Inverse Design of Tightly Woven Smart Fabrics

arXiv:2601.15746 · cond-mat.soft · Submitted 2026-01-22 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "Inverse Design of Tightly Woven Smart Fabrics".

Mira: Here's the gist of the paper. Inverse design of tightly woven smart fabrics presents a geometric framework for designing smart woven fabrics that conform to an arbitrary target geometry when actuated This approach solves a nonlinear hyperbolic partial differential equation to achieve programmable, complex three-dimensional shaping Fabric Modeling and Constraints The fabric is modeled as a two-dimensional (2D) surface representing the mid-plane of the structure, coarse-grained at the scale of the weaving unit cell The two-dimensional geometry of the fabric at a given state is captured in the metric tensor ds2 = E (u, v) du2 + 2F (u, v) dudv + G (u, v) dv2 In a tightly-woven fabric, threads lose their local freedom to change position or angle and the fabric becomes rigid and stiff Experiments show that increasing thread density rigidifies both bending and shear

24, 25: and diminishes the maximal shear angle Since the symmetry of the weaving unit-cell favors neither shear direction, and because we assume that lateral gradients at the scale of the unit cell are small, F = 0 is imposed The same jamming transition also imposes a nontrivial relationship between E and G in the tight limit due to the weaving topology Mechanical Response Simulation A three-dimensional finite element model is used to simulate the mechanical response of a plain woven fabric, focusing on a single unit cell with periodic boundary conditions The elastic energy consists of a stretching term that penalizes deviation from the springs’ rest lengths, and a bending term that penalizes angles between neighboring springs A hard-cylinder repulsive term is introduced to strongly penalize inter-thread distance smaller than 1 The results corresponding to simulations at different rest lengths of the threads are plotted in Fig.

Kai: 2 The tight weaving regime comprises a large domain of thread rest-lengths that, when woven, reach the jamming transition The metric components of such fabrics indeed collapse to a onedimensional curve, namely a strict equality relating E and G Inverse Design Formulation The inverse design problem is formulated as a second-order PDE for α (u, v) G′αuu + E′αvv = (EG)′2EGα2vE' + α2uG' - α2vE'' - α2uG'' - 2EGK(u, v) In the symmetric case of a surface of revolution, the inverse design problem reduces to a second-order ODE The solution is given by Eq.

Mira: 6: α′′(u) + A

α: α' (u)2 = B

α: K s(u) For the special case of constant Gaussian curvature K(s) ≡ K0, the system (A.6) admits a first integral w(α)2 + 4K0 G(α) = C Solution Methods and Practical Implementation The solution to a Cauchy problem guarantees a unique solution for α(u, v) in the surroundings of the initial curve For practical implementations, the target surface will be given in a discrete form, such as a triangular mesh An algorithm for numerically solving the initial value problem described above on such a mesh is presented in Sec.

Kai: A.3 An alternative approach is to search for a global coordinate parameterization of the target surface, in which the metric of the surface at any point takes the form Eq.

Paper summary: Kai: So, to recap what we just covered with "Inverse Design of Tightly-Woven Smart Fabrics," the authors are presenting this geometric framework for inverse design of smart woven fabrics composed of nonuniformly shrinking threads >

Mira: They claim that a sufficiently tight weaving structure imposes strong local criteria on material deformation and reduces the local geometry to a single scalar degree of freedom >

Lev: And they say that control over this one degree of freedom can be achieved through simple calibration for each specific material system, either through mechanical experiments or numerical simulations >

Kai: Basically, they use this reduction in degrees of freedom to inverse-design a woven smart fabric that conforms to an arbitrary target geometry when you actuate it by solving a nonlinear hyperbolic partial differential equation >

Mira: They validate this by deriving the thread-level actuation required for specific target geometries and they present both exact analytic solutions for symmetric shapes and a numerical optimization method for arbitrary freeform surfaces >

Lev: The simulation part is interesting because they model the mechanical response of a plain woven fabric using a three dimensional finite element model focusing on a single unit cell with periodic boundary conditions >

Kai: They describe the elastic energy as having stretching terms that penalize deviation from the springs rest lengths and bending terms that penalize angles between neighboring springs >

