Bi-parametric Kneading Scans

• Bi-parametric kneading scans are a method that maps system behaviors in a two-dimensional parameter space using kneading invariants and symbolic encoding.
• The approach utilizes high-resolution computational grids to reveal organizing structures such as T-points and separating saddles, enhancing insights in fields from material science to medical imaging.
• It enables empirical data collapse and process optimization by correlating performance indicators like pore volume and surface area with operational parameters.

Bi-parametric kneading scans constitute a systematic approach to mapping complex behaviors, mixing characteristics, or material properties of systems as functions of two independent parameters. This paradigm enables high-resolution exploration and visualization of intrinsic structures, organizing centers, or performance landscapes in domains as varied as dynamical systems, material processing, and medical imaging.

1. Theoretical Foundations

In symbolic dynamics, bi-parametric kneading scans leverage the concept of a kneading invariant to encode the itinerary of trajectories in a dynamical system. For a one-dimensional return map or a Poincaré map with a critical point, the critical orbit produces a symbolic sequence $s = s_0 s_1 s_2 \dots$, with $s_n \in \{+1, -1\}$. The corresponding kneading sequence is defined as the truncated power series $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$ for $z \in (0,1)$.

Milnor–Thurston theory establishes a direct link between the kneading series/determinant and dynamical entropy, with the smallest positive root of the determinant providing the topological entropy via $h_\mathrm{top} = -\ln r$ or, equivalently, through $h_\mathrm{top} = \ln(1/q^*)$ where $q^*$ is the root of $P(q) = \sum s_n q^n$. Fluctuations in the itinerary signal changes in entropy and finely delineate fractal structures in parameter space (Barrio et al., 2012, Xing et al., 2013).

For practical applications in material science and engineering, systems often exhibit responses dependent on two operational or processing parameters (e.g., accumulated deformation and pH in the kneading of alumina pastes). Here, bi-parametric scans map measured metrics—such as specific surface area or pore volume—as continuous functions over the two-dimensional control parameter space, exposing design-optimal regimes or critical transitions (Auxois et al., 2024).

2. Methodological Pipeline

The pipeline for constructing a bi-parametric kneading scan in symbolic dynamical systems comprises the following stages (Barrio et al., 2012, Xing et al., 2013):

• Define a two-parameter family $ẋ = F(x; \mu, \nu)$ with parameters $(\mu, \nu)$.
• For each mesh point $s_n \in \{+1, -1\}$0 in a high-resolution grid:
  • Initialize the trajectory near a critical manifold or saddle point.
  • Trace the forward separatrix, recording the sign assignment at each Poincaré return or wing-turn; build the first $s_n \in \{+1, -1\}$1 symbols $s_n \in \{+1, -1\}$2.
  • Compute the truncated kneading invariant $s_n \in \{+1, -1\}$3 for fixed $s_n \in \{+1, -1\}$4.
  • Map $s_n \in \{+1, -1\}$5 and assign a color for visualization.
• Iterate over all grid points (typically $s_n \in \{+1, -1\}$6 or finer).

For physical systems, the general framework is adapted accordingly:

• In granular paste processing (Auxois et al., 2024), the two parameters are accumulated deformation $s_n \in \{+1, -1\}$7 and neutralization ratio $s_n \in \{+1, -1\}$8 (proxy for pH). Textural properties (e.g., $s_n \in \{+1, -1\}$9, $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$0) are mapped versus these controls by sampling aliquots at specific process intervals, measured via nitrogen adsorption or Hg porosimetry.
• In the context of mixing in extruders, disk-stagger angle $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$1 and pitched-tip angle $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$2 parametrize the kneading geometry, while performance is captured through metrics such as pressure drop, stress distributions, or tracer mixing (Nakayama et al., 2010).
• For MR-compatible mechanical loading, normal and shear loading forces define the scan axes, and tissue deformations are measured through MRI and image registration protocols (Trebbi et al., 2021).

3. Interpretation and Structural Features

Bi-parametric kneading scans reveal several classes of organizing structures, most notably:

• Codimension-Two T-Points: These are organizing centers visible as spiral focal points in level-set diagrams of the kneading invariant. T-points represent closed heteroclinic cycles connecting key saddles and saddle-foci, with their accumulation creating a hierarchy of nested spirals in parameter space (Barrio et al., 2012, Xing et al., 2013).
• Separating Saddles: Signaled by sharp discontinuities or folds in the kneading color map, corresponding to parameter values where the symbolic itinerary and thus the kneading invariant change discontinuously. These arise from abrupt transitions in the global manifold structure.
• Fractality and Universality: Both three-dimensional (e.g., Lorenz, Shimizu–Morioka) and higher-dimensional (e.g., 6D laser) systems display nearly identical spiral assemblages, with T-points surrounded by self-similar, nested spirals and lacunae, evidencing a universal topological classification by kneading invariants (Barrio et al., 2012, Xing et al., 2013).

