3.5 Table clusters

The cluster is a group of observations with the common covariance matrix. The covariance matrix allows to express any combination of correlations among observations in cluster (including uncorrelated observations, where covariance matrix is diagonal). For explanation see Observation data and points.

In the database observations are stored in three tables: obs, coordinates and vectors. Cluster’s covariance matrix is stored in table covmat. Every observation, vector or coordinate in database has to be in some cluster.

• conf_id is id of configuration which cluster belongs to.
• ccluster identifies a cluster within one configuration.
• dim and band specify dimension and bandwidth of covariance matrix. The bandwidth of the diagonal matrix is equal to 0 and a fully-populated covariance matrix has a bandwidth of dim-1 (band maximum possible value is dim-1).
• tag specifies type of observations in cluster which also implies the table where they are stored in. obs and height-differences stand for obs table, coordinates and vectors stand for coordinates table and vectors table respectively.

Observations, vectors and coordinates are identified by configuration id (conf_id), cluster id ccluster and theirs index (indx). Observation index (indx) has to be unique within observations of one cluster (which belongs to one configuration). The same applies for vectors and coordinates.

See also Set of observations.

Example (table contents):

conf_id|ccluster|dim|band|tag
-----------------------------
1      |1       |3  |0   |obs
1      |4       |4  |0   |obs

This document was generated by Ales Cepek on June 5, 2013 using texi2html 1.82.