Skip to content
DraconiticResearch
Search Sources API
197,688 studies31 sourcesUpdated 30 min ago
Peer-reviewedCC BY 4.0International Journal of Molecular Sciences

Structure-Preserving Excess Gibbs Learning for Multicomponent Phase Equilibria: A Falsifiable and Uncertainty-Aware Synthetic Study

  • Pedro Robles1
  • Luis Rios-Colque2
  • Vanesa Bazan3,4
  • Luis Rojas-Valdivia2
4 affiliations
  1. 1Escuela de Ingeniería Química, Pontificia Universidad Católica de Valparaíso, Valparaíso 2340000, Chile
  2. 2Facultad de Ingeniería, Pontificia Universidad Católica de Valparaíso, Valparaíso 2340000, Chile
  3. 3Instituto de Investigaciones Mineras (IIM), Facultad de Ingeniería, Universidad Nacional de San Juan, San Juan 5400, Argentina
  4. 4Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), San Juan 5400, Argentina
Read full textPDFDOICite

Abstract

This fully synthetic study tests a structure-preserving surrogate on NRTL-generated phase equilibrium systems; it contains no experimental molecular validation. A symmetric neural potential represents GE/(nRT), from which activity coefficients, thermal derivatives, binodals, flashes, and tangent-plane-distance diagnostics are derived. Across ten binary training seeds, its median lnγ MAE was 1.16×10−3, better than matched direct regression (2.40×10−3) but not Redlich–Kister (9.65×10−4) or correctly specified refitted NRTL (2.36×10−4). Gibbs–Duhem residuals decayed as O(h2) under grid refinement, confirming diagnostic truncation around an analytic identity. A 41,600-point binary hull gave endpoint MAE 2.31×10−4. On 100 ternary type-I feeds, phase-count agreement with numerical reference labels was 98.0% (exact 95% interval, 93.0–99.8%), with two false negatives after recovery of one infeasible-floor solver failure. A 20-member ensemble showed pooled error ranking (ρ=0.894) but weak in-domain ranking (ρ=0.164); a temperature–distance heuristic was stronger (ρ=0.914), and nominal 90% conformal coverage fell from 88.0% in-domain to 0.0% under shift. Controlled flashes were 2.77× slower than NRTL. A hard thermodynamic structure is therefore valuable within neural modeling, but classical dominance, shift sensitivity, and absent molecular data bound the claim.

Related

  1. Magnetic control of lipid–peroxyl termination yield under spin relaxation: from an optimised dipolar-scale field to a falsifiable kinetic criterion
    Pablo R. Castello, Pablo Jimenez, Daniel R. Kattnig +1
    ChemRxivPreprintCC BYFull textBiophysicsChemRxiv
  2. Nonnative Thiol and Aniline Methylation, Including Enantioselective Dithiol Desymmetrization, by Multicomponent Cobamide‐Dependent Methyltransferases Using Unactivated Methyl Donors
    Amardeep Kumar, Osman Ishag Adam, Paras Gupta +1
    Angewandte Chemie (International Ed. in English)Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationPMC
  3. Recovery of a minor cryo-EM particle population reveals conformational equilibria linking cofactor loading, turnover, and reactivation in methionine synthase
    H. Wang, M. B. Watkins, P. Suksmith +1
    bioRxivPreprintCC BY-NC-NDFull textStructural biologybioRxiv
  4. Quantifying RNA Folding Equilibria from Unclassified Cryo-EM Particle Images
    D. Maji, L. Dingeldein, M. Li +9
    bioRxivPreprintCC BYFull textMolecular biology & gene regulationbioRxiv
  5. Innovative Approach Reveals How Receptors Control G Protein State Equilibria
    Alexandre Bouyssou, Mélanie da Silva, Julien Soulé +10
    Advanced SciencePeer-reviewedCC BY 4.0Full textGeneral biochemistryGeneral biologyHALEurope PMC

Cite

BibTeXRIS
BibTeX
@article{robles2026structurepreserving,
  title = {{Structure-Preserving Excess Gibbs Learning for Multicomponent Phase Equilibria: A Falsifiable and Uncertainty-Aware Synthetic Study}},
  author = {Pedro Robles and Luis Rios-Colque and Vanesa Bazan and Luis Rojas-Valdivia},
  journal = {International Journal of Molecular Sciences},
  year = {2026},
  month = sep,
  doi = {10.3390/ijms27198767},
  url = {https://www.mdpi.com/1422-0067/27/19/8767},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - Structure-Preserving Excess Gibbs Learning for Multicomponent Phase Equilibria: A Falsifiable and Uncertainty-Aware Synthetic Study
AU  - Pedro Robles
AU  - Luis Rios-Colque
AU  - Vanesa Bazan
AU  - Luis Rojas-Valdivia
PY  - 2026
DA  - 2026/09/30
JO  - International Journal of Molecular Sciences
DO  - 10.3390/ijms27198767
UR  - https://www.mdpi.com/1422-0067/27/19/8767
AB  - This fully synthetic study tests a structure-preserving surrogate on NRTL-generated phase equilibrium systems; it contains no experimental molecular validation. A symmetric neural potential represents GE/(nRT), from which activity coefficients, thermal derivatives, binodals, flashes, and tangent-plane-distance diagnostics are derived. Across ten binary training seeds, its median lnγ MAE was 1.16×10−3, better than matched direct regression (2.40×10−3) but not Redlich–Kister (9.65×10−4) or correctly specified refitted NRTL (2.36×10−4). Gibbs–Duhem residuals decayed as O(h2) under grid refinement, confirming diagnostic truncation around an analytic identity. A 41,600-point binary hull gave endpoint MAE 2.31×10−4. On 100 ternary type-I feeds, phase-count agreement with numerical reference labels was 98.0% (exact 95% interval, 93.0–99.8%), with two false negatives after recovery of one infeasible-floor solver failure. A 20-member ensemble showed pooled error ranking (ρ=0.894) but weak in-domain ranking (ρ=0.164); a temperature–distance heuristic was stronger (ρ=0.914), and nominal 90% conformal coverage fell from 88.0% in-domain to 0.0% under shift. Controlled flashes were 2.77× slower than NRTL. A hard thermodynamic structure is therefore valuable within neural modeling, but classical dominance, shift sensitivity, and absent molecular data bound the claim.
SN  - 1422-0067
ER  -