Nonlinear dynamic elastic moduli from equilibrium stress fluctuations

F.E. Garbuzov, Y.M. Beltukov · 2026

Read Nonlinear dynamic elastic moduli from equilibrium stress fluctuations by F.E. Garbuzov, Y.M. Beltukov (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-2605-13703.txt.

Introduction

F. E. Garbuzov and Y. M. Beltukov derive transient link forms for linear and bent dynamic elastic moduli. Wobble rules allow finding moduli from balance tiny motion tests, avoiding plain off-balance bending plans. While quasi-static and linear dynamic moduli rules are known, no wobble rule exists for bent time-linked moduli under finite strains. Their rules start from DOLLS/SLLOD rules for spin-free motion, using time links of stress and Born-kinetic terms. They recover known quasi-static and linear results in the right limits.

Catalog record

CREATORF.E. Garbuzov, Y.M. Beltukov
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25496
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.13703

Browse the full library · About the Tower of Babel

Full text

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

Nonlinear dynamic elastic moduli from equilibrium stress fluctuations F.E. Garbuzov, Y.M. Beltukov Abstract Fluctuation formulas for elastic and viscoelastic moduli allow their computation from equilibrium molecular dynamics simulations, avoiding explicit nonequilibrium deformation protocols. While such expressions are well established for the quasi-static moduli, and also the linear dynamic moduli, no fluctuation formula exists for the nonlinear time-dependent moduli that govern anharmonic viscoelastic response under finite time-dependent strains. In this work we derive transient-time correlation function expressions for both the linear and the nonlinear dynamic moduli, starting from the DOLLS/SLLOD equations of motion for irrotational motion. The resulting formulas involve equilibrium time correlations of the stress tensor and Born–kinetic terms, and they recover the known quasi-static and linear dynamic results in the appropriate limits. 1 Introduction Knowing the elastic and viscoelastic moduli of a material is essential for predicting its mechanical response to external loads. Molecular dynamics offers a way to compute these moduli directly from the underlying interatomic interactions, thereby linking macroscopic mechanical properties to the microscopic system description. To identify the elastic moduli, one can either perform nonequilibrium simulations and deduce the moduli directly from the stress response to an applied strain, or perform equilibrium simulations and extract them from the stress or strain fluctuations. While the equilibrium approach is computationally attractive, its application requires rigorous fluctuation formulas. For the linear quasi-static (zero-frequency) elastic constants, fluctuation expressions based on strain fluctuations [ 11 ] and stress fluctuations [ 13 , 9 , 14 , 1 ] were obtained more than four decades ago. The extension of fluctuation methods to quasi-static nonlinear elastic moduli followed more recently [ 8 ] . For time-dependent deformations, a fluctuation formula for the linear viscoelastic moduli is also available [ 15 ] . A gap nevertheless remains in the understanding of the nonlinear dynamic mechanical response: no fluctuation formula exists for the nonlinear time-dependent moduli that govern the leading anharmonic contributions in viscoelastic materials subjected to small but finite time-dependent strains. The present work aims to fill this gap. The paper is organized as follows. Section 2 introduces the microscopic equations of motion suitable for modeling an induced strain and gives the microscopic expression for the stress tensor. Section 3 contains the main derivation: we obtain the nonequilibrium probability distribution function in Sec. 3.1 and use it in Sec. 3.2 to evaluate the averaged stress in the regime of small but finite strains, arriving at explicit expressions for the linear and nonlinear dynamic moduli. The paper concludes with a summary and outlook in Sec. 4 . 