Skip to main content

%DeepSee.extensions.clusters.PAMSA

Class %DeepSee.extensions.clusters.PAMSA Extends %DeepSee.extensions.clusters.PAM [ System = 4 ]

This class provides an implemantation of Partitioning Around Medoids (PAM) algorithm adding Simulated Annealing (See Wikipedia article for more information) to avoid falling into a local minimum.

Properties​

T​

Property T As %Double [ Internal ];

TMin​

Property TMin As %Double [ Internal ];

TDen​

Property TDen As %Double [ InitialExpression = 1.1, Internal ];

Extensive​

Property Extensive As %Boolean [ InitialExpression = 0 ];

Methods​

New​

ClassMethod New(dsName As %String, Output sc As %Status) As PAMSA

Open​

ClassMethod Open(dsName As %String, Output sc As %Status) As PAMSA

Create​

ClassMethod Create(dsName As %String, new As %Boolean, Output sc As %Status) As PAMSA [ Private ]

Execute​

Method Execute() As %Status

cool​

Method cool() [ Internal ]

try​

Method try(Output accepted As %Boolean) As %Status [ Internal ]

tryDouble​

Method tryDouble(e0 As %Double, ByRef medoids, Output accepted As %Boolean) As %Status [ Internal ]

trySimple​

Method trySimple(e0 As %Double, Output accepted As %Boolean) As %Status [ Internal ]

AcceptReject​

Method AcceptReject(e0 As %Double, e1 As %Double) As %Boolean [ Internal ]

RollBackOnReject​

Method RollBackOnReject(medoids) [ Internal ]

TAssign​

Method TAssign() As %Status [ Internal ]

estimateInitialT​

Method estimateInitialT() [ Internal ]