project: Project

Description Usage Arguments Details Value See Also

View source: R/curve.R

Description

project

Usage

1
2
    project(c1, c2, signif = 0.05, eps=1e-5, fuzz=1e-14,
        show.details=FALSE, warn=TRUE)

Arguments

c1

curvature summary. See details.

c2

curvature summary. See details.

signif

passed to curvatureStats.

eps

significance tolerance for singular values. See details.

fuzz

tolerance for comparing parameter points. See details.

show.details

logical indicating if detailed calculations should be shown.

warn

see effectiveCurvature.

Details

Under Development

c1 and c2 should be curvature summary statistics as returned by curvature. It is assummed that c1 is a submodel of c2. The tangent space (parameter effects curvature) of the sub-model c1 is a subspace of the tangent space of the larger model. The acceleration space (first normal space, intrinsic curvature effects) of the sub-model is a subspace of the direct sum of the tangent and acceleration spaces of the larger model, so the intrinsic Tangent and acceleration vectors of the submodel c1 can be projected onto the tangent and acceration spaces of the larger model c2. These are called T1inT2, N1inT2, and N1inN2 (T for tangent, N for normal). A second projection (restriction) of c2 onto the tangent and acceleration spaces of c1 is less interesting but may be a useful check. The intrinsic curvature of the larger model should also be intrinsic on the sub-model and the parameter effects of the larger model may be partly intrinsic on the sub-model. These two projected onto the intrinsic curvature space of the submodel (N2andT2inN1) should compare with the intrinsic curvature of the submodel.

Singular values smaller than eps times the largest singular value are considered to be zero when calculating the dimension of the tangent space.

If the parameter points for the two models are not within fuzz then a warning is issued to indicate that they do not represent the same point in parameter space.

Value

xxx

See Also

curvature effectiveCurvature


curve documentation built on May 2, 2019, 5:31 p.m.

Related to project in curve...