A cooperative game for automated learning of elasto-plasticity knowledge graphs and models with AI-guided experimentation


We introduce a multi-agent meta-modeling game to generate data, knowledge, and models that make predictions on constitutive responses of elasto-plastic materials. We introduce a new concept from graph theory where a modeler agent is tasked with evaluating all the modeling options recast as a directed multigraph and find the optimal path that links the source of the directed graph (e.g. strain history) to the target (e.g. stress) measured by an objective function. Meanwhile, the data agent, which is tasked with generating data from real or virtual experiments (e.g. molecular dynamics, discrete element simulations), interacts with the modeling agent sequentially and uses reinforcement learning to design new experiments to optimize the prediction capacity. Consequently, this treatment enables us to emulate an idealized scientific collaboration as selections of the optimal choices in a decision tree search done automatically via deep reinforcement learning.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16


  1. 1.

    Andrade JE, Borja RI (2006) Capturing strain localization in dense sands with random density. Int J Numer Methods Eng 67(11):1531–1564

    MATH  Article  Google Scholar 

  2. 2.

    Asaro RJ (1983) Crystal plasticity. J Appl Mech 50(4b):921–934

    MATH  Article  Google Scholar 

  3. 3.

    Aydin A, Borja RI, Eichhubl P (2006) Geological and mathematical framework for failure modes in granular rock. J Struct Geol 28(1):83–98

    Article  Google Scholar 

  4. 4.

    Bang-Jensen J, Gutin GZ (2008) Digraphs: theory, algorithms and applications. Springer, Berlin

    Google Scholar 

  5. 5.

    Bardet JP, Choucair W (1991) A linearized integration technique for incremental constitutive equations. Int J Numer Anal Methods Geomech 15(1):1–19

    MATH  Article  Google Scholar 

  6. 6.

    Battaglia PW, Hamrick JB, Bapst V, Sanchez-Gonzalez A, Zambaldi V, Malinowski M, Tacchetti A, Raposo D, Santoro A, Faulkner R et al (2018) Relational inductive biases, deep learning, and graph networks. arXiv preprint arXiv:1806.01261

  7. 7.

    Been K, Jefferies MG, Hachey J (1991) Critical state of sands. Geotechnique 41(3):365–381

    Article  Google Scholar 

  8. 8.

    Bonabeau E (2002) Agent-based modeling: methods and techniques for simulating human systems. Proc Natl Acad Sci 99(suppl 3):7280–7287

    Article  Google Scholar 

  9. 9.

    Borja RI (2013) Plasticity: modeling and computation. Springer, Berlin

    Google Scholar 

  10. 10.

    Borja RI, Sama KM, Sanz PF (2003) On the numerical integration of three-invariant elastoplastic constitutive models. Comput Methods Appl Mech Eng 192(9–10):1227–1258

    MATH  Article  Google Scholar 

  11. 11.

    Boyce BL, Kramer SLB, Fang HE, Cordova TE, Neilsen MK, Dion K, Kaczmarowski AK, Karasz E, Xue L, Gross AJ (2014) The sandia fracture challenge: blind round robin predictions of ductile tearing. Int J Fract 186(1–2):5–68

    Article  Google Scholar 

  12. 12.

    Casagrande A (1976) Liquefaction and cyclic deformation of sands—a critical review. Harvard soil mechanics series (88). Harvard University, Cambridge

    Google Scholar 

  13. 13.

    Chomsky N (2014) Aspects of the theory of syntax, vol 11. MIT press

  14. 14.

    Choo J, Sun WC (2018a) Coupled phase-field and plasticity modeling of geological materials: from brittle fracture to ductile flow. Comput Methods Appl Mech Eng 330:1–32

    MathSciNet  Article  Google Scholar 

  15. 15.

    Choo J, Sun WC (2018b) Cracking and damage from crystallization in pores: coupled chemo-hydro-mechanics and phase-field modeling. Comput Methods Appl Mech Eng 335:347–379

    MathSciNet  Article  Google Scholar 

  16. 16.

