (add/move/delete atoms, change species), structure saved as CIF or POSCAR within the cell, and bidirectional sync with code variables in adjacent cells
  • 128.5.1.2 Implement band structure cell: configure calculation parameters (functional, k-path, energy range), execute band structure computation (via @kalika/electronic-structure or external code), and display interactive band structure plot with fat bands and DOS side panel — all within a single notebook cell
  • 128.5.1.3 Implement phase diagram cell: interactive composition explorer where user clicks on composition space, system queries database or runs calculation for stability at that composition, and progressively builds up a phase diagram with each query — combining exploration and computation
  • 128.5.1.4 Implement reaction energy cell: define chemical reaction (reactants → products with stoichiometric coefficients), automatically look up or compute energies for all phases, compute reaction enthalpy/Gibbs energy, display reaction energy diagram with transition states if available
  • 128.5.1.5 Implement screening cell: define property filters (band gap range, formation energy < threshold, element constraints), execute high- throughput screening pipeline across a database or generated candidates, display results as interactive table with sorting, filtering, and click-to- view-structure
  • 128.5.1.6 Implement comparison cell: side-by-side comparison of two or more materials — synchronized 3D viewers, overlaid property plots (band structure, DOS, phonons), property table with delta values, and radar chart for multi-property comparison

128.5.2 Smart Autocomplete & Context#

  • 128.5.2.1 Implement material name autocomplete: as user types a composition (e.g., "BaTiO"), suggest matching materials from databases (BaTiO3 perovskite, Ba2TiO4, etc.) with space group and key properties, and auto-fetch structure on selection
  • 128.5.2.2 Implement element symbol context: typing an element symbol (e.g., "Ti") triggers a context popup showing atomic number, electron configuration, common oxidation states, electronegativity, ionic radii, and links to property entries — immediate reference without leaving the notebook
  • 128.5.2.3 Implement workflow suggestion engine: based on current notebook content (e.g., "you have a structure and band structure — would you like to compute phonons, elastic constants, or defect formation energies next?"), suggest next computational steps with one-click cell insertion
  • 128.5.2.4 Implement input parameter validation: when configuring a calculation cell, validate parameters in real-time (k-mesh density sufficient for convergence, cutoff energy appropriate for pseudopotential, smearing width reasonable for insulator vs. metal), with warnings and suggested corrections

128.5.3 Template Notebooks#

  • 128.5.3.1 Implement "Battery material screening" template: structure input → stability assessment → voltage calculation (intercalation energy) → ion migration barrier (NEB) → volume change on cycling → screening summary dashboard, pre-populated with explanatory markdown cells and parameter suggestions for common chemistries (Li-ion, Na-ion, Mg-ion)
  • 128.5.3.2 Implement "Thermoelectric optimization" template: structure input → electronic structure (band gap, effective mass) → Boltzmann transport (sigma, S, kappa_e) → phonon calculation (kappa_L) → ZT computation → doping optimization → comparison with known thermoelectrics
  • 128.5.3.3 Implement "Catalyst design" template: surface slab generation → adsorption energy calculations (key intermediates) → scaling relations → volcano plot → selectivity analysis → stability assessment (dissolution potential, surface energy)
  • 128.5.3.4 Implement "Defect engineering" template: host structure → point defect generation (vacancies, interstitials, substitutionals) → defect formation energy calculation → charge transition level diagram → defect concentration vs. Fermi level → optical transition energies
  • 128.5.3.5 Implement "2D material discovery" template: bulk structure → exfoliation energy calculation → monolayer optimization → electronic structure → mechanical properties (in-plane stiffness, Poisson ratio) → optical absorption → stability assessment (phonon, AIMD)
  • 128.5.3.6 Implement "Alloy phase diagram" template: define composition space → SQS structure generation → total energy calculations → cluster expansion fitting → Monte Carlo thermodynamics → phase diagram construction → ordering temperature identification