Mira: There's also this hard cylinder repulsive term introduced to strongly penalize interthread distance smaller than one and they show in simulations that when threads are woven tightly, the metric components of such fabrics indeed collapse to a one dimensional curve, meaning a strict equality relating the stretching and bending terms >

Lev: That collapse means in the tight limit, you only have one degree of freedom left to control the shape >

Conclusion: Kai: So looking at "Inverse Design of Tightly-Woven Smart Fabrics" and what we just discussed, the title really hits on how they use this geometric framework for inverse design >

Mira: It’s about using a specific constraint—tight weaving—to simplify the problem so you can control complex three dee shaping by just calibrating the thread response >

Lev: The implication is that instead of designing every single thread actuation profile from scratch, you use this framework to get an explicit recipe for creating fabrics that deform into any shape you want locally >

Kai: It seems like they've provided a constructive method ranging from exact analytic solutions for surfaces of revolution to a global optimization method for general discrete surfaces >

Mira: The authors suggest that using the preferred non zero saddle-like extrinsic curvature of knitted fabrics gives manufacturers some control over the resulting shape choice among all possible shapes >

Lev: What they're pointing toward is that this framework provides an explicit recipe for fabrics that will deform into an arbitrary target geometry at least locally, which is what we need when building real actuators >

Einav Berin, Hillel Aharoni

Weizmann Institute of Science

cond-mat.soft

Submitted: 2026-01-22

Updated: 2026-11-09

Comments: 13 pages, 4 figures, 8 SI pages, 1 SI figure

Code: https://github.com/etcorman/RectangularSurfaceParameterization

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 79/100

The gist: Inverse design of tightly woven smart fabrics presents a geometric framework for designing smart woven fabrics that conform to an arbitrary target geometry when actuated This approach solves a…

Key concepts

Metric Tensor
The metric tensor ds² = E (u, v) du² + 2F (u, v) dudv + G (u, v) dv² describes the two-dimensional geometry of the fabric surface at any given state; it mathematically defines how distances and angles are measured on the fabric's mid-plane [10].
Tight Weaving Constraints
In tightly woven fabrics, threads lose local freedom to move or change angle, making the fabric rigid and stiff; this constraint simplifies the geometry by imposing F = 0 because lateral gradients at the unit cell scale are assumed small [24, 25].
Inverse Design PDE
The inverse design problem is formulated as a second-order PDE for $\alpha$ (u, v) that governs how the fabric's shape changes based on actuation; solving this equation yields the required deformation profile for programmable shaping [A.3].
Gaussian Curvature
Gaussian curvature $K(u, v)$ is a geometric property of a surface that measures how much the surface deviates from being flat at a point; it is crucial in relating the fabric's geometry to its physical properties, as seen in the Brioschi formula [A.2].

Terminology

Summary

Inverse design of tightly woven smart fabrics presents a geometric framework for designing smart woven fabrics that conform to an arbitrary target geometry when actuated This approach solves a nonlinear hyperbolic partial differential equation to achieve programmable, complex three-dimensional shaping

Fabric Modeling and Constraints

The fabric is modeled as a two-dimensional (2D) surface representing the mid-plane of the structure, coarse-grained at the scale of the weaving unit cell The two-dimensional geometry of the fabric at a given state is captured in the metric tensor ds2 = E (u, v) du2 + 2F (u, v) dudv + G (u, v) dv2 In a tightly-woven fabric, threads lose their local freedom to change position or angle and the fabric becomes rigid and stiff Experiments show that increasing thread density rigidifies both bending and shear [24, 25] and diminishes the maximal shear angle Since the symmetry of the weaving unit-cell favors neither shear direction, and because we assume that lateral gradients at the scale of the unit cell are small, F = 0 is imposed The same jamming transition also imposes a nontrivial relationship between E and G in the tight limit due to the weaving topology