    Coussy O (2004) Poromechanics. Wiley, New York

    Google Scholar 

  17. 17.

    Cundall PA, Strack ODL (1979) A discrete numerical model for granular assemblies. Geotechnique 29(1):47–65

    Article  Google Scholar 

  18. 18.

    Dabrowski R, Stencel K, Timoszuk G (2011) Software is a directed multigraph. In: European conference on software architecture. Springer, pp 360–369

  19. 19.

    Dafalias YF, Manzari MT (2004) Simple plasticity sand model accounting for fabric change effects. J Eng Mech 130(6):622–634

    Article  Google Scholar 

  20. 20.

    de Borst R, Heeres OM (2002) A unified approach to the implicit integration of standard, non-standard and viscous plasticity models. Int J Numer Anal Methods Geomech 26(11):1059–1070

    MATH  Article  Google Scholar 

  21. 21.

    Foerster J, Assael IA, de Freitas N, Whiteson S (2016) Learning to communicate with deep multi-agent reinforcement learning. In: 30th conference on neural information processing systems (NIPS 2016). Advances in neural information processing systems, Barcelona, Spain, pp 2137–2145

  22. 22.

    Furukawa T, Yagawa G (1998) Implicit constitutive modelling for viscoplasticity using neural networks. Int J Numer Methods Eng 43(2):195–219

    MATH  Article  Google Scholar 

  23. 23.

    Gens A, Potts DM (1988) Critical state models in computational geomechanics. Eng Comput 5(3):178–197

    Article  Google Scholar 

  24. 24.

    Ghaboussi J, Garrett JH Jr, Wu X (1991) Knowledge-based modeling of material behavior with neural networks. J Eng Mech 117(1):132–153

    Article  Google Scholar 

  25. 25.

    Ghaboussi J, Pecknold DA, Zhang M, Haj-Ali RM (1998) Autoprogressive training of neural network constitutive models. Int J Numer Methods Eng 42(1):105–126

    MATH  Article  Google Scholar 

  26. 26.

    Ghavamzadeh M, Mannor S, Pineau J, Tamar A (2015) Bayesian reinforcement learning: a survey. Found Trends® Mach Learn 8(5–6):359–483

    MATH  Article  Google Scholar 

  27. 27.

    Graham RL, Knuth DE, Patashnik O, Liu S (1989) Concrete mathematics: a foundation for computer science. Comput Phys 3(5):106–107

    MATH  Article  Google Scholar 

  28. 28.

    Hibbitt, Karlsson, Sorensen (2001) ABAQUS/standard user’s manual, vol 1. Hibbitt, Karlsson & Sorensen, Pawtucket

    Google Scholar 

  29. 29.

    Hornik K, Stinchcombe M, White H (1989) Multilayer feedforward networks are universal approximators. Neural Netw 2(5):359–366

    MATH  Article  Google Scholar 

  30. 30.

    Humboldt W (1999) On language: on the diversity of human language construction and its influence on the mental development of the human species. Cambridge University Press

  31. 31.

    Ibañez R, Abisset-Chavanne E, Aguado JV, Gonzalez D, Cueto E, Chinesta F (2018) A manifold learning approach to data-driven computational elasticity and inelasticity. Arch Comput Methods Eng 25(1):47–57

    MathSciNet  MATH  Article  Google Scholar 

  32. 32.

    Jefferies MG (1993) Nor-sand: a simle critical state model for sand. Géotechnique 43(1):91–103

    Article  Google Scholar 

  33. 33.

    Kendall MG et al (1946) The advanced theory of statistics. , 5th edn. Charles Griffin & Company, London

  34. 34.

    Kirchdoerfer T, Ortiz M (2016) Data-driven computational mechanics. Comput Methods Appl Mech Eng 304:81–101

    MathSciNet  MATH  Article  Google Scholar 

  35. 35.

    Kirchdoerfer T, Ortiz M (2017) Data driven computing with noisy material data sets. Comput Methods Appl Mech Eng 326:622–641

    MathSciNet  Article  Google Scholar 

  36. 36.

