---
title: Generalized Strain Transformation Zones (GSTZ)
url: https://www.emergentmind.com/topics/generalized-strain-transformation-zones-gstz
type: topic
---

# Generalized Strain Transformation Zones (GSTZ)

Generalized Strain Transformation Zones (GSTZ) denotes a family of extensions of shear-transformation-zone descriptions for amorphous solids in which the elementary loci of plastic or relaxation activity are generalized beyond the classic athermal, volume-preserving, single-species STZ picture. In the literature summarized here, the generalization proceeds along three distinct but related axes: a statistical description of near-threshold sites through the singular density \(P(x)\sim x^\theta\) controlled by the elastic kernel [1307.1646]; a multi-species STZ formulation with a spectrum of activation barriers \(\Delta_i\) and a shared effective temperature \(\chi\) [1203.1965]; and finite-temperature, tensorial local-state descriptions in which coupled deviatoric and volumetric strains govern thermally mediated rearrangements in undeformed or weakly deformed glasses [2509.06319, 2507.13960]. Taken together, these formulations suggest a broader GSTZ framework for amorphous plasticity in which local instability, heterogeneous activation, and tensorial strain content are treated as interlocking aspects of the same phenomenon.

## 1. Terminological scope and relation to classical STZ theory

Classical STZ theory, as explicitly contrasted in the finite-temperature silica study, “postulates localized regions (‘two-state’ defects) that carry plastic shear in athermal, zero-\(T\) limit” [2507.13960]. In the undeformed-glass formalism, classic STZ are described as assuming volume-preserving deformation, \(\epsilon_{\rm vol}=0\), whereas GSTZ “recognize that thermally induced hops involve coupled changes in both shear and density” [2509.06319]. In the high-strain-rate metallic-glass analysis, the “generalized” label refers instead to a statistical ensemble of STZ species with distinct formation or activation energies \(\Delta_i\), replacing the single-species assumption [1203.1965].

The literature therefore uses GSTZ in more than one precise sense. One sense is statistical: flow defects in amorphous solids are represented as a continuum of internal barriers, each weighted by Boltzmann factors controlled by the effective temperature \(\chi\) [1203.1965]. A second sense is kinematic and energetic: each local zone carries a tensorial strain state that can be decomposed into deviatoric and volumetric parts, with elastic energy
\[
u_{\rm el} = \tfrac12\,G\,\|e\|^2 + \tfrac12\,K\,(\epsilon_{\rm vol})^2
\]
and thermally activated transformations driven by the reduction of a local barrier [2509.06319]. A third sense, derived from the stability theory of amorphous solids, is that the population of susceptible regions is itself singular near instability, with \(P(x)\sim x^\theta\), where the exponent \(\theta\) is fixed by the non-monotonicity and range of the elastic propagator \(\mathcal G(\mathbf r)\) [1307.1646].

A plausible implication is that GSTZ is best regarded not as a single closed model but as a research program for extending STZ theory to heterogeneous disorder, nontrivial elastic kernels, and finite-temperature local strain states.

## 2. Near-threshold instability density and marginal stability

In the stability-based formulation, each mesoscopic region \(i\) has local shear stress \(\sigma_i\), a uniform yield threshold \(\sigma_i^{\rm th}=1\), and a “distance to instability”
\[
x_i \equiv 1-\sigma_i.
\]
When \(x_i<0\), the site is mechanically unstable and rearranges [1307.1646]. The density \(P(x)\) is defined as the density of sites whose distance to instability lies in \([x,x+dx]\), and near \(x=0^+\) it takes the singular form
\[
P(x)\sim x^\theta,\qquad x\to0^+,
\]
with \(\theta\ge 0\) set by the elastic interactions among rearranging regions [1307.1646].

In the fully mean-field limit, where all kicks to \(\sigma_j\) are i.i.d. random, the density \(P_1(x)\) of active sites obeys the steady-shear Fokker-Planck equation
\[
\frac{\partial P_1(x)}{\partial t}
= \dot\gamma\Bigl[D\,\frac{\partial^2P_1}{\partial x^2}
+ \lambda\,\frac{\partial P_1}{\partial x}
+ \delta(x-1)\Bigr]
-\frac{\theta(-x)\,P_1(x)}{\tau_c},
\]
where \(D\equiv \eta_0^2/6\) is the stress-diffusion constant due to random kicks of amplitude \(\eta_0\), \(\lambda\) enforces \(\langle x\rangle=1-\sigma\), and \(\tau_c\) is the local collapse timescale [1307.1646]. At the critical stress \(\sigma=\sigma_c\), where \(\dot\gamma\to0\), the absorbing boundary at \(x=0\) imposes
\[
P(x\to0^+) \propto x^1,
\]
so that \(\theta_{\rm MF}=1\) [1307.1646].