Mechanical Response Simulation

A three-dimensional finite element model is used to simulate the mechanical response of a plain woven fabric, focusing on a single unit cell with periodic boundary conditions The elastic energy consists of a stretching term that penalizes deviation from the springs’ rest lengths, and a bending term that penalizes angles between neighboring springs A hard-cylinder repulsive term is introduced to strongly penalize inter-thread distance smaller than 1 The results corresponding to simulations at different rest lengths of the threads are plotted in Fig. 2 The tight weaving regime comprises a large domain of thread rest-lengths that, when woven, reach the jamming transition The metric components of such fabrics indeed collapse to a onedimensional curve, namely a strict equality relating E and G

Inverse Design Formulation

The inverse design problem is formulated as a second-order PDE for α (u, v) G′αuu + E′αvv = (EG)′2EGα2vE' + α2uG' - α2vE'' - α2uG'' - 2EGK(u, v) In the symmetric case of a surface of revolution, the inverse design problem reduces to a second-order ODE The solution is given by Eq. 6: α′′(u) + A [α] α' (u)2 = B [α] K s(u) For the special case of constant Gaussian curvature K(s) ≡ K0, the system (A.6) admits a first integral w(α)2 + 4K0 G(α) = C

Solution Methods and Practical Implementation

The solution to a Cauchy problem guarantees a unique solution for α(u, v) in the surroundings of the initial curve For practical implementations, the target surface will be given in a discrete form, such as a triangular mesh An algorithm for numerically solving the initial value problem described above on such a mesh is presented in Sec. A.3 An alternative approach is to search for a global coordinate parameterization of the target surface, in which the metric of the surface at any point takes the form Eq. p. 4

Where the paper says it

Such a solution is not guaranteed to extend to the full mesh. An alternative approach is to search for a global coordinate parameterization of the 5

2 for given functions E and G The Corman–Crane method produces a rectangular parameterization consistent with the tight-weave constraints, to which threads can be aligned This demonstrates that the discrete construction provides a faithful, though approximate, realization of the target geometry

Discussion and Future Directions

The framework allows predicting the resulting surface shape from the thread actuation profile via a calibration curve of the type of Fig. 2 The authors provide constructive solutions for this system, ranging from an exact analytical solution for surfaces of revolution to a global optimization method for general discrete surfaces While global solutions do not always exist, they are expected to exist if they are allowed to contain point disclinations The authors suggest that utilizing the preferred non-zero saddle-like extrinsic curvature of knitted fabrics grants the manufacturer access to some degree of control over the resulting shape choice among all isometries of the target surface The framework provides an explicit recipe for fabrics that will deform into an arbitrary target geometry upon actuation, at least locally

Acknowledgments

We thank Etienne Corman and Keenan Crane for clarifications and advice on adapting their code We thank Myfanwy E. Evans and Randall D. p. 6

Where the paper says it

We thank Myfanwy E. Evans and Randall D.

Kamien for fruitful discussions This research was supported by the Israel Science Foundation (grant No. p. 6
Where the paper says it

Kamien for fruitful discussions. This research was supported by the Israel Science Foundation (grant No.

2677/20)

References

[1] Tianyong Zheng, Jing Wei, Zhengtao Shi, Tingting Li, and Zhen Wu. An overview of modeling yarn’s 3d geometric configuration

[2] Xavier Provot. Deformation constraints in a mass-spring model to describe rigid cloth behavior

[3] David Baraff and Andrew Witkin. Large steps in cloth simulation

[4] J. R. Womersley. 6.—the application of differential geometry to the study of the deformation of cloth under stress

[5] Samuel Poincloux, Mokhtar Adda-Bedia, and Fr´ed´eric Lechenault. Geometry and elasticity of a knitted fabric

[6] Philippe Boisse, Nabil Hamila, Elizabeth Guzman-Maldonado, Angela Madeo, Guillaume Hivet, and Francesco dell’Isola. The bias-extension test for the analysis of in-plane shear properties of textile composite reinforcements and prepregs: a review

[7] Hyun Kim, Suk-kyun Ahn, D. Mackie, Jinhyeong Kwon, Shi-Hyeong Kim, Changsoon Choi, Y. p. 7

Where the paper says it

[7] Hyun Kim, Suk-kyun Ahn, D. Mackie, Jinhyeong Kwon, Shi-Hyeong Kim, Changsoon Choi, Y.