Similar melting-point structures and performance transitions are observed in engineering and biomedical implementations, where the interplay of control parameters leads to sharp regime changes or master-curve data collapse (Auxois et al., 2024, Nakayama et al., 2010).

4. Implementation and Computational Aspects

Symbolic kneading scans require careful balancing of computational resolution and symbol depth:

• Typical computational grids employ $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$3 or finer resolution.
• Kneading depth $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$4–$\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$5 captures sufficient symbolic detail for structural discrimination.
• The weighing factor $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$6 is selected (usually $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$7) to balance early and late symbols' influence (Xing et al., 2013).

Filtering strategies consist of removing parameter regions with non-chaotic (trivial) regimes and accentuating zones of rapid kneading change. In practice, the computational time is order-linear in the number of grid points and kneading depth.

For experimental process characterization (Auxois et al., 2024):

• Scans span operational parameters (e.g., $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$8 revolutions, $\kappa_N(z) = \sum_{n=0}^{N-1} s_n z^n$9).
• Textural data (surface area, pore volume) are recorded at pre-defined intervals and subjected to tests of empirical collapse versus a joint parameter, $z \in (0,1)$0.

For device-driven bi-parametric scans (biomechanical, extruder, MRI):

• Physical parameters are controlled in calibrated increments, with system outputs or physical displacements recorded at each combinatorial setting (Trebbi et al., 2021, Nakayama et al., 2010, Liu et al., 2023).

5. Applications Across Domains

Bi-parametric kneading scans are deployed in diverse contexts:

• Parametric chaos diagnostics: Visualization and classification of organizing centers, bifurcation sets, and fractal boundaries in models such as Lorenz, Shimizu–Morioka, and multi-level lasers (Barrio et al., 2012, Xing et al., 2013).
• Materials engineering: Rational formulation of catalyst supports via empirical master curves that relate mechanical work and chemical environment to pore network development (Auxois et al., 2024). The bi-parametric mapping enables the targeting of specific microstructure by selecting appropriate process recipes.
• Process engineering (extrusion): Optimization of mixing machine configuration (disk-stagger and tip angles) via bi-parametric evaluation of mixing uniformity and pressure drop, guiding design toward balanced distributive/dispersive regimes (Nakayama et al., 2010).
• Biomechanical tissue response: Quantitative internal mapping of tissue deformations under controlled force application, providing ground truth for model validation in finite element simulations (Trebbi et al., 2021).
• Quantitative MRI: Bi-parametric quantitative MRI protocols (such as 3D MR-STAT adapted for T1/T2 mapping) allow accelerated acquisition and parameter mapping over spatial domains, with protocol details specifiable for a desired bi-parametric regime (Liu et al., 2023).

6. Data Collapse and Empirical Reduction

A distinguishing feature in the materials context is the demonstration that textural trends across a multidimensional experimental space collapse onto master curves when plotted versus an appropriately constructed joint parameter. For kneading of boehmite pastes (Auxois et al., 2024), the empirical combination $z \in (0,1)$1 unifies all observed trends:

• $z \in (0,1)$2
• $z \in (0,1)$3

This empirical reduction strongly streamlines process design: for a desired pore structure or surface area, any parameter pair yielding the target $z \in (0,1)$4 is valid, minimizing experimental trial-and-error.

7. Extensions and Universality

The kneading scan approach generalizes to higher-dimensional dynamical systems, provided a symbolic partition of the critical trajectory is implementable. For systems with more than two symbolic domains, multiple kneading invariants can be computed in parallel (Xing et al., 2013). Universality of the emerging structures across models demonstrates the global utility of these invariants for classifying structurally unstable regimes (Barrio et al., 2012, Xing et al., 2013).

The technique also supports real-time parameter estimation (by decoding observed sequences back to coordinates in parameter space), direct support for bifurcation theory, and adaptive control in laboratory or industrial environments. In MRI, the explicit bi-parametric protocol design ensures that image data directly encodes the simultaneous measurement of two key tissue parameters, optimizing information acquisition and workflow (Liu et al., 2023).