journal article·2026-01Open access

A generalized dual potential for inelastic Constitutive Artificial Neural Networks: A JAX implementation at finite strains

Hagen Holthusen, Kevin Linka, Ellen Kuhl, Tim Brepols

Abstract

We present a methodology for designing a generalized dual potential, or pseudo potential, for inelastic Constitutive Artificial Neural Networks (iCANNs). This potential, expressed in terms of stress invariants, inherently satisfies thermodynamic consistency for large deformations. In comparison to our previous work, the new potential captures a broader spectrum of material behaviors, including pressure-sensitive inelasticity. To this end, we revisit the underlying thermodynamic framework of iCANNs for finite strain inelasticity and derive conditions for constructing a convex, zero-valued, and non-negative dual potential. To embed these principles in a neural network, we detail the architecture’s design, ensuring a priori compliance with thermodynamics. To evaluate the proposed architecture, we study its performance and limitations discovering visco-elastic material behavior, though the method is not limited to visco-elasticity. In this context, we investigate different aspects in the strategy of discovering inelastic materials. Our results indicate that the novel architecture robustly discovers interpretable models and parameters, while autonomously revealing the degree of inelasticity. The iCANN framework, implemented in JAX , is publicly accessible at https://doi.org/10.5281/zenodo.14894687 .
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Model Reduction and Neural Networks · Neural Networks and Applications · Elasticity and Material Modeling

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

References (100+)