Moon, Ha Beom Lee, and S. p. 7
Where the paper says it

Mackie, Jinhyeong Kwon, Shi-Hyeong Kim, Changsoon Choi, Y. Moon, Ha Beom Lee, and S.

Ko. Shape morphing smart 3d actuator materials for micro soft robot

[8] Yoshitsugu Hirokawa and Toyoichi Tanaka. Volume phase transition in a nonionic gel

[9] J¨urgen K¨upfer and Heino Finkelmann. Nematic liquid single crystal elastomers

[10] Andreas Lendlein, Annette M Schmidt, and Robert Langer. Ab-polymer networks based on oligo (εcaprolactone) segments showing shape-memory properties

[11] Arda Kotikian, Ryan L Truby, John William Boley, Timothy J White, and Jennifer A Lewis. 3d printing of liquid crystal elastomeric actuators with spatially programed nematic order

[12] Jiaqi Liu, Yuchong Gao, Haihuan Wang, Ryan Poling-Skutvik, Chinedum O. Osuji, and Shu Yang. p. 7

Where the paper says it

[12] Jiaqi Liu, Yuchong Gao, Haihuan Wang, Ryan Poling-Skutvik, Chinedum O. Osuji, and Shu Yang.

Shaping and locomotion of soft robots using filament actuators made from liquid crystal elastomer–carbon nanotube composites

[13] Pedro E. S. Silva, Xueyan Lin, Maija Vaara, Mithila Mohan, Jaana Vapaavuori, and Eugene M. p. 7

Where the paper says it

[13] Pedro E. Silva, Xueyan Lin, Maija Vaara, Mithila Mohan, Jaana Vapaavuori, and Eugene M.

Terentjev. Active textile fabrics from weaving liquid crystalline elastomer filaments

[14] Jiahao Sun, Wei Liao, and Zhongqiang Yang. Additive manufacturing of liquid crystal elastomer actuators based on knitting technology

[15] Antonio Proctor Martinez, Alicia Ng, So Hee Nah, and Shu Yang. Active-textile yarns and embroidery enabled by wet-spun liquid crystalline elastomer filaments

[16] Robert F. Shepherd, Filip Ilievski, Wonjae Choi, Stephen A. p. 7

Where the paper says it

[16] Robert F. Shepherd, Filip Ilievski, Wonjae Choi, Stephen A.

Morin, Adam A. p. 7
Where the paper says it

Shepherd, Filip Ilievski, Wonjae Choi, Stephen A. Morin, Adam A.

Stokes, Aaron D. p. 7
Where the paper says it

Morin, Adam A. Stokes, Aaron D.

Mazzeo, Xin Chen, Michael Wang, and George M. p. 7
Where the paper says it

Stokes, Aaron D. Mazzeo, Xin Chen, Michael Wang, and George M.

Whitesides. p. 7
Where the paper says it

Mazzeo, Xin Chen, Michael Wang, and George M. Whitesides.

Multigait soft robot

[17] Gregory P Crawford, Scott J Woltman, and Gregory D Jay. Liquid-crystal materials find a new order in biomedical applications

[18] A. Sydney Gladman, Elisabetta A. p. 7

Where the paper says it

Nature materials, 6(12):929–938, 2007. Sydney Gladman, Elisabetta A.

Matsumoto, Ralph G. p. 7
Where the paper says it

Sydney Gladman, Elisabetta A. Matsumoto, Ralph G.

Nuzzo, L. Mahadevan, and Jennifer A. p. 7
Where the paper says it

Matsumoto, Ralph G. Mahadevan, and Jennifer A.

Lewis. Biomimetic 4d printing

[19] Mina Konakovi´c, Keenan Crane, Bailin Deng, Sofien Bouaziz, Daniel Piker, and Mark Pauly. Beyond developable: computational design and fabrication with auxetic materials

[20] Hillel Aharoni, Yu Xia, Xinyue Zhang, Randall D. Kamien, and Shu Yang. p. 7

Where the paper says it

[20] Hillel Aharoni, Yu Xia, Xinyue Zhang, Randall D. Kamien, and Shu Yang.