128.5.4 Inline Computation & Caching#

  • 128.5.4.1 Implement inline DFT computation: execute DFT calculations (single-point, relaxation, band structure, DOS) directly from notebook cells, routing to local or remote compute resources (Phase 128.10), with progress bar showing SCF convergence in real time
  • 128.5.4.2 Implement inline ML prediction: run ML property predictions (CGCNN, MEGNet, M3GNet, ALIGNN from Phase 121) directly from notebook cells, with instant results and uncertainty quantification displayed as error bars or confidence intervals
  • 128.5.4.3 Implement results caching: hash-based cache keyed on structure (Niggli-reduced cell + sorted sites) + method parameters, avoid recomputation for previously calculated structure/method combinations, cache stored locally and optionally shared across team workspace

128.6 Autonomous Materials Discovery Agents (@kalika/discovery-agents)#

128.6.1 Structure Generation Agent#

  • 128.6.1.1 Implement prototype substitution engine: given target composition and element constraints, search ICSD/AFLOW prototype database, identify matching prototypes (same stoichiometry type), substitute elements while respecting ionic radius ratios and electronegativity differences, rank candidates by Goldschmidt tolerance factor and radius ratio rules
  • 128.6.1.2 Implement evolutionary structure generation: genetic algorithm for crystal structure prediction — population of random structures, crossover (slab cut-and-splice), mutation (atom displacement, lattice strain, permutation), selection by fitness (computed energy), with convergence detection and diversity maintenance (niching)
  • 128.6.1.3 Implement generative model integration: connect to Phase 121 generative models (diffusion, flow matching) for neural structure generation, with conditional generation on target properties (band gap, formation energy), and diversity sampling to avoid mode collapse
  • 128.6.1.4 Implement structure validity filtering: Pauling's rules checker (coordination number vs. radius ratio, electrostatic valence sum, edge/face sharing restrictions), charge neutrality verification (sum of oxidation states = 0), minimum interatomic distance check (no atom overlap), and space group consistency validation

128.6.2 Stability Assessment Agent#

  • 128.6.2.1 Implement thermodynamic stability assessment: compute formation energy relative to elemental references, query/compute convex hull for the relevant chemical system, determine energy above hull (E_hull), classify as stable (E_hull = 0), metastable (E_hull < threshold, e.g., 50 meV/atom), or unstable — with uncertainty propagation from DFT errors
  • 128.6.2.2 Implement dynamic (phonon) stability check: compute phonon dispersion at Gamma point (fast) or full BZ (thorough), check for imaginary frequencies (soft modes indicating structural instability), report unstable mode character (which atoms, which direction), suggest distortion along soft mode eigenvector to find stable structure
  • 128.6.2.3 Implement mechanical stability check: compute elastic constant tensor C_ij, verify Born stability criteria (positive definite C_ij for the crystal system: e.g., cubic requires C11>0, C44>0, C11>|C12|, C11+2C12>0), report which criterion fails and by how much
  • 128.6.2.4 Implement AIMD stability check: run short ab initio molecular dynamics (e.g., 5 ps at target temperature), monitor for structural decomposition (bond breaking, phase separation), amorphization (loss of long-range order via Lindemann criterion), or phase transition (symmetry change) — report pass/fail with trajectory visualization

128.6.3 Property Prediction Agent#

  • 128.6.3.1 Implement property routing engine: given a target property (band gap, formation energy, elastic moduli, thermal conductivity, etc.) and accuracy requirement, select the optimal prediction method: fast ML model (ms, ~0.3 eV MAE for band gap), semi-empirical (seconds, ~0.2 eV), DFT-PBE (minutes-hours, ~0.5 eV systematic error), or hybrid functional/GW (hours-days, ~0.1 eV) — trading off speed vs. accuracy
  • 128.6.3.2 Implement ML-DFT cross-validation: when ML model prediction uncertainty exceeds threshold, automatically trigger DFT calculation for validation, compare ML vs. DFT result, and flag cases where ML is unreliable (out-of-distribution structures), feeding back to active learning loop
  • 128.6.3.3 Implement multi-property prediction: given a material, predict an ensemble of properties simultaneously (band gap, formation energy, elastic moduli, Debye temperature, dielectric constant), using appropriate models for each, and present results as a property card with confidence intervals
  • 128.6.3.4 Implement property trend analysis: across a set of candidate materials, identify structure-property correlations (e.g., band gap increases with electronegativity difference), visualize as scatter plots with regression lines, and highlight outliers for further investigation

