Definition of class pPCA in RvtkStatismo

I finally made up my mind about the exact contents of the S3 class pPCA (see my manual, page 6).

It is now specified as a list containing the following required entries:

• PCA: a list containing
  • sdev: the square roots of the covariance matrix’ eigenvalues
  • rotation matrix containing the orthonormal PCBasis vectors
  • x the scores within the latent space (scaled by 1/sdev)

  • center a vector of the mean shape with coordinates ordered

    (x1,y1,z1, x2, y2,z2, ..., xn,yn,zn)

• scale: logical. indicates whether the data were aligned including scaling

• representer : an object of class mesh3d or a list with entry vb being a matrix with the columns containing coordinates and it a 0x0 matrix
• sigma: the noise estimation of the data
• Variance: a data.frame containing the Variance, cumulative Variance and Variance explained by each Principal component

And optionally (and missing if data is imported using statismoLoadModel)

• rawdata: optional data: a matrix with rows containing the mean centred coordinates in order

   ```(x1,y1,z1, x2, y2,z2, ..., xn,yn,zn)```
  

I also spent quite some time, ironing out minor flaws and writing documentation.

TODO:

the class pPCAconstr should be a either a subclass of pPCA or removed and replaced by functions manipulating pPCA, which would, however, be less convenient. We’ll see…, it is probably the best way to implement the constrained model from statismo first, and then look if things play nicely together.