Constitutive-fibre mechanics: mechanics-directed uncertainty in constitutive function space

A structural prediction rests on constitutive functions that are known only through a finite number of tests. The tests fix the response along the states they visit; together with a declared regularity bound they leave a set of admissible constitutive functions, which we call the constitutive fibre. Constitutive-fibre mechanics (CFM) computes the range of a quantity of interest over this set. Adjoint solves of the mechanical problem represent the test observations and the quantity of interest as elements of the constitutive function space. The part of the sensitivity of the quantity of interest that is orthogonal to all observation sensitivities is hidden from the tests; for exact data it alone determines the half-width of the prediction, and for data with tolerance the half-width is a finite convex problem in the span of the adjoint representers. Comparing the two values shows whether a new type of test or a more precise test is needed, and a rank-one update ranks candidate tests before they are performed. A learned constitutive class can only under-represent the range; the deficit is computable. Beyond the linear range, an optimality system with state-dependent representers gives the extreme constitutive functions. The same set decides whether the prediction is defined for every admissible constitutive function, which is a different question from its precision. For a phase-field bar with unknown degradation function, 3% of the sensitivity of the stability eigenvalue is hidden and produces 70% of its uncertainty. For an open-hole composite lamina, the coupon that sharpens the compliance most is not the one that makes the prediction valid; validity is proved on a declared cell by a certificate accepted in exact arithmetic.

Authors

Publication Details

Journal
arXiv (Cornell University)
Published
2026-11-17
DOI
https://doi.org/10.48550/arxiv.2611.22187
Primary Topic
Probabilistic and Robust Engineering Design
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Constitutive-fibre mechanics: mechanics-directed uncertainty in constitutive function space

Timon Rabczuk, Xiaoying Zhuang, Altaf Ahmad Lone, R. T. Durai Prabhakaran
arXiv (Cornell University)
Probabilistic and Robust Engineering Design
preprint

Constitutive-fibre mechanics: mechanics-directed uncertainty in constitutive function space

Timon Rabczuk, Xiaoying Zhuang, Altaf Ahmad Lone, R. T. Durai Prabhakaran
preprint en

Abstract

A structural prediction rests on constitutive functions that are known only through a finite number of tests. The tests fix the response along the states they visit; together with a declared regularity bound they leave a set of admissible constitutive functions, which we call the constitutive fibre. Constitutive-fibre mechanics (CFM) computes the range of a quantity of interest over this set. Adjoint solves of the mechanical problem represent the test observations and the quantity of interest as elements of the constitutive function space. The part of the sensitivity of the quantity of interest that is orthogonal to all observation sensitivities is hidden from the tests; for exact data it alone determines the half-width of the prediction, and for data with tolerance the half-width is a finite convex problem in the span of the adjoint representers. Comparing the two values shows whether a new type of test or a more precise test is needed, and a rank-one update ranks candidate tests before they are performed. A learned constitutive class can only under-represent the range; the deficit is computable. Beyond the linear range, an optimality system with state-dependent representers gives the extreme constitutive functions. The same set decides whether the prediction is defined for every admissible constitutive function, which is a different question from its precision. For a phase-field bar with unknown degradation function, 3% of the sensitivity of the stability eigenvalue is hidden and produces 70% of its uncertainty. For an open-hole composite lamina, the coupon that sharpens the compliance most is not the one that makes the prediction valid; validity is proved on a declared cell by a certificate accepted in exact arithmetic.

arXiv (Cornell University)
Probabilistic and Robust Engineering Design
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon