Atoms for new UCDs

In the scheme discussed in RefactoringUCDs, UCDs will be composed -- by application writers and end-users, not by committee -- from atomic parts. The atoms remain to be chosen. This document is a list of possible atoms.

Notation

I have written out the atoms as strings of words linked by hyphens. In the examples of composed UCDS, I separate the atoms with forward slashes.

The words are given in full so as to make the atoms as clear as possible without needing supporting notes. This, I hope, helps the discussion of the selection of atoms. When the atoms are chosen, we might also choose standard abbreviations for some of the words.

Grammar

A UCD is like a phrase in natural language, but written in a contracted form, with the prepositions removed. E.g. we might represent "the central wavelength of the V band in the Johnson photometric-system" as "length/absolute/wavelength/johnson-v"; we could then re-expand that UCD as "a length which is an absolute value in its own right (as opposed to an offset from or an uncertainty in some other measurement); and is also a wavelength (of light); and is also to do with the V-band of the Johnson system."

In this system of composition, there is one primary noun -- length -- and a number of complementary words and phrases. The noun is chosen to have strongly-defined meaning by itself and the complements explain the context in which the noun is used. An atom can be a primary noun if it has an "is-a" relationship with the data it describes; otherwise, it is a complement. "X is a flux" is meaningful. "Y is a Johnson V-band" isn't proper usage, and one should write instead something like "Y is a flux measured in the Johnson V-band." Hence, "Johnson V-band is always a complement.

The primary grammatical rule for UCDs seems to be that they should include exactly one primary noun and zero or more complements. The noun is the basis of defintion and, for numeric data, identifies the physical dimensions of the quantity.

Some complements can qualify any numeric data and apply within and without astronomy: e.g. mean, upper-limit, uncertainty. Others apply generally to non-numeric data: e.g. member-of-set, alphabetic. Other complements apply only in specific contexts in astronomy. The general complements are a small set and the contextual complements a much larger one. The general complements have a large bearing on the usage of the data and the contextual complements generally have a lesser effect. E.g. the distinction between measured value and uncertainty for flux measurements is more important than distinctions between photometric systems. General complements tend to be mutally exclusive.

This suggests a second rule for UCDs: there should be zero or one general complements in a UCD; if zero, then a general complement is assumed: absolute value for numeric data and arbitrary string for non-numeric data.

Examples of atoms

This list was derived by scanning through the current list of UCDs and noting down words that looked significant. I have made no particular effort to make the list complete; that effort is not really worthwhile until we agree the general principals for choosing and combining atoms. In particular, the atoms naming passbands for photometry have been left incomplete to save time and space. I guess that of the atoms needed to represent the ~1,500 UCDs in the old, hierarchical system, roughly half are listed here.

Primary nouns for quantities

• frequency
• length
• time
• angle
• velocity
• angular-velocity
• count
• spatial-frequency
• area
• volume
• solid-angle
• temperature
• angular-scale
• energy
• power
• force
• energy-flux
• number-flux
• energy-flux-density
• count-rate
• intensity
• number-density
• mass-density
• pressure
• luminosity

Primary nouns for qualities

• classification
• name
• code
• reference

General complements for numeric data

• absolute (i.e. value, e.g. central wavelength of spectral bin)
• relative (e.g. angular distance from search centre)
• extent (e.g. width of spectral bin)
• upper-limit
• lower-limit
• uncertainty
• maximum (i.e. among a statistical set of values)
• minimum (i.e. among a statistical set of values)
• mean
• median
• mode
• ratio (NB: changes dimensionality of the numbers w.r.t. the noun)

General complements for non-numeric data

• String (i.e. any character data)
• member-of-set
• alphabetic
• alphanumeric
• uniform-resource-identifier

Complements that are adjectives

• ruling
• alternative
• instrumental
• modelled
• observed
• radial
• tangental
• lunar
• solar
• planetary
• stellar
• galactic
• terrestial
• celestial
• absolute
• normalized
• estimated
• linear
• logarithmic
• ecliptic
• bibliographic
• integrated
• comoving
• barycentric
• geocentric
• heliocentric

