Stress Asymmetry in Hard Magnetic Soft Materials Preprint: To appear in Journal of Applied Mechanics ( DOI:10.1115/1.4071532 )

H. Gökçen Güner, Francois Barthelat, John D. Clayton, Carlos Mora-Corral, Noel Walkington, Kaushik Dayal · 2026

Read Stress Asymmetry in Hard Magnetic Soft Materials Preprint: To appear in Journal of Applied Mechanics ( DOI:10.1115/1.4071532 ) by H. Gökçen Güner, Francois Barthelat, John D. Clayton, Carlos Mora-Corral, Noel Walkington, Kaushik Dayal (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-00238.txt.

Introduction

Güner and five colleagues ask whether Cauchy stress stays symmetric in hard magnetic soft materials, plastics filled with hard magnetic grains. Two energy-wise matching versions, linked by switching terms between referential and current magnetization, usually give unlike Cauchy stresses. The first version yields symmetric stress, while the current version often yields asymmetric stress. At the lowest-energy point, however, both stresses turn symmetric and their divergences match.

Catalog record

CREATORH. Gökçen Güner, Francois Barthelat, John D. Clayton, Carlos Mora-Corral, Noel Walkington, Kaushik Dayal
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34710
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.00238

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Stress Asymmetry in Hard Magnetic Soft Materials Preprint: To appear in Journal of Applied Mechanics ( DOI:10.1115/1.4071532 ) H. Gökçen Güner, Francois Barthelat, John D. Clayton, Carlos Mora-Corral, Noel Walkington, Kaushik Dayal Abstract Hard magnetic soft materials – soft polymers embedded with hard magnetic particles – are modeled using continuum magnetomechanical formulations in which the deformation and the magnetization field are the primary kinematic variables. A recent question in such formulations is whether the Cauchy stress is symmetric, which is directly related to frame invariance and angular momentum balance. This note discusses energetically equivalent formulations, related by a change of variables between referential and current descriptions of the magnetization, and shows that they generally yield different Cauchy stresses, including a change in their symmetry. Specifically, the formulation based on a referential magnetization produces a symmetric Cauchy stress, while that based on a current magnetization generally yields an asymmetric Cauchy stress. We highlight that when the internal variable (magnetization field) is at the energy-minimizing equilibrium configuration, the divergences of these stresses are the same, and both stresses are symmetric. Dedicated to Prof. David J. Steigmann 1 Introduction Hard magnetic soft materials (HMSM), composed of hard magnetic particles that are embedded in a soft polymer matrix, have recently received attention in mechanics due to their potential for applications in robotics and related fields KYZ+ 18 ; ZKC+ 19 ; ZZ 22 ; LWSZ 22 ; WHZ+ 20 ; WZD+ 20 . Homogenized continuum models – that is, constitutive magnetomechanical models that do not consider individual particles but only their collective response at larger scales – use the deformation and magnetization fields as the primary kinematic variables ZKC+ 19 . Recent work in HMSM has brought up the issue of frame invariance and the consequent symmetry of the Cauchy stress, e.g., RJT+ 23 ; DO 24 . In this short note, we discuss that equivalent energetic formulations – which are related by an elementary change of variables – can generally lead to different Cauchy stresses, including changing whether they are symmetric. The class of changes of variables that we consider are between referential and current descriptions of the internal variables; the magnetization