randRangeFinder

Carry out principal component analysis (PCA) of very large pedigrees such as found
in breeding populations! This package exploits sparse matrices and randomised
linear algebra to deliver a gazillion-times speed-up compared to naive singular value decoposition (SVD)
(and eigen decomposition).

Hannes Becher

randPedPCA

Fast PCA for Large Pedigrees

Hanbin Lee

Ros Craddock

Gregor Gorjanc

randRangeFinder function

<dl><dt>L</dt>
<dd>a pedigree's L inverse matrix in sparse 'spam' format</dd>
<dt>rank</dt>
<dd>An <code>integer</code>, how many principal components to return</dd>
<dt>depth</dt>
<dd><code>integer</code>, number of iterations for generating the range matrix</dd>
<dt>numVectors</dt>
<dd>An <code>integer &gt; rank</code>, to specify the oversampling for the</dd>
<dt>cent</dt>
<dd><code>logical</code> whether or not to (implicitly) 'centre' the additive
relationship matrix, or more precisely, its underlying 'data matrix' L</dd></dl>

Arguments

Generate range matrix for SVD — randRangeFinder

<dl>

<dt>L</dt>
<dd>a pedigree's L inverse matrix in sparse 'spam' format</dd>


<dt>rank</dt>
<dd>An <code>integer</code>, how many principal components to return</dd>


<dt>depth</dt>
<dd><code>integer</code>, number of iterations for generating the range matrix</dd>


<dt>numVectors</dt>
<dd>An <code>integer &gt; rank</code>, to specify the oversampling for the</dd>


<dt>cent</dt>
<dd><code>logical</code> whether or not to (implicitly) 'centre' the additive
relationship matrix, or more precisely, its underlying 'data matrix' L</dd>

</dl>

randRangeFinder: Generate range matrix for SVD

Description

Usage

Value

Arguments