Skip to content

qgrav roadmap

A high-level view of where qgrav is and where it is headed. For the detailed, version-by-version history see CHANGELOG.md.

Current status

qgrav is a working, validated simulation and analysis platform:

  • Emergent-gravity simulation. The Mach-Zehnder gravity phase is computed from a ballistic atom trajectory under a chirped Raman laser, not injected analytically, and every result carries an explicit study-scope label (fully simulated / hybrid / analytical only).
  • Sub-pulse finite-τ Raman integration (v1.5, opt-in via raman_substeps). Finite-duration pulses are stepped in sub-intervals with ballistic free-fall between slices; the result converges to the Bertoldi 2019 closed form within 2×10⁻³ relative, so finite-pulse physics is computed rather than predicted.
  • Realistic measurement cycle. A multi-drop simulator with a per-shot noise budget (seismic vibration, detection noise, Raman-phase noise, AC Stark, wavefront aberrations) and a fringe-lock servo (PID with anti-windup) produces ASD and Allan-deviation curves like a real instrument. Seismic vibration is drop-to-drop correlated: one continuous Peterson NLNM/NHNM series spans the campaign and every drop samples its three pulse times from it.
  • Emergent quantum projection noise (v1.5). In multi-drop mode the shot-noise floor emerges from Binomial(N_det, P) single-atom statistics and matches the analytic σ_g = 1/(√N_det·k_eff·T²) to 0.2 %.
  • Emergent gravity-gradient physics. GravityFreePropagator integrates a linear gradient to first order in γ, and the emergent Mach-Zehnder gradient phase reproduces the textbook k_eff·γ·T²·[(z₀−z_ref) + v₀T − (7/12)gT²] closed form (Peters/Chung/Chu 2001) to ~0.1 % — regression-locked in tests/test_gravity_gradient_emergent.py.
  • Time-domain mirror motion (vibration_model: "time_domain"). The vibration sweep can drive the pulse sequence with a synthesized Peterson NLNM/NHNM displacement record instead of a sinusoidal amplitude axis; the realized per-shot phase noise reproduces the frequency-domain Cheinet budget ∫S_a·|H_a|²df end-to-end (ratio 0.92–1.13 across seeds, truth-checked on every run).
  • Validation. Automated regressions reproduce the short-term sensitivity of five published transportable gravimeters (Freier 2016 as the primary target, plus Hu 2013, Ménoret 2018, Xu 2022, and Wu 2019), and an independent QuTiP integration cross-checks the single-pulse Raman dynamics.
  • Real data. Superconducting-gravimeter (IGETS) time series can be ingested through the analysis pipeline.
  • Tooling. A six-tab desktop GUI, a single-YAML configuration schema, auto-generated HTML reports, continuous integration on Linux and Windows, and a documentation site.

Near-term directions

  • Sub-pulse integration by default. raman_substeps (v1.5) already computes finite-τ pulse physics numerically; the remaining step is making it the default and retiring the empirical pulse-timing calibration entirely.
  • Four- and five-pulse sequences for gradiometry and differential gravimetry.
  • Measured-wavefront input. Drive the Zernike wavefront model from real wavefront-sensor coefficients.
  • Real interferometer output. Ingest raw fringe / CSV output from an interferometer through the existing analysis pipeline, where such data is available.

Longer-term vision

  • Bayesian estimation of g from the multi-drop time series with a full ensemble likelihood.
  • Hardware bench. When a real apparatus is available, ingest live photodiode data through the same interface as the virtual bench and run a closure test: compare the measured Allan deviation against the qgrav prediction. This is the long-term goal, validating the simulation against a specific instrument.

Contributing

Contributions are welcome. See CONTRIBUTING.md for the development setup, test commands, and the project's guiding principles.