field, in HMSM. Here, we emphasize that we are not switching between Lagrangian and Eulerian descriptions; all quantities are Lagrangian. What does change, however, is that the referential and current descriptions transform in different ways under a rigid rotation of the current configuration: the referential description is invariant, whereas the current description is generally not invariant. A simple example is the temperature gradient: the gradient with respect to referential coordinates is invariant under a rigid rotation of the current configuration, whereas the gradient with respect to current coordinates is not invariant; however, the descriptions are equivalent in that it is possible to transform between them Gur 82 . An elementary example shows the main point of this note. Consider a function \displaystyle w(f,n) , and define \displaystyle w_{0}(f,n_{0}(f,n)):=w(f,n) through the change of variables \displaystyle n_{0}(f,n) . The derivatives are related by the expression: \begin{Bmatrix}\frac{\partial w}{\partial f}\\[10.00002pt] \frac{\partial w}{\partial n}\end{Bmatrix}=\begin{bmatrix}1&\frac{\partial n_{0}}{\partial f}\\[10.00002pt] 0&\frac{\partial n_{0}}{\partial n}\end{bmatrix}\begin{Bmatrix}\frac{\partial w_{0}}{\partial f}\\[10.00002pt] \frac{\partial w_{0}}{\partial n_{0}}\end{Bmatrix} (1) It is clear that \displaystyle\frac{\partial w}{\partial f}\neq\frac{\partial w_{0}}{\partial f} in general, unless \displaystyle\frac{\partial w}{\partial n}=0 or equivalently \displaystyle\frac{\partial w_{0}}{\partial n_{0}}=0 . Our central point is that, analogously, the stresses – defined through the variational derivatives of the energy – are sensitive to a transformation of the internal variables. Section 2 provides the general framework; Section 3 discusses simple examples from liquid crystal elastomers; and Section 4 discusses hard magnetic soft matter. 1.A Notation. • We use boldface to denote vectors and tensors. • We use subscripts of \displaystyle 0 to represent quantities related to the reference configuration; e.g., \displaystyle\nabla_{0} represents differentiation with respect to the referential position \displaystyle{\mathbfit x}_{0} , and \displaystyle\nabla to represent differentiation with respect to the deformed position \displaystyle{\mathbfit x} . The deformation map is denoted \displaystyle\chi , i.e., \displaystyle{\mathbfit x}=\chi({\mathbfit x}_{0}) . We use \displaystyle{\mathbfit F}=\nabla_{0}\mathbfit{\chi} , \displaystyle J=\det{\mathbfit F} , and \displaystyle{\mathbfit C}={\mathbfit F}^{T}{\mathbfit F} . The standard relation between differentiation with respect to \displaystyle{\mathbfit x}_{0} and to \displaystyle{\mathbfit x} is then \displaystyle{\mathbfit F}^{-T}\nabla_{0}(\cdot)=\nabla(\cdot) . • For a 2nd order tensor \displaystyle\mathbfit{\sigma} , we use \displaystyle\skw(\mathbfit{\sigma}):=\tfrac{1}{2}\left(\mathbfit{\sigma}-\mathbfit{\sigma}^{T}\right) . • We use \displaystyle\delta{\mathbfit x} to denote the variation of a function \displaystyle{\mathbfit x} , and \displaystyle\updelta_{{\mathbfit x}}A to denote the variational derivative of the functional \displaystyle A[{\mathbfit x}] with respect to the function \displaystyle{\mathbfit x} . 