Show source-provided references
1. Abdolazizi · 2025
2. Viscoelastic constitutive artificial neural networks (vcanns) – A framework for data-driven anisotropic nonlinear finite viscoelasticity10.1016/j.jcp.2023.112704 ↗
3. Input convex neural networks
4. On the robustness and efficiency of integration algorithms for a 3D finite strain phenomenological SMA constitutive model10.1002/nme.2964 ↗
5. A mechanics-informed artificial neural network approach in data-driven constitutive modeling10.1002/nme.6957 ↗
6. A mechanics-informed deep learning framework for data-driven nonlinear viscoelasticity10.1016/j.cma.2023.116463 ↗
7. A mechanics-informed neural network framework for data-driven nonlinear viscoelasticity
8. A neural network-based enrichment of reproducing kernel approximation for modeling brittle fracture10.1016/j.cma.2023.116590 ↗
9. The effect of lode parameter on the yield surface of schoen’s iwp triply periodic minimal surface lattice10.1016/j.mechmat.2022.104473 ↗
10. A Bayesian-based approach for constitutive model selection and calibration using diverse material responses10.1016/j.actamat.2025.120796 ↗
11. Algorithms for hyper-parameter optimization
12. A framework for data-driven analysis of materials under uncertainty: Countering the curse of dimensionality10.1016/j.cma.2017.03.037 ↗
13. Hyperparameter optimization: Foundations, algorithms, best practices, and open challenges
14. Boes · 2024
15. Bradbury · 2018
16. Strength of concrete under combined stresses
17. A class of methods for solving nonlinear simultaneous equations10.1090/S0025-5718-1965-0198670-6 ↗
18. Thermodynamic formulation of constitutive equations and application to the viscoplasticity and viscoelasticity of metals and polymers10.1016/S0020-7683(96)00162-X ↗
19. Thermodynamics with internal state variables10.1063/1.1711937 ↗
20. Foundations of linear viscoelasticity10.1103/RevModPhys.33.239 ↗
21. The thermodynamics of elastic materials with heat conduction and viscosity10.1007/BF01262690 ↗
22. Collins · 1993
23. A thermomechanical analysis of a family of soil models10.1680/geot.2002.52.7.507 ↗
24. On the theoretical and numerical modelling of armstrong–frederick kinematic hardening in the finite strain regime10.1016/j.cma.2003.09.005 ↗
25. Soil mechanics and plastic analysis or limit design10.1090/qam/48291 ↗
26. The thermodynamics of irreversible processes. IV. The theory of elasticity and anelasticity10.1103/PhysRev.73.373 ↗
27. Model-free data-driven inelasticity10.1016/j.cma.2019.02.016 ↗
28. NN-EVP: A physics informed neural network-based elasto-viscoplastic framework for predictions of grain size-aware flow response10.1016/j.ijplas.2024.104072 ↗
29. Finding structure in time10.1207/s15516709cog1402_1 ↗
30. Agent-based collaborative random search for hyperparameter tuning and global function optimization10.3390/systems11050228 ↗
31. Ferle · 2024
32. Discovering plasticity models without stress data10.1038/s41524-022-00752-4 ↗
33. Automated discovery of generalized standard material models with EUCLID10.1016/j.cma.2022.115867 ↗
34. Flaschel · 2024
35. A review on data-driven constitutive laws for solids
36. Fuhg · 2024
37. Functional concepts in continuum mechanics10.1023/A:1004304224398 ↗
38. Continuum thermodynamics10.1115/1.3167184 ↗
39. Knowledge-based modeling of material behavior with neural networks10.1061/(ASCE)0733-9399(1991)117:1(132) ↗
40. Ghouli · 2025
41. Girardi · 2018
42. Does convexity of yield surfaces in plasticity have a physical significance?10.1177/1081286517721599 ↗
43. Sur les matériaux standard généralisés
44. Polyconvexity of generalized polynomial-type hyperelastic strain energy functions for near-incompressibility10.1016/S0020-7683(03)00086-6 ↗
45. Automated model discovery for tensional homeostasis: Constitutive machine learning in growth and remodeling10.1016/j.compbiomed.2025.109691 ↗
46. Theory and implementation of inelastic constitutive artificial neural networks10.1016/j.cma.2024.117063 ↗
47. Polyconvex inelastic constitutive artificial neural networks10.1002/pamm.202400032 ↗
48. Inelastic material formulations based on a co-rotated intermediate configuration—Application to bioengineered tissues10.1016/j.jmps.2022.105174 ↗
49. A novel anisotropic stress-driven model for bioengineered tissues accounting for remodeling and reorientation based on homeostatic surfaces10.1002/pamm.202200015 ↗
50. A thermomechanical framework for analysis of microstructural evolution: Application to olivine rocks at high temperature10.1029/2018JB015613 ↗
51. Multilayer feedforward networks are universal approximators10.1016/0893-6080(89)90020-8 ↗
52. Experimental study and numerical modelling of VHB 4910 polymer10.1016/j.commatsci.2012.02.027 ↗
53. Closed-form representation of matrix functions in the formulation of nonlinear material models10.1016/j.finel.2015.12.002 ↗
54. Automated model discovery of finite strain elastoplasticity from uniaxial experiments10.1016/j.cma.2024.117653 ↗
55. Neural network-based multiscale modeling of finite strain magneto-elasticity with relaxed convexity criteria10.1016/j.cma.2023.116739 ↗
56. Hypersparse neural network analysis of large-scale internet traffic
57. Kestin · 1969 · 275
58. Kingma · 2017
59. Parametrized polyconvex hyperelasticity with physics-augmented neural networks10.1017/dce.2023.21 ↗
60. Allgemeine kontinuumstheorie der versetzungen und eigenspannungen10.1007/BF00281393 ↗
61. Computational homogenization of nonlinear elastic materials using neural networks10.1002/nme.4953 ↗
62. Lemaitre · 1994
63. Potential-based constitutive models for cohesive interfaces: Theory, implementation and examples10.1016/j.compositesb.2014.08.024 ↗
64. Neural networks meet hyperelasticity: A guide to enforcing physics10.1016/j.jmps.2023.105363 ↗
65. Constitutive artificial neural networks: A fast and general approach to predictive data-driven constitutive modeling by deep learning10.1016/j.jcp.2020.110010 ↗
66. Discovering uncertainty: Bayesian constitutive artificial neural networks10.1016/j.cma.2024.117517 ↗
67. A new family of constitutive artificial neural networks towards automated model discovery10.1016/j.cma.2022.115731 ↗
68. Best-in-class modeling: A novel strategy to discover constitutive models for soft matter systems10.1016/j.eml.2024.102181 ↗
69. Multiscale modeling of inelastic materials with thermodynamics-based artificial neural networks (TANN)10.1016/j.cma.2022.115190 ↗
70. Evolution TANN and the identification of internal variables and evolution equations in solid mechanics10.1016/j.jmps.2023.105245 ↗
71. Thermodynamics-based artificial neural networks for constitutive modeling10.1016/j.jmps.2020.104277 ↗
72. A constitutive model for anisotropic damage in fiber-composites10.1016/0167-6636(94)00053-0 ↗
73. On sparse regression, L-regularization, and automated model discovery10.1002/nme.7481 ↗
74. Deep learning predicts path-dependent plasticity10.1073/pnas.1911815116 ↗
75. A machine learning approach to predict in vivo skin growth10.1038/s41598-024-67056-z ↗
76. Hyperparameter optimization with approximate gradient
77. Stress update algorithm for elastoplastic models with nonconvex yield surfaces10.1002/nme.2407 ↗
78. A POD-TANN approach for the multiscale modeling of materials and macro-element derivation in geomechanics10.1002/nag.3891 ↗
79. Direct data-driven algorithms for multiscale mechanics10.1016/j.cma.2024.117525 ↗
80. Model-free data-driven inference in computational mechanics10.1016/j.cma.2022.115704 ↗
81. A theory of finite viscoelasticity and numerical aspects10.1016/S0020-7683(97)00217-5 ↗
82. Stress-dependent finite growth in soft elastic tissues10.1016/0021-9290(94)90021-3 ↗
83. Viscoelasticty with physics-augmented neural networks: model formulation and training methods without prescribed internal variables10.1007/s00466-024-02477-1 ↗
84. A mathematical theory of communication10.1002/j.1538-7305.1948.tb01338.x ↗
85. A fast algorithm for finding the first intersection with a non-convex yield surface10.1016/j.compgeo.2011.02.008 ↗
86. Adjustment of an inverse matrix corresponding to a change in one element of a given matrix10.1214/aoms/1177729893 ↗
87. UN modele viscoelastique non lineaire avec configuration intermediaire
88. Part III. Theory of invariants
89. Flow and fracture of materials according to a new limiting condition of yelding10.1007/BF02128173 ↗
90. Completely automated CNN architecture design based on blocks10.1109/TNNLS.2019.2919608 ↗
91. On the modelling of anisotropic elastic and inelastic material behaviour at large deformation10.1016/S0020-7683(01)00140-8 ↗
92. Data-driven continuum damage mechanics with built-in physics10.1016/j.eml.2024.102220 ↗
93. Data-driven tissue mechanics with polyconvex neural ordinary differential equations10.1016/j.cma.2022.115248 ↗
94. Failure surfaces in principal stress space10.1002/polc.5070320113 ↗
95. Physics-informed data-driven discovery of constitutive models with application to strain-rate-sensitive soft materials
96. Mechanik der festen körper im plastisch- deformablen zustand
97. Watson · 2024
98. Efficient hyperparameter optimization through model-based reinforcement learning10.1016/j.neucom.2020.06.064 ↗
99. Discovering non-associated pressure-sensitive plasticity models with EUCLID10.1186/s40323-024-00281-3 ↗
100. Xu · 2023