    Koeppe A, Bamer F, Padilla CAH, Markert B (2017) Neural network representation of a phase-field model for brittle fracture. PAMM 17(1):253–254

    Article  Google Scholar 

  37. 37.

    Kuhn MR, Sun WC, Wang Q (2015) Stress-induced anisotropy in granular materials: fabric, stiffness, and permeability. Acta Geotech 10(4):399–419

    Article  Google Scholar 

  38. 38.

    Lake Brenden M, Ullman Tomer D, Tenenbaum Joshua B, Gershman Samuel J (2017) Building machines that learn and think like people. Behav Brain Sci 40:e253

    Article  Google Scholar 

  39. 39.

    Lange M (2012) What makes a scientific explanation distinctively mathematical? Br J Philos Sci 64(3):485–511

    MathSciNet  MATH  Article  Google Scholar 

  40. 40.

    Lefik M, Schrefler BA (2002) Artificial neural network for parameter identifications for an elasto-plastic model of superconducting cable under cyclic loading. Comput Struct 80(22):1699–1713

    Article  Google Scholar 

  41. 41.

    Lefik M, Schrefler BA (2003) Artificial neural network as an incremental non-linear constitutive model for a finite element code. Comput Methods Appl Mech Eng 192(28–30):3265–3283

    MATH  Article  Google Scholar 

  42. 42.

    Li XS, Dafalias YF (2011) Anisotropic critical state theory: role of fabric. J Eng Mech 138(3):263–275

    Article  Google Scholar 

  43. 43.

    Ling HI, Liu H (2003) Pressure-level dependency and densification behavior of sand through generalized plasticity model. J Eng Mech 129(8):851–860

    Article  Google Scholar 

  44. 44.

    Ling HI, Yang S (2006) Unified sand model based on the critical state and generalized plasticity. J Eng Mech 132(12):1380–1391

    Article  Google Scholar 

  45. 45.

    Liu Y, Sun WC, Fish J (2016) Determining material parameters for critical state plasticity models based on multilevel extended digital database. J Appl Mech 83(1):011003

    Article  Google Scholar 

  46. 46.

    Liu Y, Sun WC, Yuan Z, Fish J (2016) A nonlocal multiscale discrete-continuum model for predicting mechanical behavior of granular materials. Int J Numer Methods Eng 106(2):129–160

    MathSciNet  MATH  Article  Google Scholar 

  47. 47.

    Liu Z, Fleming M, Liu WK (2018) Microstructural material database for self-consistent clustering analysis of elastoplastic strain softening materials. Comput Methods Appl Mech Eng 330:547–577

    MathSciNet  Article  Google Scholar 

  48. 48.

    Lubliner J, Auricchio F (1996) Generalized plasticity and shape-memory alloys. Int J Solids Struct 33(7):991–1003

    MATH  Article  Google Scholar 

  49. 49.

    Malcher L, Pires FMA, de Sá JMAC, Andrade FXC (2009) Numerical integration algorithm of a new model for metal plasticity and fracture including pressure and lode angle dependence. Int J Mater Form 2(1):443–446

    Article  Google Scholar 

  50. 50.

    Malmgren RD, Ottino JM, Amaral LAN (2010) The role of mentorship in protégé performance. Nature 465(7298):622

    Article  Google Scholar 

  51. 51.

    Manzari MT, Dafalias YF (1997) A critical state two-surface plasticity model for sands. Geotechnique 47(2):255–272

    Article  Google Scholar 

  52. 52.

    Miehe C, Schröder J (2001) A comparative study of stress update algorithms for rate-independent and rate-dependent crystal plasticity. Int J Numer Methods Eng 50(2):273–298

    MATH  Article  Google Scholar 

  53. 53.

    Mira P, Tonni L, Pastor M, Merodo JAF (2009) A generalized midpoint algorithm for the integration of a generalized plasticity model for sands. Int J Numer Methods Eng 77(9):1201–1223

    MathSciNet  MATH  Article  Google Scholar 

  54. 54.