2 General Framework Consider an energy \displaystyle E[{\mathbfit x},{\mathbfit n}] with a corresponding energy density \displaystyle W(\nabla_{0}{\mathbfit x},{\mathbfit n}) ; that is 1 1 1 Boundary conditions do not play a significant role in this paper, and consequently we assume that they are such that the boundary terms vanish. , \displaystyle E[{\mathbfit x},{\mathbfit n}]=\int_{\Omega}W(\nabla_{0}{\mathbfit x},{\mathbfit n}) . We use \displaystyle{\mathbfit n} as a vector internal variable that describes the current value and is written in the Lagrangian sense. We next introduce \displaystyle{\mathbfit n}_{0} as the referential description of \displaystyle{\mathbfit n} . They are related by the mapping \displaystyle{\mathbfit n}=\hat{\mathbfit n}({\mathbfit F},{\mathbfit n}_{0}) , where \displaystyle\hat{\mathbfit n} is the function that relates \displaystyle{\mathbfit n}_{0} to \displaystyle{\mathbfit n} . The mapping is simply a change of variables and can be fairly general; for now, we only require that it be invertible and that there exist \displaystyle\hat{\mathbfit n}_{0} such that \displaystyle{\mathbfit n}_{0}=\hat{\mathbfit n}_{0}({\mathbfit F},{\mathbfit n}) . We use the mapping between \displaystyle{\mathbfit n} and \displaystyle{\mathbfit n}_{0} to introduce the energy density \displaystyle W_{0}({\mathbfit F},{\mathbfit n}_{0}) , defined by: W_{0}({\mathbfit F},{\mathbfit n}_{0}):=W({\mathbfit F},{\mathbfit n}=\hat{\mathbfit n}({\mathbfit F},{\mathbfit n}_{0})) (2) This change of variables leads us to introduce \displaystyle E_{0}[{\mathbfit x},{\mathbfit n}_{0}]=\int_{\Omega}W_{0}(\nabla_{0}{\mathbfit x},{\mathbfit n}_{0}) . 2.A Implication of Frame Invariance We now consider \displaystyle{\mathbfit n} to be an objective vector. Then, any rigid rotation \displaystyle{\mathbfit Q} applied to the current configuration leads to the transformations \displaystyle{\mathbfit F}\to{\mathbfit Q}{\mathbfit F} and \displaystyle{\mathbfit n}\to{\mathbfit Q}{\mathbfit n} . From the usual arguments of frame invariance of \displaystyle E , we require: W({\mathbfit F},{\mathbfit n})=W({\mathbfit Q}{\mathbfit F},{\mathbfit Q}{\mathbfit n}),\quad\forall{\mathbfit Q}\in SO(3) (3) Next, we require that \displaystyle\hat{\mathbfit n}_{0} satisfies \displaystyle\hat{\mathbfit n}_{0}({\mathbfit Q}{\mathbfit F},{\mathbfit Q}{\mathbfit n})=\hat{\mathbfit n}_{0}({\mathbfit F},{\mathbfit n}),\forall{\mathbfit Q}\in SO(3) . Consequently, any rigid rotation \displaystyle{\mathbfit Q} of the current configuration leaves \displaystyle{\mathbfit n}_{0} invariant. This gives: \begin{split}&W_{0}({\mathbfit F},\hat{\mathbfit n}_{0}({\mathbfit F},{\mathbfit n}))=W({\mathbfit F},{\mathbfit n})=W({\mathbfit Q}{\mathbfit F},{\mathbfit Q}{\mathbfit n})=W_{0}({\mathbfit Q}{\mathbfit F},\hat{\mathbfit n}_{0}({\mathbfit Q}{\mathbfit F},{\mathbfit Q}{\mathbfit n}))=W_{0}({\mathbfit Q}{\mathbfit F},\hat{\mathbfit n}_{0}({\mathbfit F},{\mathbfit n}))\\ &\implies W_{0}({\mathbfit F},{\mathbfit n}_{0})=W_{0}({\mathbfit Q}{\mathbfit F},{\mathbfit n}_{0})\end{split} (4) Following the structure of the Noether principle, we apply frame invariance to obtain conditions that relate to angular momentum balance LM 00 ; KS 72 ; Eri 76 ; SH 21 ; Gol 81 . We parameterize the rotation \displaystyle{\mathbfit Q} by \displaystyle s such that \displaystyle{\mathbfit Q}(0)={\mathbfit I},{\mathbfit Q}^{\prime}(0)={\mathbfit S} , where \displaystyle{\mathbfit S} is any skew tensor; explicitly, \displaystyle{\mathbfit Q}(s)=e^{s{\mathbfit S}} . We then differentiate ( 3 ) with \displaystyle s and evaluate at \displaystyle s=0 to get: {\mathbfit S}:\left(\frac{\partial W}{\partial{\mathbfit F}}{\mathbfit F}^{T}+\frac{\partial W}{\partial{\mathbfit n}}\otimes{\mathbfit n}\right)=0\implies\skw\left(\frac{\partial W}{\partial{\mathbfit F}}{\mathbfit F}^{T}+\frac{\partial W}{\partial{\mathbfit n}}\otimes{\mathbfit n}\right)=\mathbf{0} (5) where we have used the arbitrariness of \displaystyle{\mathbfit S} to conclude that the quantity in parentheses must be symmetric. A similar procedure applied to ( 4 ) gives: \skw\frac{\partial W_{0}}{\partial{\mathbfit F}}{\mathbfit F}^{T}=\mathbf{0} (6) 2.B Symmetry and Asymmetry of the

Read the complete text