Find evidence first. Spend computation second. Let physics decide.
VESPER (Validation Engine for Stellar Photometric Evidence and Recovery; named for the evening star) is a research program testing whether evidence-first routing can reduce the computational cost of exoplanet transit detection without sacrificing recall. Instead of folding every star's light curve at thousands of trial periods (the BLS/TLS approach), it detects individual transit-like events directly, infers the orbital period from their spacing, confirms candidates with a physics-based transit model, and reserves a full periodogram search only for the minority of stars showing no local evidence.
Status: Phase I — Scientific Validation (pre-registered). · Data: TESS. · Charter author: Ansul Suryawanshi.
| Path | Contents |
|---|---|
docs/ |
Canonical specifications and theory (start with docs/VESPER.md) |
CLAUDE.md |
Operating rules for AI-agent sessions (non-negotiables, doc map) |
REPOSITORY_GAP_ANALYSIS.md |
Critical cross-document review |
PHASE1_REMEDIATION.md |
Fix plan for the critical / must-fix findings |
src/ · data/ · research/ · results/ · notebooks/ · papers/ |
Project scaffold (no code yet — Phase I is validation-only) |
archive/ |
Prior-project forensic audit & scientific review (historical reference) |
docs/VESPER.md— the master charter (mission, philosophy, architecture, milestones).docs/SCIENTIFIC_HYPOTHESIS.md— the formal, falsifiable claims.docs/VESPER_PHASE1_VALIDATION.md— the pre-registered experiment.docs/VESPER_MATHEMATICAL_FOUNDATIONS.md— the theory.docs/VESPER_ARCHITECTURE.md— the full system design.
A transit search normally asks "is there a planet at period P?" ten-thousand times per star. VESPER asks once: "does this star show any dimming the noise can't explain?" — and only searches for a period where the evidence says to. The scientific test is whether that routing preserves recall (within 2 percentage points of full TLS) while measurably reducing compute.
Phase I is at document/specification stage. The immediate next step is to resolve the four critical/must-fix items in PHASE1_REMEDIATION.md, reissue the pre-registration as v2, then freeze the TESS sector/target manifest (milestone M0). No project code has been written yet — by design, validation comes first.
Math in these documents uses LaTeX ($…$); view in a math-aware Markdown renderer (Obsidian, VS Code, GitHub) for proper formatting.