E Models in Integrability, Cosmology & Security

• E Models are mathematical constructs that define system dynamics via linear or involutive operators, unifying integrable systems, inflationary cosmology, and formal security methods.
• They employ rigorous algebraic and geometric techniques to derive integrability conditions, renormalization flow equations, and invariant observables across diverse fields.
• Extensions include degenerate and finite-dimensional generalizations, enabling applications from energy-based probabilistic modeling to secure, formal analysis of software systems.

E models are a class of mathematical constructs that appear prominently in disparate fields, most notably integrable systems, path-space statistical modeling, algebraic and geometric analysis of sigma models, supergravity inflationary scenarios, and even formal models of software security systems. The unifying conceptual thread is the specification of dynamics, measures, or transition rules through a (typically linear or involutive) operator E, which encodes symmetries, constraints, or energetics in the respective domain. This article presents a technical survey of E-models across these principal domains.

1. Lorentzian and Euclidean $\mathcal E$-Models in Integrable Sigma Models

The $\mathcal E$-model, as formulated by Klimčík and others, encodes integrable two-dimensional sigma models in a first-order formalism defined on the loop group $LD$ of a $2n$-dimensional real Lie group $D$ whose Lie algebra $\mathcal D$ admits an invariant symmetric bilinear form of split signature $(n,n)$. The dynamical action is

$$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$

where $E:\mathcal D\to\mathcal D$ is real-linear, symmetric with respect to $(\cdot,\cdot)_\mathcal D$, and satisfies $\mathcal E$0.

• For Lorentzian $\mathcal E$1-models ($\mathcal E$2), $\mathcal E$3 is required to be positive definite, producing real Lorentzian world-sheet actions for the associated sigma models.
• Euclidean $\mathcal E$4-models ($\mathcal E$5) are structurally distinct. The resulting second-order $\mathcal E$6-models possess real Euclidean world-sheet actions but fail to yield unitary Lorentzian spectra after Wick rotation. The bilinear form $\mathcal E$7 then has split signature (Klimcik, 22 Mar 2026).

After reduction to coset sigma models via integration over maximally isotropic subgroups, one obtains families of integrable models, including deformations such as bi-Yang–Baxter models. Classical integrability is guaranteed via the construction of a Lax pair, with the structure of $\mathcal E$8 entering crucially into the zero-curvature equations and Maillet $\mathcal E$9-matrices (Klimcik, 2023, Klimcik, 22 Mar 2026).

Renormalization group flow equations for $LD$0 are also well-defined; the one-loop beta function exhibits distinct behavior in Lorentzian and Euclidean cases. For $LD$1,

$LD$2

while $LD$3 yields an overall sign flip for the $LD$4 term.

2. Degenerate and Point Particle $LD$5-Model Generalizations

The $LD$6-model framework admits degenerate and finite-dimensional generalizations:

• Degenerate $LD$7-models, constructed from 4d Chern–Simons theory on $LD$8, appear when the involution $LD$9 has a nontrivial kernel. This leads to Hamiltonian systems on cosets $2n$0 via dressing coset geometry. The algebraic solution proceeds by decomposing the defect Lie algebra $2n$1 via $2n$2's eigenspaces, with projection operators $2n$3 solving the necessary constraints (Liniado et al., 2023).
• Point particle $2n$4-models correspond to reductions from stringy models, yielding dynamics on finite-dimensional symmetric spaces or flag manifolds (e.g., $2n$5), with Hamiltonians $2n$6. Their integrability, Lax structures, and physical interpretation as classical mechanical models are fully characterized by the same algebraic data $2n$7 (Klimcik, 2023).

These extensions systematically produce new classical integrable systems, including all classical $2n$8-models arising as dressing cosets from arbitrary meromorphic differentials $2n$9 on $D$0.

3. E-Models in $D$1-Attractor Inflation and Supergravity

In cosmology, the term E-model also denotes a class of single- and multi-field inflationary potentials, part of the $D$2–attractor framework. An E–model is realized in $D$3 supergravity using a nilpotent goldstino superfield $D$4 and an orthogonal inflaton multiplet $D$5, constrained by

$D$6

This constrains the spectrum to a single real inflaton scalar $D$7.

The Kähler potential and superpotential are

$D$8

for suitable constants $D$9 (the attractor parameter), $\mathcal D$0, $\mathcal D$1, and $\mathcal D$2 (Ellgan, 2020).

Modifying the superpotential by a constant shift ensures all derivatives of the F-term scalar potential vanish at the de Sitter vacuum. The E–model potential (after field redefinitions and imposing constraints) is

$\mathcal D$3

with $\mathcal D$4 tunable and $\mathcal D$5 a model parameter. Generalization to $\mathcal D$6 inflaton fields produces a multi-field potential, with canonical metrics on field space and analytic control of entropic/adiabatic-mode dynamics. Observables, such as the spectral index $\mathcal D$7, tensor-to-scalar ratio $\mathcal D$8, and isocurvature fraction $\mathcal D$9, are directly computable and consistent with Planck constraints for suitable choice of parameters (Ellgan, 2020).