    Mooney MA, Finno RJ, Viggiani MG (1998) A unique critical state for sand? J Geotech Geoenviron Eng 124(11):1100–1108

    Article  Google Scholar 

  55. 55.

    Munafò MR, Nosek BA, Bishop DVM, Button KS, Chambers CD, du Sert NP, Simonsohn U, Wagenmakers E-J, Ware JJ, Ioannidis JPA (2017) A manifesto for reproducible science. Nat Hum Behav 1:0021

    Article  Google Scholar 

  56. 56.

    Na SH, Sun WC (2017) Computational thermo-hydro-mechanics for multiphase freezing and thawing porous media in the finite deformation range. Comput Methods Appl Mech Eng 318:667–700

    MathSciNet  Article  Google Scholar 

  57. 57.

    Na SH, Sun WC (2018) Computational thermomechanics of crystalline rock, part I: a combined multi-phase-field/crystal plasticity approach for single crystal simulations. Comput Methods Appl Mech Eng 338:657–691

    MathSciNet  Article  Google Scholar 

  58. 58.

    Olivier A, Smyth AW (2018) A marginalized unscented kalman filter for efficient parameter estimation with applications to finite element models. Comput Methods Appl Mech Eng 339:615–643

    MathSciNet  Article  Google Scholar 

  59. 59.

    Ortiz M, Pandolfi A (1999) Finite-deformation irreversible cohesive elements for three-dimensional crack-propagation analysis. Int J Numer Methods Eng 44(9):1267–1282

    MATH  Article  Google Scholar 

  60. 60.

    Pack K, Luo M, Wierzbicki T (2014) Sandia fracture challenge: blind prediction and full calibration to enhance fracture predictability. Int J Fract 186(1–2):155–175

    Article  Google Scholar 

  61. 61.

    Pandolfi ANNA, Krysl P, Ortiz M (1999) Finite element simulation of ring expansion and fragmentation: the capturing of length and time scales through cohesive models of fracture. Int J Fract 95(1–4):279–297

    Article  Google Scholar 

  62. 62.

    Park K, Paulino GH (2011) Cohesive zone models: a critical review of traction-separation relationships across fracture surfaces. Appl Mech Rev 64(6):060802

    Article  Google Scholar 

  63. 63.

    Pastor M, Zienkiewicz OC, Chan AHC (1990) Generalized plasticity and the modelling of soil behaviour. Int J Numer Anal Methods Geomech 14(3):151–190

    MATH  Article  Google Scholar 

  64. 64.

    Pestana JM, Whittle AJ, Salvati LA (2002) Evaluation of a constitutive model for clays and sands: part I—sand behaviour. Int J Numer Anal Methods Geomech 26(11):1097–1121

    MATH  Article  Google Scholar 

  65. 65.

    Raileanu R, Denton E, Szlam A, Fergus R (2018) Modeling others using oneself in multi-agent reinforcement learning. arXiv preprint arXiv:1802.09640

  66. 66.

    Rutqvist J, Ijiri Y, Yamamoto H (2011) Implementation of the barcelona basic model into tough-flac for simulations of the geomechanical behavior of unsaturated soils. Comput Geosci 37(6):751–762

    Article  Google Scholar 

  67. 67.

    Salinger AG, Bartlett RA, Bradley AM, Chen Q, Demeshko IP, Gao X, Hansen GA, Mota A, Muller RP, Nielsen E et al (2016) Albany: using component-based design to develop a flexible, generic multiphysics analysis code. Int J Multiscale Comput Eng 14(4):415–438

    Article  Google Scholar 

  68. 68.

    Sánchez M, Gens A, Guimarães LN, Olivella S (2005) A double structure generalized plasticity model for expansive materials. Int J Numer Anal Methods Geomech 29(8):751–787

    MATH  Article  Google Scholar 

  69. 69.

    Schofield A, Wroth P (1968) Critical state soil mechanics, vol 310. McGraw-Hill, London

    Google Scholar 

  70. 70.