128.6.4 Literature Search Agent#

  • 128.6.4.1 Implement materials database search: query Materials Project API (mp-api), AFLOW REST API, JARVIS-DFT API, OQMD API for existing data on a composition or structure, unify results into standard format, and report: known phases, measured/computed properties, and references
  • 128.6.4.2 Implement literature search: given a composition or structure type, search arXiv (cond-mat, materials science), Google Scholar, and journal APIs (APS, ACS, Elsevier) for prior publications, extract key findings (experimental properties, synthesis conditions), and generate a concise literature summary with citations
  • 128.6.4.3 Implement similar materials finder: given a target material, find structurally similar materials (same space group, similar lattice parameters), compositionally similar (isoelectronic substitutions), and functionally similar (similar band gap, similar elastic properties) — for analogy-based property prediction and inspiration
  • 128.6.4.4 Implement prior art check: before proposing a "new" material, verify it hasn't been previously reported in databases or literature, including structural polymorphs and solid solutions, to avoid redundant discovery claims

128.6.5 Synthesis Feasibility Agent#

  • 128.6.5.1 Implement thermodynamic synthesis assessment: given target material and precursors, compute reaction Gibbs energy at synthesis conditions (temperature, pressure, atmosphere), check if target is thermodynamically accessible, and identify competing phases that might form instead
  • 128.6.5.2 Implement synthesis route suggestion: based on target material chemistry, suggest appropriate synthesis methods (solid-state reaction, sol-gel, co-precipitation, hydrothermal, CVD, MBE, sputtering), with recommended precursors, temperatures, atmospheres, and time scales — using rules from synthesis databases (e.g., text-mined recipes from literature)
  • 128.6.5.3 Implement synthesis temperature estimation: from phase diagram data (Phase 125), estimate minimum synthesis temperature (lowest T where target phase is stable and kinetically accessible), melting point prediction, and decomposition temperature — with confidence intervals
  • 128.6.5.4 Implement earth-abundance and cost scoring: given a composition, compute elemental abundance score (Clarke number weighted), raw material cost estimate (from USGS commodity data), toxicity flags (Cd, Pb, Tl, Be, As, Hg), and supply chain risk indicators (critical minerals list) — essential for practical material selection

128.6.6 Multi-Agent Orchestration#

  • 128.6.6.1 Implement discovery loop orchestrator: coordinate the full generate → screen → compute → validate → report pipeline across agents, with configurable pipeline stages, conditional branching (skip expensive DFT if ML predicts E_hull > 200 meV), and parallel execution of independent stages
  • 128.6.6.2 Implement human-in-the-loop checkpoints: at configurable pipeline stages (e.g., after initial screening, before expensive DFT), pause and present candidate summary to researcher for approval/rejection/ modification, with suggested actions and risk assessment
  • 128.6.6.3 Implement compute budget management: track cumulative computational cost (CPU-hours for DFT, GPU-hours for ML, API calls for database queries), enforce budget limits per discovery session, prioritize candidates by expected information gain per compute cost (active learning acquisition function)
  • 128.6.6.4 Implement progress reporting and visualization: real-time dashboard showing discovery pipeline status, candidate count at each stage (funnel visualization), property distribution of surviving candidates, computation queue status, and estimated time to completion
  • 128.6.6.5 Implement candidate ranking engine: multi-objective ranking of candidates using Pareto frontier analysis (e.g., optimize both ZT and stability simultaneously), weighted scoring with user-defined property importance weights, and uncertainty-aware ranking (candidates with high predicted value but high uncertainty ranked for further computation)