4. Energy-Based Models (EBMs) for Functional Data (𝓕-EBM)

E-models in the context of probabilistic modeling include $(n,n)$0-EBMs, which extend energy-based models to function space. Given $(n,n)$1 realizations $(n,n)$2 of random functions in $(n,n)$3, an $(n,n)$4-EBM specifies a measure $(n,n)$5 by tilting a Gaussian process (GP) path measure $(n,n)$6: $(n,n)$7 with $(n,n)$8 parametrized via a latent generator map $(n,n)$9. A spectral (Karhunen–Loève) decomposition underlies the parameterization; expansion coefficients are regularized by the decay of GP kernel eigenvalues, enforcing Sobolev-type smoothness (Lim et al., 2022).

The model supports irregularly-sampled observations and naturally outputs predictions at arbitrary resolutions. The core training objective is marginal log-likelihood maximization: $$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$0 with gradients computed via contrastive-divergence–style schemes and regularized by the GP prior, ensuring non-overfitting between observed locations.

5. E-Net Models (Extended Petri Nets) in Formal Systems Security

E-Net models, or Extended Petri-Net models, constitute a formalism for describing and analyzing secure software system workflows, particularly in cryptographic protection for email systems. An E-Net is specified as a 7-tuple

$$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$1

with sets of places ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$2), peripheral places ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$3), permissive/inhibitor places ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$4), transitions ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$5), input/output arc incidences ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$6, $$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$7), and initial marking ($$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$8). Tokens traverse the net according to input/output matrices and event firing rules. The enablement of transitions depends not only on available tokens in input places but also on the presence of tokens in permissive places.

This structure can rigorously enforce properties such as confidentiality, integrity, and non-repudiation in transactional workflows by ensuring that cryptographic and logging operations cannot be bypassed and that audit trails are non-erasable (Stoianov et al., 2010).

6. The EP (E4) Model: Exotic Invariants in Supersymmetric Field Theory

The "E4" or EP model is a minimal supersymmetric field theory with two chiral supermultiplets $$S_E(\ell) = \int dt\,d\sigma\,\bigl(\langle\partial_t\ell\,\ell^{-1},\partial_\sigma\ell\,\ell^{-1}\rangle_\mathcal{D} -\langle\partial_\sigma\ell\,\ell^{-1},E(\partial_\sigma\ell\,\ell^{-1})\rangle_\mathcal{D} \bigr) +\Gamma_{\rm WZ}(\ell)$$9 and $E:\mathcal D\to\mathcal D$0, representing electron and positron-like fields, respectively. The action contains canonical kinetic terms, a mass superpotential

$E:\mathcal D\to\mathcal D$1

and an exotic invariant $E:\mathcal D\to\mathcal D$2 constructed to enforce a BRS-stable constraint: $E:\mathcal D\to\mathcal D$3 where $E:\mathcal D\to\mathcal D$4 are the fermionic components of $E:\mathcal D\to\mathcal D$5 and $E:\mathcal D\to\mathcal D$6. The constraint is solved by setting the fermions proportional. Completion terms at $E:\mathcal D\to\mathcal D$7 are required to maintain BRS invariance after including the exotic invariant in the action. This structure directly informs the more complicated exotic model extensions for supersymmetric Standard Model generalizations (Dixon, 4 Feb 2026).

7. E-Models in Supergravity: T/E Model Families and Inflationary Observables

The T/E model formalism systematically explores inflationary effective field theories within the context of supergravity and $E:\mathcal D\to\mathcal D$8-attractors. The distinction between T and E models lies in the respective forms of the superpotential and Kähler potential, though both are engineered to enforce orthogonality of nilpotent superfields and the decoupling of heavy sinflaton and inflatino states: $E:\mathcal D\to\mathcal D$9 This yields a class of models with universal predictions for $(\cdot,\cdot)_\mathcal D$0 and $(\cdot,\cdot)_\mathcal D$1 that interpolate between various single- and multi-field inflationary scenarios. The augmentation of the superpotential by specific constants can render the cosmological vacuum extremal in all directions, and the formalism is readily extended to multiple interacting fields with canonical kinetic terms and diagonal potential contributions (Ellgan, 2020).

---

References:

• Energy-Based Models for Functional Data using Path Measure Tilting (Lim et al., 2022)
• Integrable degenerate $(\cdot,\cdot)_\mathcal D$2-models from 4d Chern–Simons theory (Liniado et al., 2023)
• Point particle E-models (Klimcik, 2023)
• Euclidean E-models (Klimcik, 22 Mar 2026)
• Modification of T/E models and their multi-field versions (Ellgan, 2020)
• E-Net Models of a Software System for E-Mail Security (Stoianov et al., 2010)
• The EP Model and its Completion Terms (E4) (Dixon, 4 Feb 2026)