    Silver D, Hubert T, Schrittwieser J, Antonoglou I, Lai M, Guez A, Lanctot M, Sifre L, Kumaran D, Graepel T et al (2017) Mastering chess and shogi by self-play with a general reinforcement learning algorithm. arXiv preprint arXiv:1712.01815

  71. 71.

    Silver D, Hubert T, Schrittwieser J, Antonoglou I, Lai M, Guez A, Lanctot M, Sifre L, Kumaran D, Graepel T et al (2017) Mastering chess and shogi by self-play with a general reinforcement learning algorithm. arXiv preprint arXiv:1712.01815

  72. 72.

    Silver D, Schrittwieser J, Simonyan K, Antonoglou I, Huang A, Guez A, Hubert T, Baker L, Lai M, Bolton A et al (2017c) Mastering the game of go without human knowledge. Nature 550(7676):354

    Article  Google Scholar 

  73. 73.

    Simo JC, Hughes TJR (2006) Computational inelasticity, vol 7. Springer, Berlin

    Google Scholar 

  74. 74.

    Sloan SW (1987) Substepping schemes for the numerical integration of elastoplastic stress–strain relations. Int J Numer Methods Eng 24(5):893–911

    MATH  Article  Google Scholar 

  75. 75.

    Sloan SW, Abbo AJ, Sheng D (2001) Refined explicit integration of elastoplastic models with automatic error control. Eng Comput 18(1/2):121–194

    MATH  Article  Google Scholar 

  76. 76.

    Šmilauer V, Catalano E, Chareyre B, Dorofeenko S, Duriez J, Gladky A, Kozicki J, Modenese C, Scholtès L, Sibille L et al (2010) Yade reference documentation. Yade Doc 474(1):1–161

    Google Scholar 

  77. 77.

    Smith J, Xiong W, Yan W, Lin S, Cheng P, Kafka OL, Wagner GJ, Cao J, Liu WK (2016) Linking process, structure, property, and performance for metal-based additive manufacturing: computational approaches with experimental support. Comput Mech 57(4):583–610

    MATH  Article  Google Scholar 

  78. 78.

    Sun Q, Tao Y, Du Q (2018) Stochastic training of residual networks: a differential equation viewpoint. arXiv preprintarXiv:1812.00174

  79. 79.

    Sun WC, Kuhn MR, Rudnicki JW et al (2014) A micromechanical analysis on permeability evolutions of a dilatant shear band. In: 48th US rock mechanics/geomechanics symposium. American Rock Mechanics Association

  80. 80.

    Sun WC (2013) A unified method to predict diffuse and localized instabilities in sands. Geomech Geoeng 8(2):65–75

    Article  Google Scholar 

  81. 81.

    Sun WC (2015) A stabilized finite element formulation for monolithic thermo-hydro-mechanical simulations at finite strain. Int J Numer Methods Eng 103(11):798–839

    MathSciNet  MATH  Article  Google Scholar 

  82. 82.

    Sun WC, Kuhn MR, Rudnicki JW (2013) A multiscale DEM-LBM analysis on permeability evolutions inside a dilatant shear band. Acta Geotech 8(5):465–480

    Article  Google Scholar 

  83. 83.

    Sun WC, Ostien JT, Salinger AG (2013) A stabilized assumed deformation gradient finite element formulation for strongly coupled poromechanical simulations at finite strain. Int J Numer Anal Methods Geomech 37(16):2755–2788

    Google Scholar 

  84. 84.

    Tampuu A, Matiisen T, Kodelja D, Kuzovkin I, Korjus K, Aru J, Aru J, Vicente R (2017) Multiagent cooperation and competition with deep reinforcement learning. PLoS ONE 12(4):e0172395

    Article  Google Scholar 

  85. 85.

    Tan M (1993) Multi-agent reinforcement learning: independent vs. cooperative agents. In: Proceedings of the tenth international conference on machine learning, pp 330–337

  86. 86.