128.6.7 Goal-Directed Discovery#

  • 128.6.7.1 Implement natural language goal parser: parse discovery goals like "Find a material with ZT > 2 at 600K containing earth-abundant elements" into structured constraints: target property (ZT), threshold (> 2), condition (T=600K), element filter (earth-abundant), with disambiguation dialog for ambiguous goals
  • 128.6.7.2 Implement constraint-to-workflow compiler: given structured constraints, generate an optimal discovery workflow: choose element palette, select structure prototypes, configure screening filters, select property prediction methods, define validation criteria — outputting a complete executable pipeline
  • 128.6.7.3 Implement "stable oxide semiconductor" discovery preset: pre-configured goal for finding stable oxide materials with band gap 1.5-3.0 eV, including element selection (Ti, Zn, Sn, In, Ga, etc.), structure prototypes (perovskite, spinel, delafossite, bixbyite), and screening criteria (E_hull < 25 meV/atom, no imaginary phonons)
  • 128.6.7.4 Implement "Li-ion conductor" discovery preset: pre-configured goal for solid electrolyte discovery, with constraints on ionic conductivity (sigma_Li > 1 mS/cm at 300K), electrochemical window (> 4V vs. Li/Li+), mechanical stability, and air/moisture stability
  • 128.6.7.5 Implement "high-entropy alloy" discovery preset: search for single-phase HEA compositions in 4-6 element spaces, with screening by Hume-Rothery rules (atomic size difference < 6.6%, electronegativity difference, VEC range), CALPHAD phase stability, and target mechanical properties (hardness, ductility)
  • 128.6.7.6 Implement discovery report generation: at pipeline completion, generate comprehensive report including: top candidates with properties, crystal structures (3D visualizations), stability analysis, synthesis recommendations, comparison with known materials, and literature context — formatted for research group presentation or publication SI

128.7 Collaborative Research Infrastructure (@kalika/materials-collab)#

128.7.1 Project Management#

  • 128.7.1.1 Implement materials project workspace: create, configure, and manage research projects containing structures, calculations, notebooks, and results — with project-level metadata (PI, funding, keywords), access control (owner, admin, contributor, viewer roles), and activity timeline
  • 128.7.1.2 Implement shared structure library: team-curated collection of crystal structures with tags (substrate, target, precursor, reference), search by composition/space group/property, version history per structure (track relaxation steps, manual edits), and import from CIF/POSCAR/XYZ files
  • 128.7.1.3 Implement calculation provenance tracking: for every computation, record: input structure (hash), method parameters (functional, basis set, k-mesh, convergence criteria), code version (VASP 6.4.1, QE 7.2, Kalika native), machine (hostname, cores, GPU), wall-time, and output files — enabling full reproducibility and audit trail
  • 128.7.1.4 Implement calculation workflow templates: save and share multi-step workflows (e.g., relax → SCF → band structure → DOS → phonon) as reusable templates, with parameter presets for different accuracy levels (fast screening, standard, high accuracy, benchmark)
  • 128.7.1.5 Implement dataset management: organize results into named datasets (e.g., "ternary oxide screening Q4 2026"), with metadata, DOI assignment, and export to standard formats (JSON, CSV, HDF5) for external sharing and archival

128.7.2 Comparison & Benchmarking#

  • 128.7.2.1 Implement cross-functional comparison: for the same material, compare results from different DFT functionals (LDA, PBE, PBEsol, SCAN, HSE06, PBE0) — side-by-side property tables, overlaid band structures and DOS, with experimental reference values highlighted and MAE/RMSE metrics
  • 128.7.2.2 Implement cross-code comparison: compare results from different DFT codes (VASP, Quantum ESPRESSO, ABINIT, Kalika native) for the same structure and equivalent settings — verifying consistency and identifying code-specific artifacts
  • 128.7.2.3 Implement reproducibility report: given two calculations of the same material with nominally identical settings, quantify agreement (energy difference per atom, force RMSE, band gap difference, phonon frequency MAE) and flag discrepancies above threshold — essential for computational materials science quality assurance
  • 128.7.2.4 Implement benchmark suite: curated set of materials with well-established experimental properties (Si, NaCl, MgO, Fe, Cu, GaAs, SrTiO3, LiFePO4), run standard calculations and compare with reference values — for validating new methods or code versions

