Iopt
- class pyoptex.doe.fixed_structure.splitk_plot.metric.Iopt(n=10000, cov=None, complete=True)[source]
The I-optimality criterion. Computes the average (average) prediction variance if multiple Vinv are provided.
Attributes
- momentsnp.array(2d)
The moments matrix.
- samplesnp.array(2d)
The covariate expanded samples for the moments matrix.
- nint
The number of samples.
- Minvnp.array(3d)
The inverse of the information matrix. Used as a cache.
- Mupnp.array(3d)
The update to the inverse of the information matrix. Used as a cache.
- __init__(n=10000, cov=None, complete=True)[source]
Creates the metric
Parameters
- nint
The number of samples to compute the moments matrix.
- covfunc(Y, X)
The covariance function
- completebool
Whether to only use the coordinates or completely randomly initialize the samples to generate the moments matrix.
Methods
Iopt.accepted(Y, X, params, update)Updates the internal state when the updated design was accepted (and therefore better).
Iopt.call(Y, X, params)Computes the I-optimality criterion.
Iopt.init(Y, X, params)Initializes the metric for each random initialization of the coordinate-exchange algorithm.
Iopt.preinit(params)Pre-initializes the metric
Iopt.update(Y, X, params, update)Computes the update to the metric according to update.