Where does anomaly collide?

In celestial mechanics, an anomaly is an angle that fixes where a body sits along its orbit, reckoned from the point of closest approach. Kepler's 1609 Astronomia nova gave the relation now called Kepler's equation, tying the steadily increasing mean anomaly (a uniform timekeeping angle) to the eccentric anomaly and then to the true anomaly, the body's real angular position; the word for irregular celestial motion is older still, used by Ptolemy for the uneven apparent motion of the planets. Murray and Dermott's 1999 Solar system dynamics is a standard modern treatment. Here the referent is a geometric angle in an orbit, not a data point that breaks a pattern, a gap in measured gravity, or an observation a theory cannot absorb.[1][2]

Unit

An angle in an orbit, in degrees or radians measured from perihelion

Recovery means

Not applicable

Measured by

Computed from orbital elements by solving Kepler's equation, or derived from an observed position

In geophysics and geodesy, a gravity anomaly is the difference between gravity measured at a point and the value predicted by a reference model of the Earth, after standard corrections. The idea traces to Bouguer's 1749 La figure de la terre, which reported that gravity near the Andes fell short of expectation because the mountains attracted less than their bulk implied, the observation behind the correction and the reduction that still bear his name. Hackney and Featherstone's 2003 review sets out how geodesists and geophysicists define the quantity in different ways. The referent is a mismatch between measured and modelled gravity over the ground, not an orbital angle, a rejected data point, or a fact a framework cannot hold.[3][4]

Unit

A gap between observed and model gravity, in milligal

Recovery means

Not applicable

Measured by

Observed gravity from a gravimeter minus a reference value, after free-air, Bouguer, and terrain corrections

In the philosophy and history of science, an anomaly is an observation or result that the reigning framework cannot accommodate. Kuhn's 1962 The structure of scientific revolutions put the term at the centre: normal science proceeds by puzzle solving until anomalies accumulate that resist the accepted paradigm, and a crisis can then open the way to a new one. The referent is a violation of expectation within a body of theory, judged against what practitioners take for granted, not a geometric angle in an orbit, a shortfall in measured gravity, or a lone deviant number in a dataset.[5]

Unit

An observation a prevailing framework cannot explain

Recovery means

Not applicable

Measured by

Identified qualitatively, as a persistent mismatch between an accepted theory's predictions and observed results

In statistics and machine learning, an anomaly (or outlier) is a data point that departs markedly from the pattern of the rest of the data. Grubbs's 1969 Procedures for detecting outlying observations in samples gave formal tests for flagging such points under a distributional assumption; Chandola, Banerjee, and Kumar's 2009 survey Anomaly detection mapped the many methods that now carry the name across domains. The referent is an aberrant observation relative to a fitted model or distribution, not an angle locating a body in orbit, a gravity mismatch, or an observation that overturns a theory.[6][7]

Unit

A single observation far from the bulk of the data

Recovery means

Not applicable

Measured by

Flagged by statistical tests, distance or density thresholds, or model reconstruction error