128.7.3 Publication-Ready Output#

  • 128.7.3.1 Implement publication figure generator: export band structure, DOS, phonon dispersion, phase diagrams as publication-quality vector graphics (SVG/PDF) with journal-appropriate formatting (font sizes, line widths, color schemes for colorblind accessibility), axis labels with proper Unicode symbols and units
  • 128.7.3.2 Implement property table generator: generate formatted tables of lattice parameters (a, b, c, alpha, beta, gamma), band gaps (direct/ indirect), elastic constants (C_ij matrix), phonon frequencies at Gamma, effective masses, and other computed properties — in LaTeX, Markdown, or HTML format with configurable significant figures
  • 128.7.3.3 Implement supplementary information generator: compile convergence tests (k-mesh, cutoff energy), computational details (functional, pseudopotential, parameters), raw data tables, and additional figures into a structured SI document — in LaTeX or docx format
  • 128.7.3.4 Implement CIF file generator: export crystal structures in Crystallographic Information Framework format, with all symmetry operations, atomic positions (fractional), anisotropic displacement parameters, occupancy factors, and metadata (author, reference, data collection conditions) — for submission to ICSD, COD, or journal SI
  • 128.7.3.5 Implement data deposition formatter: prepare calculation results in NOMAD archive format (JSON-LD with NOMAD metainfo schema), Materials Cloud AiiDA archive format, or OPTIMADE API format — for FAIR data sharing and compliance with journal data availability requirements

128.7.4 Sharing & Collaboration#

  • 128.7.4.1 Implement notebook sharing: share materials science notebooks with collaborators via link, with permission levels (view only, comment, execute, edit), optional expiration date, and password protection for sensitive pre-publication data
  • 128.7.4.2 Implement result sharing: generate shareable link to specific calculation result set, including interactive 3D structure viewer, property plots, and raw data download — accessible without login for public results, authenticated for private results
  • 128.7.4.3 Implement Jupyter export: convert materials science notebooks to standard .ipynb format, with 3D viewers replaced by static images and interactive widgets replaced by matplotlib equivalents — for sharing with collaborators using Jupyter ecosystem
  • 128.7.4.4 Implement annotation and discussion: inline commenting on structures, plots, and notebook cells, with @mention notifications, threaded discussions, and resolution tracking — for collaborative interpretation of results

128.7.5 Experimental Characterization & Lab Data#

  • 128.7.5.1 Define SampleRecord, SynthesisBatch, CharacterizationRun, and FacilitySession types: encode sample lineage, processing history, operator, instrument/facility, calibration state, environment (temperature, pressure, gas, bias), and links to notebook and workflow artifacts — making experimental provenance first-class
  • 128.7.5.2 Implement experimental data import pipeline: ingest XRD/XRR, Raman/PL maps, SEM/TEM/STEM images, EBSD maps, atom probe tomography, and synchrotron/neutron beamline exports with raw files, reduced data, and instrument metadata preserved
  • 128.7.5.3 Implement microscopy and mapping analysis workspace: support annotation, segmentation, grain/defect/phase labeling, region-of-interest extraction, scale-bar calibration, and derived feature tables for EBSD, TEM/STEM, SEM, and spectroscopy maps
  • 128.7.5.4 Implement correlative experiment-simulation alignment: register experimental datasets to computed structures, diffraction/spectrum predictions, and phase-diagram regions — enabling "this image/spectrum is most consistent with this structure, defect chemistry, and processing path" workflows
  • 128.7.5.5 Implement operando and in situ experiment timelines: align time-resolved XRD, spectroscopy, microscopy, electrochemistry, or thermal measurements with simulation states, process steps, and agent decisions, with synchronized playback and event markers
  • 128.7.5.6 Implement ELN/LIMS/facility connectors: sync sample metadata, instrument bookings, beamtime IDs, run manifests, and access controls from external lab systems while preserving Kalika-side provenance and sharing permissions