Universal inverse design of surfaces with thin nematic elastomer sheets

[21] Jungwook Kim, James A. Hanna, Myunghwan Byun, Christian D. p. 7

Where the paper says it

[21] Jungwook Kim, James A. Hanna, Myunghwan Byun, Christian D.

Santangelo, and Ryan C. p. 7
Where the paper says it

Hanna, Myunghwan Byun, Christian D. Santangelo, and Ryan C.

Hayward. Designing responsive buckled surfaces by halftone gel lithography

[23] Manfredo P do Carmo. Differential Geometry of Curves and Surfaces: Revised and Updated Second Edition

[24] Md Samsu Alam, Abhijit Majumdar, Anindya Ghosh, et al. Role of fibre, yarn and fabric parameters on bending and shear behaviour of plain woven fabrics

[25] Jinlian. Hu and England) Textile Institute (Manchester. p. 8

Where the paper says it

[25] Jinlian. Hu and England) Textile Institute (Manchester.

Structure and mechanics of woven fabrics

[26] John Skelton. Fundamentals of fabric shear

[27] F. Barsi, B. Carboni, and W. p. 8

Where the paper says it

Textile Research Journal, 46(12):862–869, 1976. Carboni, and W.

Lacarbonara. p. 8
Where the paper says it

Carboni, and W. Lacarbonara.

A new mechanical model of short wire ropes: Theory and experimental validation

[28] Mohammed Adnan, Robert A Pinnick, Zhao Tang, Lauren W Taylor, Sushma Sri Pamulapati, Gianni Royer Carfagni, and Matteo Pasquali. Bending behavior of cnt fibers and their scaling laws

[29] Jason Cantarella, Joseph H G Fu, Rob Kusner, John M Sullivan, and Nancy C Wrinkle. Criticality for the gehring link problem

[30] Yehuda Pinchover and Jacob Rubinstein. An Introduction to Partial Differential Equations

[31] Ligang Liu, Lei Zhang, Yin Xu, Craig Gotsman, and Steven J. Gortler. A local/global approach to mesh parameterization

[32] Andrew O. Sageman-Furnas, Albert Chern, Mirela Ben-Chen, and Amir Vaxman. p. 8

Where the paper says it

[32] Andrew O. Sageman-Furnas, Albert Chern, Mirela Ben-Chen, and Amir Vaxman.

Chebyshev nets from commuting polyvector fields

[33] Etienne Corman and Keenan Crane. Rectangular surface parameterization

[34] Lauren Niu, Genevieve Dion, and Randall D. Kamien. Geometric modeling of knitted fabrics

[35] Arif Kurbak and Ozgur Ekmen. Basic studies for modeling complex weft knitted fabric structures part i: A geometrical model for widthwise curlings of plain knitted fabrics

[36] Code available at https://doi.org/10.5281/zenodo. p. 8

Where the paper says it

[36] Code available at https://doi. 5281/zenodo.

18324284

[37] John M Sullivan and Nancy C Wrinkle. Some ropelength-critical clasps

[9] Manfredo P do Carmo. Differential Geometry of Curves and Surfaces: Revised and Updated Second Edition

[10] Gauss’s Theorema Egregium connects the metric tensor and the Gaussian curvature of a surface K (u, v)

[A.2] Given a general two-dimensional metric ds2 = E(u, v) du2 + 2F(u, v) du dv + G(u, v) dv2, the Gaussian curvature is given by the Brioschi formula The metric tensor that describes polar weaving is, in polar coordinates: ds2 = E (α (r, θ)) dr2 + r2G (α (r, θ)) dθ2, and the Gaussian curvature is then K(r, θ) = [EG′ + E′G] α2θE' + r2α2rG' - 2GE α2θE'' + α2r3G'' + αrr G' + 2αθrr G'

[A.3] The inverse design problem for a tight smart fabric whose local deformation obeys the metric ds2 = E (α (u, v)) du2 + G (α (u, v)) dv2, can be written as a hyperbolic PDE G′αuu + E′αvv = (EG)′2EGα2vE' + α2uG' - α2vE'' - α2uG'' - 2EGK(u, v) The associated Cauchy problem requires, in addition to Eq. A.2, also the value of α (u, 0) and αv (u, 0) on an initial curve v = 0 (transverse to the characteristic curves) Geometrically, the solution is constructed as follows: we start with an arbitrary curve on the target surface, which we identify with the u-curve v = 0 Iterating this construction produces a local rectangular coordinate system on the surface, which obeys the metric tensor Eq. p. 3