    Tang S, Zhang L, Liu WK (2018) From virtual clustering analysis to self-consistent clustering analysis: a mathematical study. Comput Mech 62(6):1443–1460

    MathSciNet  MATH  Article  Google Scholar 

  87. 87.

    Truesdell C (1959) The rational mechanics of materials—past, present, future. Appl Mech Rev 12:75–80

    MathSciNet  Google Scholar 

  88. 88.

    Truesdell C, Noll W (2004) The non-linear field theories of mechanics. Springer, Berlin, Heidelberg, pp 1–579

  89. 89.

    Tu X, Andrade JE, Chen Q (2009) Return mapping for nonsmooth and multiscale elastoplasticity. Comput Methods Appl Mech Eng 198(30–32):2286–2296

    MATH  Article  Google Scholar 

  90. 90.

    Tvergaard V (1990) Effect of fibre debonding in a whisker-reinforced metal. Mater Sci Eng A 125(2):203–213

    Article  Google Scholar 

  91. 91.

    Ulven OI, Sun WC (2018) Capturing the two-way hydromechanical coupling effect on fluid-driven fracture in a dual-graph lattice beam model. Int J Numer Anal Methods Geomech 42(5):736–767

    Article  Google Scholar 

  92. 92.

    Wang K, Sun WC (2016) A semi-implicit discrete-continuum coupling method for porous media based on the effective stress principle at finite strain. Comput Methods Appl Mech Eng 304:546–583

    MathSciNet  MATH  Article  Google Scholar 

  93. 93.

    Wang K, Sun WC (2017) Data-driven discrete-continuum method for partially saturated micro-polar porous media. In: Poromechanics VI, pp 571–578

  94. 94.

    Wang K, Sun WC (2018) A multiscale multi-permeability poroplasticity model linked by recursive homogenizations and deep learning. Comput Methods Appl Mech Eng 334:337–380

    MathSciNet  Article  Google Scholar 

  95. 95.

    Wang K, Sun WC (2019) Meta-modeling game for deriving theory-consistent, microstructure-based traction-separation laws via deep reinforcement learning. Comput Methods Appl Mech Eng 346:216–241

    MathSciNet  Article  Google Scholar 

  96. 96.

    Wang K, Sun WC (2019) An updated Lagrangian LBM–DEM–FEM coupling model for dual-permeability fissured porous media with embedded discontinuities. Comput Methods Appl Mech Eng 344:276–305

    MathSciNet  Article  Google Scholar 

  97. 97.

    Wang K, Sun W, Salager S, Na S, Khaddour G (2016) Identifying material parameters for a micro-polar plasticity model via X-ray micro-CT images: lessons learned from the curve-fitting exercises. Int J Multiscale Comput Eng 14(4):389–413

    Article  Google Scholar 

  98. 98.

    West DB et al (2001) Introduction to graph theory, vol 2. Prentice Hall, Upper Saddle River

    Google Scholar 

  99. 99.

    Wollny I, Sun WC, Kaliske M (2017) A hierarchical sequential ale poromechanics model for tire–soil–water interaction on fluid-infiltrated roads. Int J Numer Methods Eng 112(8):909–938

    MathSciNet  Article  Google Scholar 

  100. 100.

    Wood DM (1990) Soil behaviour and critical state soil mechanics. Cambridge University Press, Cambridge

    Google Scholar 

  101. 101.

    Xin H, Sun WC, Fish J (2017) Discrete element simulations of powder-bed sintering-based additive manufacturing. Int J Mech Sci 149:373–392

    Article  Google Scholar 

  102. 102.

    Zhao J, Guo N (2013) Unique critical state characteristics in granular media considering fabric anisotropy. Géotechnique 63(8):695

    Article  Google Scholar 

  103. 103.

    Zienkiewicz OC, Mroz Z (1984) Generalized plasticity formulation and applications to geomechanics. Mech Eng Mater 44(3):655–680

    Google Scholar 

  104. 104.

    Zienkiewicz Olgierd C, Chan AHC, Pastor M, Schrefler BA, Shiomi T (1999) Computational geomechanics. Citeseer, New York

    Google Scholar 

  105. 105.

    Zohdi TI (2013) Rapid simulation of laser processing of discrete particulate materials. Arch Comput Methods Eng 20(4):309–325

    Article  Google Scholar 

Download references


The work of KW and WCS is supported by the Earth Materials and Processes program from the US Army Research Office under Grant Contract W911NF-18-2-0306, the Dynamic Materials and Interactions Program from the Air Force Office of Scientific Research under Grant Contract FA9550-17-1-0169, the nuclear energy university program from Department of Energy under Grant Contract DE-NE0008534, the Mechanics of Material program at National Science Foundation under Grant Contract CMMI-1462760, and the Columbia SEAS Interdisciplinary Research Seed Grant. The work of QD is supported in part by NSF CCF-1704833, DMS-1719699, DMR-1534910, and ARO MURI W911NF-15-1-0562. These supports are gratefully acknowledged. The views and conclusions contained in this document are those of the authors, and should not be interpreted as representing the official policies, either expressed or implied, of the sponsors, including the Army Research Laboratory or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein.

Author information



Corresponding author

Correspondence to WaiChing Sun.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix: Traction-separation example

Appendix: Traction-separation example

In this appendix, we provide an additional simple example for traction-separation law to (1) demonstrate how to selection components from a combinations of existing hand-crafted and machine-generated operators/mappings, (2) provide one possible example to form the the directed labeled multi-graph from existing mappings in the plasticity literature and (3) show the inclusive nature of the multi-graph approach by demonstrating how to merge multiple sub-graph into a multi-graph.

Example 1

Traction-separation Law. Given a pre-defined objective function, assume that the only known theoretical traction-separation model incorporated in the labeled directed multigraph are the Tvergaard model (cf. [90]) and the Ortiz–Pandolfi model (cf. [61]). In addition, we also consider using a neural network that incorporates porosity to predict traction-separation relations. Define the labeled directed multi-graph that provides all the options available.

Fig. 17

The generation of directed multi-graph by expanding action space using previous models

First, we convert the traction-separation laws into directed graphs where the relative displacement vector is the input and the traction is the output. Notice that both Tvergaard [90] and Pandolfi et al. [61] are effective displacement models where an effective displacement \(\overline{\Delta }\) is used as additional input to predict the traction. In [90],

$$\begin{aligned} T_{n}= & {} \frac{\overline{T}(\overline{\Delta })}{\overline{\Delta }}\frac{\Delta _{n}}{\delta _{n}}, \end{aligned}$$
$$\begin{aligned} T_{t}= & {} \frac{\overline{T}(\overline{\Delta })}{\overline{\Delta }} \alpha \frac{\Delta _{n}}{\delta _{t}} \end{aligned}$$

and the effective displacement and effective traction are scalars defined as,

$$\begin{aligned} \overline{\Delta }= & {} \sqrt{ (\Delta _{n}/ \delta _{n})^{2} + (\Delta _{t} / \delta _{t})^{2}}, \end{aligned}$$
$$\begin{aligned} \overline{T}(\overline{\Delta })= & {} \frac{27}{4} \sigma _{\max } \overline{\Delta }(1 - 2 \overline{\Delta } + \overline{\Delta }^{2}), \end{aligned}$$

where \(\delta _{n}\) and \(\delta _{t}\) are the characteristic length corresponding to the fracture energy and cohesive strength of the normal and tangential opening modes, \(\alpha \) is a non-dimensional material parameter. As pointed out in [62], the traction-separation model in [61] can be expressed in the forms of Eqs. (57) and (58) with the alternative definition of effective displacement and traction separation law, i.e.,

$$\begin{aligned} \overline{\Delta }= & {} \tilde{\Delta } / \delta _{n} \, , \, \tilde{\Delta } = \sqrt{ \Delta _{n}^{2} + \beta ^{2} \Delta _{t}^{2}} \end{aligned}$$
$$\begin{aligned} \overline{T}(\overline{\Delta })= & {} k \overline{\Delta } + c \end{aligned}$$

