apodex/FrontierChallenge

1# Input data dictionary2 3Input is mounted at `/app/input/` in the agent container (copied from `data/` at build time).4 5## Files6 7| File | Rows / count | Description |8|------|--------------|--------------|9| `metadata.csv` | 9 | One row per XRD scan |10| `xrd_peak_table.csv` | 63 | Deconvolved peak table (27 gamma + 27 alpha + 9 Si) |11| `phase_reference.csv` | 6 | h,k,l and RIR weight factor per alpha/gamma peak |12| `analysis_constants.json` | — | Cu Kα1 wavelength, Si 111 reference 2θ, formulas |13| `patterns/*.csv` | 9 files | Raw XRD patterns (two_theta_deg, intensity_counts), for visualization only |14| `README_input.md` | — | Annotator's input structure notes |15 16## `metadata.csv`17 18| Column | Description |19|--------|--------------|20| `scan_id` | Unique scan ID (e.g. `DUAL_LASER_R1`) |21| `sample_id` | `DUAL_LASER`, `LPBF_FAST`, or `SA_1100C` — 3 scans each |22| `condition` | Human-readable processing condition |23| `replicate` | {1, 2, 3} |24| `pattern_file` | Relative path under `/app/input/patterns/` |25 26## `xrd_peak_table.csv`27 28| Column | Units | Description | Observed range (computed) |29|--------|-------|--------------|----------------------------|30| `scan_id` | — | Links to metadata | — |31| `phase` | categorical | `Si` (9 rows, internal standard), `alpha` (27), `gamma` (27) | — |32| `hkl` | — | Miller indices, e.g. `111` | — |33| `center_2theta_raw_deg` | degree 2θ | Peak center, uncorrected | 28.39–82.21 |34| `integrated_area_counts_deg` | counts·deg | Peak area | 215.46–1215.40 |35| `fwhm_deg` | degree | Peak width | — |36 37Each scan has exactly one `Si` peak (111, zero-shift reference only — **must not** enter38phase-fraction or lattice-constant statistics), 3 `gamma` peaks (111/200/220), and 339`alpha` peaks (110/200/211). The gamma(111) and alpha(110) peaks overlap near 44° in the40raw pattern; this table already ships them deconvolved as separate rows — do not re-pick41peaks from the raw pattern's tallest local maximum near 44°.42 43## `phase_reference.csv`44 45| phase | hkl | h | k | l | rir_weight_factor |46|-------|-----|---|---|---|--------------------|47| gamma | 111 | 1 | 1 | 1 | 1.000 |48| gamma | 200 | 2 | 0 | 0 | 0.575 |49| gamma | 220 | 2 | 2 | 0 | 0.390 |50| alpha | 110 | 1 | 1 | 0 | 1.000 |51| alpha | 200 | 2 | 0 | 0 | 0.285 |52| alpha | 211 | 2 | 1 | 1 | 0.345 |53 54## `analysis_constants.json`55 56| Key | Value |57|-----|-------|58| `cu_kalpha1_wavelength_A` | 1.5406 |59| `si_111_reference_2theta_deg` | 28.442 |60| `zero_shift_formula` | `zero_shift_deg = measured_Si111_center - 28.442`; `corrected_2theta = raw - zero_shift` |61| `lattice_formula` | `d = lambda/(2*sin(theta))`; `a = d*sqrt(h^2+k^2+l^2)`, `theta = corrected_2theta/2` |62| `phase_fraction_formula` | `sum(area/rir_weight_factor)` by phase; `gamma_wt_pct = 100*gamma/(gamma+alpha)` |63 64**Replicate-variance note:** the 9 scans (3 conditions × 3 replicates) carry real65measurement noise, so the per-sample `sd_gamma_wt_pct`/`sd_alpha_wt_pct` you compute66should be nonzero and the mean/SD/n summary is statistically meaningful, not trivial.67 68Provenance: synthetic/de-identified duplex stainless steel XRD exercise (`Liu MZ-2`).69