128.8 Materials Science Knowledge Graph (@kalika/materials-graph)#

128.8.1 Materials Ontology#

  • 128.8.1.1 Implement materials ontology schema: define entity types (Material, Structure, Phase, Element, Property, Calculation, Paper, Researcher, SynthesisMethod), relationship types (has_structure, has_property, synthesized_by, computed_with, reported_in, isostructural_to, composed_of), and property types (scalar, tensor, spectrum, function of T/P) — compatible with EMMO (European Materials Modelling Ontology) and MatOnto standards
  • 128.8.1.2 Implement structure-property linkage graph: encode which structural features (bond lengths, coordination numbers, orbital overlap, band topology) determine which properties (band gap, conductivity, hardness, thermal expansion), with quantified correlation strengths from data mining across the knowledge graph
  • 128.8.1.3 Implement processing-structure-property triangle: encode relationships between synthesis/processing conditions (temperature, atmosphere, annealing, doping), resulting structure (phase, defects, grain size, texture), and measured/computed properties — enabling "given this processing, predict this structure and these properties" reasoning
  • 128.8.1.4 Implement material genealogy graph: track which materials inspired which — parent materials (structure prototypes), derivative materials (doped variants, solid solutions, surface modifications), and analogous materials (isoelectronic, isostructural) — enabling "materials inspired by perovskite SrTiO3" queries

128.8.2 Graph-Based Property Prediction#

  • 128.8.2.1 Implement compositional neighbor search: given a target composition, find materials in the knowledge graph with similar compositions (Earth mover's distance in element space), retrieve their known properties, and estimate target properties by composition-weighted interpolation with uncertainty
  • 128.8.2.2 Implement structural analog search: given a target structure, find materials with the same space group and similar Wyckoff positions, retrieve their properties, and predict target properties by structural analogy — leveraging the observation that isostructural materials often have similar property trends
  • 128.8.2.3 Implement isoelectronic compound prediction: identify isoelectronic substitutions (e.g., GaAs → ZnSe → CuBr), predict property changes along the isoelectronic series (band gap trend, effective mass trend), and suggest unexplored isoelectronic compounds
  • 128.8.2.4 Implement graph neural network property prediction: train GNN on the materials knowledge graph (material nodes with features, property edges with values) to predict missing properties, with attention mechanisms highlighting which graph neighbors are most informative for each prediction

128.8.3 Literature Integration#

  • 128.8.3.1 Implement paper-material linking: for each material in the graph, link to papers that report experimental or computational data, with extracted property values (band gap = 1.1 eV from Smith et al. 2024), experimental conditions (T=300K, single crystal), and confidence assessment (peer-reviewed journal, preprint, conference proceeding)
  • 128.8.3.2 Implement experimental property aggregation: for materials with multiple experimental reports, aggregate property values with uncertainty estimation (weighted mean, outlier detection, systematic error correction between measurement techniques), and present as recommended values with provenance trail
  • 128.8.3.3 Implement computational benchmark database: collect reported DFT/GW/DMFT results from literature for benchmark materials, organize by method and code, compute statistics (mean, std, range) across studies, and identify systematic biases (PBE underestimates band gaps by ~40% for standard semiconductors)