2 Microscopic equations The dynamics of a system of interacting particles is described by the Hamiltonian H(\{\vec{r}_{a}\},\{\vec{p}_{a}\})=\sum_{a=1}^{N}\frac{\vec{p}_{a}\cdot\vec{p}_{a}}{2m_{a}}+U(\{\vec{r}_{a}\}), (1) where \vec{r}_{a} , \vec{p}_{a} , and m_{a} are the coordinates, momenta, and masses of particle a , respectively. The phase space point is denoted by (\{\vec{r}_{a}\},\{\vec{p}_{a}\})=(\vec{r}_{1},\dots,\vec{r}_{N},\vec{p}_{1},\dots\vec{p}_{N}) , U is the interaction potential, and the central dot denotes the scalar (inner) product. The equations of motion derived from this Hamiltonian describe a closed system at equilibrium. To model nonequilibrium behavior corresponding to deformation, these equations must be modified. 2.1 Nonequilibrium equations of motion Several formulations of the equations of motion for homogeneous nonequilibrium flows have been developed, and there is ongoing discussion in the literature regarding their validity. An early Hamiltonian approach is the DOLLS method [ 7 ] , where the total Hamiltonian in Eq. ( 1 ) is augmented by a term that couples the particle momenta to the streaming velocity field: H_{\text{DOLLS}}=H+\sum_{a}\vec{p}_{a}\vec{r}_{a}:L, (2) with the streaming velocity gradient tensor L=\nabla\vec{v}^{\,T} . The dyadic (outer) product is implied by the juxtaposition of two vectors, and : denotes double contraction, so that the second term reads \sum_{a}\sum_{ij}p_{a,i}r_{a,j}L_{ij} in index notation. The corresponding Hamiltonian equations of motion are (3) ItwassoondiscoveredthattheDOLLSmethodproduceserroneousnormalstressesinthenonlinearregimewhensimulatingplanarshear [ 2 , 3 ] . ThisissuewasaddressedbytheSLLODalgorithm , whichreplacesthetranspose L^T by L inthemomentumequation:\begin{aligned} \dot{\vec{r}}_{a}&=\frac{\vec{p}_{a}}{m_{a}}+L\!\cdot\!\vec{r}_{a},\\ \dot{\vec{p}}_{a}&=-\frac{\partial U}{\partial\vec{r}_{a}}-L\!\cdot\!\vec{p}_{a}.\end{aligned}AlthoughtheseequationsarenotHamiltonian,theypreservephasespacevolume~\cite[cite]{[\@@bibref{}{EvansMorrissStatisticalMechanics2008}{}{}]}.\par TheSLLODequationshavebeenwidelyusedinnonequilibriummoleculardynamics,particularlyforplanarshearflow~\cite[cite]{[\@@bibref{}{EvansMorrissNumberDependence1989, SandbergHeyesSelfDiffusion1995, MundySiepmannDecaneUnder1995, AoyagiDoiMolecularDynamics2000, FadaeiBjorlingUnravelingPressure2023}{}{}]}.Forelongationalandmoregeneralflowstheproper‑SLLOD(p‑SLLOD)algorithm~wasintroduced~\cite[cite]{[\@@bibref{}{BaigEdwardsProperApproach2005}{}{}]},whichhoweverraisedanumberofobjections~\cite[cite]{[\@@bibref{}{DaivisToddSimpleDirect2006}{}{}]}.Moreover,p-SLLODproducesthestresstensorthatdiffersfromthestandardonethatfollowsfromtheIrving–Kirkwoodprocedure~\cite[cite]{[\@@bibref{}{IrvingKirkwoodStatisticalMechanical1950}{}{}]}.Forthesereasonswedonotadoptp‑SLLODinthepresentwork.\par DOLLSandSLLODdifferonlywhenthevelocity‑gradienttensor L isnon‑symmetric,i.e.,whentheflowinvolvesrotation(non‑zerovorticity).Becauseourfocusisonelasticresponse,werestrictourselvestoirrotationaldeformations,where$L$issymmetric($L=L^{T}$).UnderthisconditiontheDOLLSandSLLODequationscoincideidentically.Therefore,throughoutthispaperweemploythefollowingDOLLS/SLLODequationsofmotion:\begin{aligned} \dot{\vec{r}}_{a}&=\frac{\vec{p}_{a}}{m_{a}}+L(t)\!\cdot\!\vec{r}_{a},\\ \dot{\vec{p}}_{a}&=-\frac{\partial U}{\partial\vec{r}_{a}}-L^{T}(t)\!\cdot\!\vec{p}_{a},\qquad L=L^{T}.\end{aligned}Althoughweassume L tobesymmetricinwhatfollows,weretain L^T sothatourderivationsremainapplicableintheDOLLSframeworkevenwithoutthesymmetryconditionon L . 