where k is typically negative and c is the effective cohesive strength. Finally, we consider a neural network model in which the traction depends on the porosity \(\phi ^{f}\) [16, 82, 92], i.e.,

$$\begin{aligned} T_{n}= & {} f^{\text {LSTM}}(\phi ^{f}, \Delta _{n}), \end{aligned}$$
$$\begin{aligned} T_{t}= & {} g^{\text {LSTM}}(\phi ^{f}, \Delta _{t}), \end{aligned}$$

where the exact expression of the function \(f^{\text {LSTM}}\) and \(g^{\text {LSTM}}\) are determined by adjusting the weight of the neurons in the recurrent neural network [36, 94]. Assuming that the solid constituent is incompressible, the porosity reads,

$$\begin{aligned} \phi ^{f} = \phi ^{f}_{o} (1 + \Delta _{n} \Delta _{t}) \end{aligned}$$

The multi-graph that combines all the possible choices of the three traction separation laws can therefore be defined by multi-graph statement with the following sets,

$$\begin{aligned} \mathbb {V}= & {} \{\Delta _{n}, \Delta _{t}, T_{n}, T_{t}, \overline{\Delta }, \overline{T}, \phi ^{f}\} \end{aligned}$$
$$\begin{aligned} \mathbb {E}= & {} \mathbb {E}_{1} \cup \mathbb {E}_{2} \cup \mathbb {E}_{3} \end{aligned}$$
$$\begin{aligned} \mathbb {E}_{1}= & {} \{ \Delta _{n} \rightarrow \overline{\Delta }, \Delta _{t} \rightarrow \overline{\Delta }, \Delta _{n} \rightarrow \phi ^{f}, \Delta _{t} \rightarrow \phi ^{f}, \Delta _{n} \nonumber \\&\quad \rightarrow T_{n}, \Delta _{t} \rightarrow T_{t} \} \end{aligned}$$
$$\begin{aligned} \mathbb {E}_{2}= & {} \{ \overline{\Delta } \rightarrow \overline{T}, \phi ^{f} \rightarrow T_{n}, \phi ^{f} \rightarrow T_{t}, \Delta _{n} \rightarrow T_{n} \} \end{aligned}$$
$$\begin{aligned} \mathbb {E}_{3}= & {} \{ \overline{T} \rightarrow T_{n}, \overline{T} \rightarrow T_{t} \} \end{aligned}$$
$$\begin{aligned} \mathbb {L_{V}}= & {} \{\text {normal disp., tan. disp., normal traction, tan.}, \nonumber \\&\quad \text {traction eff. disp., eff. traction, porosity} \} \end{aligned}$$
$$\begin{aligned} \mathbb {L_{E}}= & {} \{ \text {Eq.} (57), \text {Eq.} (58), \text {Eq.} (59), \text {Eq.} (60), \text {Eq.} (61), \text {Eq.} (62),\nonumber \\&\quad \text {Eq.} (63), \text {Eq.} (64), \text {Eq.} (65) \} \end{aligned}$$

Since \(\varvec{n}_{\mathbb {V}}\) is a bijective mapping, the labeling of the vertices is trivial. The rest of the mappings, i.e. \(\varvec{s}\), \(\varvec{t}\) and \(\varvec{n}_{\mathbb {E}}\) can be visualized in a labeled directed multigraph as shown in Fig. 17. Essentially, the process of creating the directed multigraph is to mathematically represent all the possible options modelers can have when they are tasked to create a constitutive model for a data set. \(\square \)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, K., Sun, W. & Du, Q. A cooperative game for automated learning of elasto-plasticity knowledge graphs and models with AI-guided experimentation. Comput Mech 64, 467–499 (2019). https://doi.org/10.1007/s00466-019-01723-1

Download citation


  • Directed multigraph
  • Data-driven constitutive modeling
  • Multi-agent deep reinforcement learning
  • Combinatorial optimization
  • Computational combinatorics