Notation & Glossary

A consolidated lookup for every symbol and term used across the site. Symbols defined on their source pages are still listed here — this page exists so you never have to hunt for where a definition first appeared.

Physics notation

Symbol Meaning Where introduced
\(\Phi\) Newtonian gravitational potential (negative near a mass); \(-M/r\) for a point mass Clocks as Gravimeters
\(G\), \(c\) Newton’s constant and the speed of light; the simulation sets \(G = c = 1\) Units and Scales
\(\tau\), \(t\) Proper time vs coordinate time; a clock’s tick rate is \(d\tau/dt\) Clocks as Gravimeters
\(r\) Bare \(r\) is always a distance on this site Clocks as Gravimeters
\(r_c\) Tick rate of clock \(c\) in the filter equations — same letter as distance, distinguished by the clock index The Particle Filter
\(r_{ij}\) Distance from clock \(i\) to mass \(j\) Into the Plane
\(r_s\) Schwarzschild radius, \(r_s = 2M\) in simulation units Units and Scales
\(R\) Range from the clock lattice to an exterior mass; the differential signal falls as \(1/R^2\), the curvature term as \(1/R^3\) Gravitational Echolocation
\(x\), \(y\) Mass position coordinates (subscripted \(x_1, x_2\) with several masses) The Search in One Dimension (\(x\)); Into the Plane (\(y\))
\(M\), \(M_j\) Mass (of mass \(j\)) Clocks as Gravimeters
\(\mu\), \(\sigma_{\text{density}}\), \(A\) Center, width, and peak amplitude of a continuous Gaussian mass profile Beyond Point Masses

Index conventions — context-dependent: \(i\) indexes clocks/evaluation points in the potential sum (\(r_{ij}\)) but particles in the filter equations (\(w_i\), \(\theta_i\)); \(j\) indexes masses; \(c\) indexes clocks (\(r_c\)).

Inference notation

Symbol Meaning Where introduced
\(\sigma_{\text{obs}}\) Observation noise standard deviation (0.005 in the demos); written plain \(\sigma\) on the inference pages One Clock Is Not Enough
\(N\) Number of particles in the filter’s cloud The Particle Filter
\(w_i\) Weight of particle \(i\) (weights sum to one) The Particle Filter
\(\theta_i\) Particle \(i\)’s full parameter hypothesis, e.g. \((x, M)\) The Particle Filter
\(\beta\) Tempering exponent for the newest observation, advanced adaptively from 0 to 1 The Particle Filter
\(\pi_{t,\beta}\) Intermediate target after observations before \(t\) and a fraction \(\beta\) of observation \(t\) The Particle Filter
\(K\) Number of masses a model assumes How Many Masses?
ESS Effective sample size, \(1/\sum_i w_i^2\) The Particle Filter
evidence, log-evidence SMC Monte Carlo estimate of a model’s marginal likelihood, accumulated before resampling at each tempering stage How Many Masses?
NoteTwo overloaded symbols

\(\sigma\) means observation noise on the inference pages but Gaussian profile width in Beyond Point Masses (which flags the reuse itself); this page writes \(\sigma_{\text{obs}}\) vs \(\sigma_{\text{density}}\) to keep them apart. \(\tau\) means proper time in \(d\tau/dt\), but in the echolocation coda it is a clock’s averaging time: there \(\sigma_y(\tau) = 10^{-16}/\sqrt{\tau}\) is the clock’s fractional-frequency instability after averaging for time \(\tau\).

Terms

  • Weak field — the regime where gravity is a small correction (\(|2\Phi/c^2| \ll 1\)) and the tick rate \(\sqrt{1 + 2\Phi/c^2}\) is used as a pedagogical surrogate; this project enforces the conservative policy \(|2\Phi| \le 0.1\) in simulation units; see Clocks as Gravimeters.
  • Time dilation — clocks deeper in a gravitational well tick slower relative to distant ones; the measurable effect this whole site is built on (Clocks as Gravimeters).
  • Chronometric leveling — surveying height differences by comparing clock rates, as in the Tokyo Skytree experiment (Clocks as Gravimeters).
  • Forward model — the physics direction: from a mass configuration to the clock rates it produces (Clocks as Gravimeters).
  • Inverse problem — the detection direction: from noisy clock rates back to the masses that caused them (concept: One Clock Is Not Enough).
  • Mass–distance degeneracy — a small mass nearby and a heavy mass far away can produce identical readings; reduced or broken by sufficiently informative array geometry (One Clock Is Not Enough).
  • Particle filter — the site’s inference engine: adaptive tempered resample-move SMC over a cloud of hypotheses (The Particle Filter).
  • Prior — what the filter assumes before any data: a uniform rectangle over the configured position and mass ranges, conditioned on ordering and weak-field validity (The Particle Filter).
  • Likelihood — how well a hypothesis predicts an observed set of clock rates; Gaussian in the observation noise (The Particle Filter).
  • Posterior — the belief after data. The particle cloud approximates it; the cloud’s spread is the filter’s claimed uncertainty (The Search in One Dimension).
  • Evidence — the marginal likelihood a model assigned to the data it actually saw, estimated consistently by SMC but with finite-particle sampling error; what model comparison ranks (The Particle Filter).
  • Resampling — redrawing the cloud in proportion to weight at an adaptive tempering stage; systematic, stratified, or residual (The Particle Filter).
  • Rejuvenation — symmetric random-walk Metropolis-Hastings moves after resampling; accept/reject preserves the current tempered target (The Particle Filter).
  • Effective sample size\(1/\sum_i w_i^2\); how many particles are meaningfully alive (The Particle Filter).
  • Model comparison — running one filter per candidate mass count \(K\) and comparing evidence (How Many Masses?).
  • Label switching — with several masses the exchangeable posterior is symmetric under relabeling; the library represents its normalized form strictly ordered by the first spatial coordinate (Two Hidden Masses).
  • Observation noise — the Gaussian noise \(\sigma_{\text{obs}}\) added to every clock reading; the floor that makes many observations necessary (One Clock Is Not Enough).