128.9 Educational & Onboarding (@kalika/materials-education)#

128.9.1 Interactive Tutorials#

  • 128.9.1.1 Implement "Your first DFT calculation" tutorial: guided notebook walking through silicon band structure calculation — import structure (diamond cubic Si), set up calculation parameters (PBE, PAW, 400 eV cutoff, 8x8x8 k-mesh), run SCF, compute band structure along Gamma-X-W-K-Gamma-L path, analyze results (indirect gap ~0.6 eV with PBE), compare with experiment (1.12 eV), discuss band gap problem
  • 128.9.1.2 Implement "Understanding phonons" tutorial: diamond phonon dispersion — build diamond cubic carbon structure, compute force constants (finite displacement or DFPT), construct dynamical matrix, diagonalize for phonon frequencies, plot dispersion and DOS, animate optical/acoustic modes, compute thermal properties C_v(T), compare with experiment
  • 128.9.1.3 Implement "Defect calculations 101" tutorial: silicon vacancy — generate supercell (2x2x2 or 3x3x3), create vacancy, relax structure, compute formation energy (neutral vacancy), introduce charge states (V^0, V^-, V^2-), compute charge transition levels, apply finite-size corrections (Freysoldt/Kumagai), plot transition level diagram
  • 128.9.1.4 Implement "High-throughput screening" tutorial: find stable ternary oxides in the A-B-O system (A=alkaline earth, B=transition metal) — define composition space, generate candidate structures from prototypes, compute formation energies, build convex hull, identify stable phases, compute properties of hits, rank by target metric
  • 128.9.1.5 Implement "ML potentials" tutorial: train a machine learning interatomic potential for silicon — generate training data (AIMD trajectories at multiple temperatures), extract atomic descriptors (SOAP, ACE), train GAP/NequIP/MACE model, validate (energy/force parity plots, phonon comparison), run MD with ML potential (1000x faster than AIMD), analyze thermal properties

128.9.2 Contextual Help & Documentation#

  • 128.9.2.1 Implement tooltips throughout UI: hover over any parameter (k-mesh density, cutoff energy, smearing width, convergence threshold) to see explanation, recommended range, and link to relevant documentation — context-sensitive, adapting to the current calculation type
  • 128.9.2.2 Implement method comparison guide: interactive decision tree for choosing computational method — "What property do you need?" → "What accuracy?" → "What system size?" → recommended method (DFT-PBE, HSE06, GW, DMFT, ML potential) with estimated computational cost and expected accuracy
  • 128.9.2.3 Implement materials science glossary: searchable dictionary of materials science terms (Brillouin zone, density of states, formation energy, convex hull, phonon, Fermi surface, etc.) with definitions, equations, diagrams, and cross-references — accessible from any point in the UI via Ctrl+Shift+G hotkey
  • 128.9.2.4 Implement error diagnosis assistant: when a calculation fails or produces suspicious results, analyze the output and suggest remedies — "SCF not converging" → suggest mixing parameter reduction, different diagonalization algorithm, or charge density restart; "Negative phonon frequencies" → suggest tighter relaxation, larger supercell, or symmetry reduction

128.10 Performance & Scalability (@kalika/materials-infra)#

128.10.1 Computation Queue Management#

  • 128.10.1.1 Implement local job queue: manage calculation jobs on the local machine with configurable parallelism (max concurrent DFT jobs, max CPU cores, max memory), priority queue (user-assigned priority + estimated wall-time), and job lifecycle management (submit, queue, run, complete, fail, cancel)
  • 128.10.1.2 Implement SLURM integration: submit calculations to SLURM- managed HPC clusters via SSH, with automatic job script generation (partition selection, resource request, module loading, input file staging), job monitoring (squeue polling), output retrieval, and error handling (resubmit on node failure)
  • 128.10.1.3 Implement PBS/Torque integration: equivalent to SLURM integration for PBS/Torque-managed clusters, with qsub script generation, qstat monitoring, and output retrieval
  • 128.10.1.4 Implement SGE integration: equivalent to SLURM integration for Sun Grid Engine clusters, with qsub/qstat/qdel support
  • 128.10.1.5 Implement job dependency management: define job chains where later jobs depend on earlier jobs completing successfully (e.g., relaxation → SCF → band structure), with automatic input file transfer between dependent jobs and rollback on failure

