Discretized Fast–Slow Systems with Canards in Two Dimensions

We study the problem of preservation of maximal canards for time discretized fast–slow systems with canard fold points. In order to ensure such preservation, certain favorable structure-preserving properties of the discretization scheme are required. Conventional schemes do not possess such properties. We perform a detailed analysis for an unconventional discretization scheme due to Kahan. The analysis uses the blow-up method to deal with the loss of normal hyperbolicity at the canard point. We show that the structure-preserving properties of the Kahan discretization for quadratic vector fields imply a similar result as in continuous time, guaranteeing the occurrence of maximal canards between attracting and repelling slow manifolds upon variation of a bifurcation parameter. The proof is based on a Melnikov computation along an invariant separating curve, which organizes the dynamics of the map similarly to the ODE problem.

Locality-Preserving Partial Least Squares Regression

This chapter proposes another nonlinear PLS method, named as locality-preserving partial least squares (LPPLS), which embeds the nonlinear degenerative and structure-preserving properties of LPP into the PLS model. The core of LPPLS is to replace the role of PCA in PLS with LPP. When extracting the principal components of $$\boldsymbol{t}_i$$ t i and $$\boldsymbol{u}_i$$ u i , two conditions must satisfy: (1) $$\boldsymbol{t}_i$$ t i and $$\boldsymbol{u}_i$$ u i retain the most information about the local nonlinear structure of their respective data sets. (2) The correlation between $$\boldsymbol{t}_i$$ t i and $$\boldsymbol{u}_i$$ u i is the largest. Finally, a quality-related monitoring strategy is established based on LPPLS.

Scientific Variables

Despite their centrality to the scientific enterprise, both the nature of scientific variables and their relation to inductive inference remain obscure. I suggest that scientific variables should be viewed as equivalence classes of sets of physical states mapped to representations (often real numbers) in a structure preserving fashion, and argue that most scientific variables introduced to expand the degrees of freedom in terms of which we describe the world can be seen as products of an algorithmic inductive inference first identified by William W. Rozeboom. This inference algorithm depends upon a notion of natural kind previously left unexplicated. By appealing to dynamical kinds—equivalence classes of causal system characterized by the interventions which commute with their time evolution—to fill this gap, we attain a complete algorithm. I demonstrate the efficacy of this algorithm in a series of experiments involving the percolation of water through granular soils that result in the induction of three novel variables. Finally, I argue that variables obtained through this sort of inductive inference are guaranteed to satisfy a variety of norms that in turn suit them for use in further scientific inferences.