Where the paper says it

To complete a well-posed Cauchy initial-value problem, the hyperbolic PDE requires a Cauchy-type initial condition [30]. Thus, the problem is fully set given the following: (i) a target surface, (ii) a curve on the target surface, and

A.1

[A.4] In the symmetric case of a surface of revolution, Eq. p. 3

Where the paper says it

In some cases, additional symmetries allow for significantly simplifying Eq. This is the case for target surfaces of revolution, where we assume that the thread structure shares the same symmetry, namely that threads lie along latitudes and longitudes of the surface.

6 reads α′′(u) + A[α(u)] α' (u)2 = B[α(u)] K s(u), s(u) = Z u0qEα˜u d˜u, where A[α] = d/dα log G'(α) p E(α)G(a)!, B[a] = -2E(a)G(a) G'(a).

[A.5] To generate global parameterizations on discrete triangular meshes (as shown in Fig. p. 4

Where the paper says it

Therefore, given a robust and precise technique for the manufacturing of smart fabrics with given thread-shrinkage profiles, we provide an explicit recipe for fabrics that will deform into an arbitrary target geometry upon actuation, at least locally. For practical implementations of the inverse problem, the target surface will be given in a discrete form, such as a triangular mesh.

4), we utilized the Rectangular Surface Parameterization framework developed by Corman and Crane [33] The modification involves calculating the error potential err diag and its associated Jacobian derivatives to guide the optimizer toward the tight-weave solution

Table 1: List of Appendices and Contents

Code File Content Description

constrained optimization.py Numerical optimization Python code

woven fabric optimization.jl Energy minimization Julia code

Table 2: List of code files available in [36]

Code available at https://doi.org/zenodo. p. 8

Where the paper says it

[36] Code available at https://doi. 5281/zenodo.

18324284

The paper is 2601.15746. EVERY sentence must end with a reference token; the token names the page and quotes the words it rests on, copied word for word from that page; N is the page number counting from 1 — the same N as that page's Page 1 line, and N is the same as that page's Page 1 line

The paper is 2601.15746. EVERY sentence must end with a reference token; the token names the page and quotes the words it rests on, copied word for word from that page; N is the same as that page's Page 1 line, and N is the same as that page's Page 1 line

The paper is 2601.15746.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Precise Inverse Design of Smart Fabrics: The improved system can solve for an explicit recipe for fabrics that will deform into an arbitrary target geometry upon actuation, at least locally, by solving a nonlinear hyperbolic partial differential equation.

  2. Geometric Constraint Enforcement: The system can enforce the tight-weave constraints by minimizing a penalty term that enforces the specific coupling between the metric components E and G derived from our physical model during numerical optimization for rectangular surface parameterizations.

  3. Solution Construction for Target Shapes: The system can obtain an explicit function alpha(u, v) which is then transcribed to the explicit shrinkage of each thread at each point using Eq. 5, allowing it to predict the resulting surface shape from a given thread actuation profile.

Abstract

We present a geometric framework for the inverse design of smart woven fabrics composed of non-uniformly shrinking threads. A sufficiently tight weaving structure imposes strong local criteria on the material deformation and reduces the local geometry to a single scalar degree of freedom. Control over this degree of freedom can be achieved through simple calibration for each specific material system, via either mechanical experiments or numerical simulations. This reduction allows us to inverse-design a woven smart fabric, that conforms to an arbitrary target geometry when actuated, by solving a nonlinear hyperbolic partial differential equation. We validate this approach by deriving the thread-level actuation required for specific target geometries. We present both exact analytic solutions for symmetric shapes and a numerical optimization method for arbitrary freeform surfaces. Our manuscript provides motivation for weaving smart threads into fabrics, and provides a practical framework for achieving programmable, complex three-dimensional shaping of these novel metamaterials.

Related papers