2.2 Pulled-back variables In solids, deformation is described by the deformation gradient F , mapping reference positions \vec{R}_{a} to current positions \vec{r}_{a} : \vec{r}_{a}=F(t)\cdot\vec{R}_{a}. (4) The canonically conjugate momenta \vec{P}_{a} corresponding to \vec{R}_{a} are defined as \vec{p}_{a}=F^{-T}(t)\cdot\vec{P}_{a}, (5) and the velocity gradient is related to F by L(t)=\dot{F}(t)\cdot F^{-1}(t). (6) We refer to \vec{R}_{a} and \vec{P}_{a} as the pulled-back coordinates and momenta. In the literature, it has been a common practice to apply the canonical transformation defined by Eqs. ( 4 ) and ( 5 ) to the Hamiltonian ( 1 ) to derive fluctuation formulas for quasi-static (zero-frequency) linear [ 10 , 1 , 9 ] and nonlinear elastic moduli [ 8 ] . In the dynamic case, the DOLLS/SLLOD equations ( 3 )–( 2.1 ) are consistent with this transformation in the sense that, for a suddenly applied deformation, F(t)=\begin{cases}I,&t<0,\\ F,&t>0,\end{cases}\qquad\dot{F}(t)=(F-I)\delta(t), (7) the pulled-back variables \vec{R}_{a} and \vec{P}_{a} remain continuous at t=0 . This implies that the laboratory coordinates and momenta must undergo the jumps \displaystyle\vec{r}_{a}(0+) \displaystyle=F\cdot\vec{r}_{a}(0-), (8) \displaystyle\vec{p}_{a}(0+) \displaystyle=F^{-T}\cdot\vec{p}_{a}(0-). (9) We note that relations ( 8 ) and ( 9 ) hold in both DOLLS and SLLOD frameworks under the irrotational flow assumption adopted here. However, when rotational motion is present (i.e., when L is nonsymmetric), the two approaches diverge: in SLLOD the pulled-back momenta suffer a jump determined by the skew-symmetric part of L , and the jump relation for the laboratory momenta cannot be expressed as simply as in Eq. ( 9 ). Since we restrict ourselves to irrotational motion, this distinction lies outside our scope. By virtue of Eqs. ( 8 ) and ( 9 ), a suddenly applied strain produces an instantaneous change in the laboratory coordinates and momenta; thus, in this sense, they are strain-dependent. The pulled-back variables, in contrast, are free of such jumps: their strain dependence enters only through their time evolution. Expressing the DOLLS/SLLOD equations ( 3 ) and ( 2.1 ) in terms of these pulled-back variables yields (10) where C istherightCauchy--Greendeformationtensor.\par\par\par 2.3 Strain For small but finite strains, we use the Lagrangian finite strain tensor E=\frac{1}{2}(C-I)=\frac{1}{2}(F^{T}\cdot F-I). (11) Due to the pullback transformation in Eqs. ( 4 ) and ( 5 ), the Hamiltonian H becomes an explicit function of the strain: H(\Gamma,E)=\sum_{a}\frac{\vec{P}_{a}\cdot C^{-1}\cdot\vec{P}_{a}}{2m_{a}}+U(\{F\cdot\vec{R}_{a}\}), (12) where \Gamma=(\{\vec{R}_{a}\},\{\vec{P}_{a}\}) denotes the set of all pulled-back particle coordinates and momenta, and we assume that U depends on F only through C (and therefore through E ). The simplest example of such a potential is a pair potential which depends only on intermolecular distances (\vec{r}_{a}-\vec{r}_{b})^{2}=\vec{r}_{ab}^{2}=\vec{R}_{ab}\cdot C\cdot\vec{R}_{ab} , thus C plays the role of a metric tensor for the deformed space [ 10 ] . However, we keep U in unspecified form for generality. The stress conjugate to the Lagrangian strain is the second Piola–Kirchhoff stress tensor: S(\Gamma,E)=\frac{1}{V_{0}}\frac{\partial H}{\partial E}=\frac{1}{V_{0}}\sum_{a}\biggl(-C^{-1}\cdot\frac{\vec{P}_{a}\vec{P}_{a}}{m_{a}}\cdot C^{-1}+C^{-1}\cdot\frac{\partial U}{\partial\vec{R}_{a}}\vec{R}_{a}\biggr) (13) where V_{0} is the reference (undeformed) volume. Using the usual expression for the second Piola-Kirchhoff stress through the Cauchy stress \sigma S=(\det F)F^{-1}\cdot\sigma\cdot F^{-T}, (14) and the canonical transformation (Eqs. ( 4 ), ( 5 )) we recover the expression for the virial stress: \sigma=\frac{1}{V}\sum_{a}\left(\frac{\partial U}{\partial\vec{r}_{a}}\vec{r}_{a}-\frac{\vec{p}_{a}\vec{p}_{a}}{m_{a}}\right),\quad V=V_{0}\det F, (15) thus making it a microscopic expression for the Cauchy stress. We note that the virial stress is not symmetric in general, but under our assumptions it is

Read the complete text