128.10.2 Cloud Compute Integration#

  • 128.10.2.1 Implement AWS compute integration: provision EC2 instances (CPU for DFT, GPU for ML) on demand, configure with required software (VASP, QE, PyTorch), submit jobs, monitor, retrieve results, and terminate instances when idle — with cost tracking ($/calculation) and budget alerts
  • 128.10.2.2 Implement GCP compute integration: equivalent to AWS for Google Cloud Platform Compute Engine instances, with preemptible/spot instance support for cost reduction on fault-tolerant calculations
  • 128.10.2.3 Implement Azure compute integration: equivalent to AWS for Azure Virtual Machines, with Azure Batch support for embarrassingly parallel screening workflows
  • 128.10.2.4 Implement hybrid local-cloud scheduling: intelligently route jobs to local cluster or cloud based on queue wait time, job urgency, cost budget, and data locality — with automatic failover from full local queue to cloud burst

128.10.3 Data Performance#

  • 128.10.3.1 Implement result caching system: content-addressable cache keyed on structure fingerprint (Niggli cell + sorted species + Wyckoff positions) + method parameters hash, with LRU eviction, configurable cache size, and cache hit statistics — avoiding redundant recomputation across users and projects
  • 128.10.3.2 Implement lazy loading for large datasets: charge density grids (millions of points), MD trajectories (thousands of frames × thousands of atoms), and phonon band structures (thousands of q-points) loaded on demand with streaming decompression, not upfront — keeping memory usage bounded
  • 128.10.3.3 Implement progressive crystal structure rendering: level-of- detail (LOD) system for large structures (> 10,000 atoms): at maximum zoom, render full detail (spheres, bonds, labels); at medium zoom, simplify to billboard sprites and reduced bond count; at minimum zoom, render as point cloud — maintaining 60fps regardless of atom count
  • 128.10.3.4 Implement WebSocket real-time updates: push calculation progress to dashboard in real time (SCF convergence per iteration, ionic relaxation steps, MD timestep counter), phonon computation progress (q-point completion), and discovery agent status — without polling, using persistent WebSocket connections

128.10.4 Import/Export & Interoperability#

  • 128.10.4.1 Implement CIF parser: read Crystallographic Information Framework files (CIF 1.1 and CIF 2.0), handling: cell parameters, atom positions (fractional and Cartesian), symmetry operations, space group, anisotropic displacement parameters, disorder/partial occupancy, and multi-block CIF files — with robust handling of non-standard CIF variants
  • 128.10.4.2 Implement POSCAR/CONTCAR parser: read VASP structure files (direct and Cartesian coordinates, selective dynamics flags, velocity block), with automatic species detection from POTCAR or header line
  • 128.10.4.3 Implement XYZ/extXYZ parser: read XYZ format (simple and extended), handling: lattice vectors in comment line (ASE convention), per-atom properties (forces, charges, velocities), and multi-frame trajectories
  • 128.10.4.4 Implement Quantum ESPRESSO input/output parser: read/write QE pw.x input files (CELL_PARAMETERS, ATOMIC_POSITIONS, ATOMIC_SPECIES, K_POINTS, &CONTROL/&SYSTEM/&ELECTRONS namelists), parse pw.x output for energies, forces, eigenvalues, and convergence data
  • 128.10.4.5 Implement VASP output parser: parse OUTCAR (energies, forces, stress, eigenvalues, DOS, dielectric function), vasprun.xml (comprehensive structured output), EIGENVAL (band structure), DOSCAR (density of states), CHGCAR (charge density), and WAVECAR (wavefunctions) — for importing existing VASP calculations into the platform
  • 128.10.4.6 Implement pymatgen interoperability: bidirectional conversion between Kalika CrystalStructure and pymatgen Structure/Molecule objects, enabling use of pymatgen's extensive analysis tools (symmetry analysis, structure matching, phase diagram construction) alongside Kalika's native capabilities