goldilocks zones a fine-grained exoplanet taxonomy patrick j. talbot (presenter) dennis r. ellis...
DESCRIPTION
Objective ● Organize information about exoplanets ● Input exoplanets characteristics into a hierarchical, frame-based ontology, like Protege ● Identify the minimum set of characteristics that span the problem space, so that each exoplanet has, to the extent practical, a unique signature. ● Input the knowledge base into data analytics tools, such as Weka Rule Induction, to automatically discover interesting pattterns. ● Quantify missing and conflicting data as sources of uncertainty to drive further refinementTRANSCRIPT
Goldilocks Zones
– a Fine-Grained Exoplanet Taxonomy
Patrick J. Talbot (Presenter)
Dennis R. Ellis (Analysis)
Introduction and Summary
https://github.com/OpenExoplanetCatalogue/open_exoplanet_catalogue
* http://exoplanets.org/, as of 1 September 2015.
Situation: Since the 1990's, over 5,000exoplanets have been identified* and the rate ofdiscovery is accelerating:
1,569 Planets with good orbits + 24 Microlensing and imaged planets = 1,593 Total confirmed planets + 3,751 Unconfirmed Kepler candidates = 5,344 Total exoplanets + Kelper Candidates
Need: a robust, extensible organization schemefor understanding, processing, and pattern discovery
Approach:● A classic trade study to rank and score candidate attributes● A critic, a sequential optimizer, and software tests● Iteration to find a minimum spanning taxonomy
●Result:● A minimum spanning taxonomy provides a uniques set of characteristics for each exoplanet.● To the extent practical, each exoplanet has a unique set of characteristics, or signature.
Objective● Organize information about exoplanets● Input exoplanets characteristics into a hierarchical,
frame-based ontology, like Protege● Identify the minimum set of characteristics that span
the problem space, so that each exoplanet has, to the extent practical, a unique signature.
● Input the knowledge base into data analytics tools, such as Weka Rule Induction, to automatically discover interesting pattterns.
● Quantify missing and conflicting data as sources of uncertainty to drive further refinement
Approach
Compute Signatures
Unique?
Reduce #States
Increase #States
Yes
Minimal?
No
VerifyUniqueness
Done
Yes
Perform Trades● Mission● Functional Filter Attributes
● Birthday ProblemDiscretize● Rules
No
Tag Description UnitPlanet a single planet. May be a free floating (orphan) planetStar A single star. A star can be host to one or more planetsBinary two stars, star/binary or two binaries.Declination Declination +/- dd mm ssRightascension Right ascension hh mm ssDistance Distance from the Sun parsecName Used multiple times for objects with multiple names.Semimajoraxis Semi-major axis of a planet (heliocentric coordinates) AUSeparation Projected separation of planet from its host AU, arcsecPositionangle Position angle degreeEccentricity EccentricityPeriastron Longitude of periastron degreeLongitude Mean longitude at a given Epoch (all planets in a system) degreeMeananomaly Mean anomaly at a given Epoch (all planets in one system) degreeAscendingnode Longitude of the ascending node degreeInclinatioN Inclination of the orbit degreeEpoch Epoch for the orbital elements BJDPeriod Orbital period dayTransittime Time of the center of a transit BJDPeriastrontime Time of periastron BJDMass Mass (or m sin(i) for radial velocity planets) Jupiter/SolarRadius Physical radius Jupiter/SolarTemperature Temperature (surface or equilibrium) KelvinAge Age GyrMetallicity Stellar metallicity log, rel/ solarSpectraltype Spectral typeMagB B magnitudeMagV Visual magnitudeMagR R magnitudeMagI I magnitudeMagJ J magnitudeMagH H magnitudeMagK K magnitudeDiscoverymethod Discovery method : timing, RV, transit, imaging.Istransiting Whether the planet is transiting (1) or not (0).Description Short description of the planetDiscoveryyear Year of the planet's discovery yyyyLastupdate Date of the last (non-trivial) update yy/mm/ddSpinorbitalignment Rossiter-McLaughlin Effect. degree
Data Structures, MIT Open Exoplanet Catalog
Mission Trade Study
Functional Trade Study
Node Composite Score
Sample Rules
Probability of Duplicates vs Catalog Size
Exoplanet Characteristics HierarchyExoplanets
Elements Physical Host star
Period
Binary
Metallicity
Semi-major axis
SpectralType
Mass
Age
RadiusSeparation
Eccentricity
Inclination
Temperature
Distance
ExoplanetName
Input
Header Only Header Only Header Only
H H 2 M L S L N L O L A
Temperature(Hot)
Distance(Far)
Separation(Moderate)
Binary(2 stars)
Semi-major Axis(Large)
Mass(Small)
Inclination(Low)
Eccentricity(Nearly Circular)
Period(Long)
Spectral type(Hottest)
Physical Radius(Large)
Age(Ancient)
Exoplanet Signature
Physical Orbital
InfluentialEnvironmental
Size
Mass
Composition[FeHg]
Period
Inclination
Host Spectral TypeHost Age
Host Variability
Kepler 423cHabitabilityIndex = .3
GoldilocksZone
Exoplanet Kiviat Diagram
Coming in the Full Paper
● Percent of Duplicate Signatures
● Sensitivity of Duplicates to Attribute States
● Optimum Set of Attributes and # States
● Weka Rule Induction to Identify Patterns
Summary● Characteristics of an exoplanet taxonomy were
identified● Trade studies ranked attributes● Rules discretized attributes● Taxonomy was sized (Birthday Problem)● A fine-grained taxonomy was portrayed:
● Taxonomy Hierarchy● Taxonomy Signature● Taxonomy Kiviat
Backup:Example Histograms