Complements that are or include nouns

• data-quality
• abell-richness
• abell-distance
• major-axis
• minor-axis
• full-width-at-half-maximum-signal
• fit-to-model
• residual
• author
• region-on-sky
• candidate-identification
• catalogue
• chart
• constellation
• cross-identification
• database
• observation
• location (e.g. of observatory)
• survey
• table
• data-set
• file
• plate (i.e. photographic)
• group
• antenna
• aperture
• background
• passband
• baseline (i.e. of interferometer)
• beam (e.g. of radio telescope)
• calibration
• correction
• line-width
• emulsion (e.g. describing photograph)
• signal-to-noise
• point-spread-function
• pixel
• responsive-quantum-efficiency
• detective-quantum-efficiency
• spectral-order
• wavelength
• occultation
• continuum
• extinction
• morphology
• spiral-arms
• nutation
• precession
• spectral-line
• method
• altitude (e.g. of observatory)
• longitude
• latitude
• eccentricity
• comet
• elongation
• periastron
• perihelion
• orbital-period
• phase
• radius
• diameter
• absolute-magnitude (i.e. in photometry only)
• apparent-magnitude (i.e. in photometry only)
• colour-index (i.e. in photometry only)
• cousins-i
• cousins-r
• optical
• radio
• infrared
• johnson-b
• johnson-v
• johnson-r
• visual
• profile
• sky
• stromgren-b
• stromgren-r
• stromgren-y
• stromgren-h-beta
• tycho-b
• tycho-v
• element-abundance
• opacity
• optical-depth
• reflectance
• refractive-index
• separation
• star-formation
• wind
• polarization
• linear-polarization
• circular-polarization
• position-angle
• direction-cosine
• x-axis
• y-axis
• z-axis
• parallax
• zenith-distance
• data-reduction
• balmer-line
• equivalent-width (of spectral line)
• spectral-feature
• spectral-index
• mk-spectral-type
• rejection
• percentage
• completeness
• correlation
• degrees-of-freedom
• sampling
• telescope
• focal-length
• julian-date
• resolution
• units
• amplitude (of quasi-periodic signal)
• period
• barycentre
• escape-velocity
• total
• right-ascension
• declination
• proper-motion

Examples of composed UCDs (with English translations)

angle/right-ascension

right ascension of a point on the sky. POS_EQ_RA.

angle/right-ascension/ruling

right ascension of a point on the sky; preferred value from a set of otherwise-equivalent measurements. POS_EQ_RA_MAIN.

angle/declination

declination of a point in the sky. POS_EQ_DEC.

angular-velocity/declination/proper-motion

component of proper motion in the declination direction. POS_EQ_PMDEC.

energy-flux-density/johnson-v

flux-density derived from photometry in the Johnson V-band. Not expressible in old system: PHOT_JHN_V implies a value on a magnitude scale, not a linear measurement of flux.

energy-flux-density/uncertainty/johnson-v

measurement/processing uncertainty in flux-density derived from Johnson-V-band photometry. Not expressible in old system: there are no UCDs in that system for uncertainties on specific quantities.

length/wavelength/johnson-v

effective or central wavelength of a bin in an SED. INST_PASS_BAND?

length/wavelength/johnson-v

effective or central wavelength of the V-band in the Johnson system. Not expressible in old system: INST_BAND_PASS comes close, but there is no way to make this specific to a particular band.

length/wavelength/extent/full-width-at-half-maximum-signal

bandwidth to half power of a bin in an SED. INST_BANDWIDTH?.

length/wavelength/extent/full-width-at-half-maximum-signal/johnson-v

bandwidth to half power of Johnson V-band. Not expressible in old system. INST_BANDWIDTH comes closest, but there is no way to specify that this is a FWHM value, nor to specify which band is being described.

energy-flux/ratio/linear/sloan-r/johnson-i

Ratio of fluxes of obtained from Sloan r-band and Johnson I-band measurements. Not expressible in old system.

energy-flux/ratio/colour-index/johnson-r/johnson-i

r-I colour, using johnson system. PHOT_JHN_R-I.

energy-flux/ratio/colour-index/sloan-r/johnson-i

r-I colour, using mixed filter system. Not expressible in old system: PHOT_JHN_R-I is close but implies a Johnson R-magnitude.

energy-flux-density/circular-polarization

Flux density of circularly-polarized light. POL_FLUX_CIRC.

energy-flux-density/circular-polarization/johnson-v

Flux density of circularly-polarized light measured in Johnson V-band. Not expressibl in old system: needs a combination of POL_FLUX_CIRC and PHOT-JHN-V.

Counting the atoms

There are 198 atoms in the lists above, derived ~1,500 UCDs in the hierarchical system. If my guess at 50% completeness of the atoms is accurate, then we have nearly a four-fold reduction in the number of terms needed to maintain the status quo.

In addition, the atoms listed above allow many UCDs to be formed that do not exist in the old system. Some of these UCDs (e.g. those describing bins in a composite SED) are urgently needed for AVO's January demonstration.

-- GuyRixon - 07 Oct 2002