The crucial physical distinction from depinning is the non-monotonicity of the Eshelby kernel. For amorphous plasticity, the far-field propagator \(\mathcal G(\mathbf r)\) changes sign with angle, so a distant plastic event can either stabilize or destabilize another site. This re-stabilization depletes the density of near-threshold excitations and produces \(\theta>0\) [1307.1646]. The same paper derives a lower bound on \(\theta\) from the requirement that a plastic event not trigger runaway avalanches. If a kick scales as \(\delta\sigma(r)\sim \eta/r^\alpha\), then the mean number of secondary events scales as
\[
m \sim \int_{r=1}^L\!\Bigl[\frac{\eta}{r^\alpha}\Bigr]^{\theta+1}r^{d-1}\,dr
\sim \eta^{\theta+1}L^{\,d-\alpha(\theta+1)}.
\]
Marginal stability, \(m\lesssim 1\) as \(L\to\infty\), yields
\[
\theta\ge
\begin{cases}
1, & d\ge 2\alpha,\\
\frac{d}{\alpha}-1, & \tfrac d2\le \alpha \le d,\\
0, & \alpha>d.
\end{cases}
\]
For quadrupolar interactions, \(\alpha=d\), simulations give \(\theta_{d=2}\approx 0.6\) and \(\theta_{d=3}\approx 0.4\), both at the yield stress \(\sigma_c\) and after an athermal quench to \(\sigma=0\) [1307.1646].

This stability theory provides a precise statistical meaning for GSTZ density: the population of zones close to failure is not regular at threshold but pseudogapped. The same work also relates \(\theta\) to the smallest gap \(x_{\min}=\min_i x_i\), for which weak-correlation arguments give \(\langle x_{\min}\rangle\sim N^{-1/(\theta+1)}\equiv N^{-\xi}\), with numerical estimates \(\xi_{d=2}\approx 0.63\) and \(\xi_{d=3}\approx 0.71\), implying \(\theta\approx 0.59\) and \(\theta\approx 0.41\), respectively [1307.1646].

## 3. Multi-species GSTZ and effective-temperature thermodynamics

A different generalization appears in the high-strain-rate metallic-glass theory, where the single-species STZ model is extended to a spectrum of species \(i\) labeled by activation energies \(\Delta_i\) [1203.1965]. The order parameters are the number densities \(n_+^{(i)}\) and \(n_-^{(i)}\) of STZ’s whose internal transition axes are aligned parallel or antiparallel to the applied shear stress \(s\), their sum \(\Lambda_i=n_+^{(i)}+n_-^{(i)}\), the effective temperature \(\chi\equiv k_B T_{\rm eff}/e_Z\), the kinetic temperature \(\theta\equiv k_B T/e_Z\), the thermal noise strength \(\rho(\theta)\), the mechanically generated noise \(\Gamma(s,\chi)\), and the dimensionless plastic strain rate \(q\equiv \tau_0 \dot\gamma^{pl}\) [1203.1965].

For a single species, the steady-state master equation is
\[
\tau_0 \dot n_\pm
= R(\pm s)\,n_\mp - R(\mp s)\,n_\pm
+ [\rho(\theta)+\Gamma(s,\chi)]\,[n_\infty(\chi)-n_\pm],
\]
with
\[
n_\infty(\chi)=\frac{1}{v_0}e^{-1/\chi}.
\]
The plastic strain rate is
\[
q=\epsilon_0 v_0 [R(+s)n_- - R(-s)n_+],
\]
and the transition rates are
\[
R(\pm s)=\Gamma_0 e^{-\Delta/k_B T}\,e^{\pm v_0 s/(e_Z\chi)}.
\]
From these, one defines
\[
{\cal C}(s)=\tfrac12 [R(+s)+R(-s)],\qquad
{\cal T}(s,\chi)=\frac{R(+s)-R(-s)}{R(+s)+R(-s)}=\tanh\!\bigl(v_0s/(e_Z\chi)\bigr).
\]
The mechanical noise is
\[
\Gamma(s,\chi)=\frac{q\,s}{\epsilon_0 s_0}e^{1/\chi},
\]
and the effective-temperature equation can be written as
\[
c_{\rm eff}\,\dot\chi = s\,\dot\gamma^{pl}-\kappa(\chi-T)
\]
or, near steady state,
\[
\chi = \frac{\Gamma(s,\chi)\,\hat\chi(q)+\kappa \rho(\theta)\theta}{\Gamma(s,\chi)+\kappa \rho(\theta)}.
\]
The steady-state driving-rate value \(\hat\chi(q)\) is defined through
\[
\frac{1}{q}=\frac{1}{q_0}\exp[A/\hat\chi+\alpha_{\rm eff}(\hat\chi)],
\]
with
\[
\alpha_{\rm eff}(\hat\chi)= \frac{\chi_1}{\hat\chi-\chi_0}\exp\!\Bigl[-b(\hat\chi-\chi_0)/(\chi_A-\chi_0)\Bigr]
\]
[1203.1965].

The generalized, multi-species version replaces the single Boltzmann factor \(e^{-1/\chi}\) by \(e^{-\Delta_i/\chi}\), introduces a continuous distribution \(p(\Delta)\) if desired, and sums or integrates over species:
\[
\Lambda=\sum_i \Lambda_i=\sum_i e^{-\Delta_i/\chi},
\]
\[
q=\epsilon_0 v_0 \sum_i [R_i(+s)\,n_-^{(i)}-R_i(-s)\,n_+^{(i)}].
\]
The mechanical noise and effective-temperature evolution retain the same thermodynamic form, but with weighted contributions from all species [1203.1965].

This formalism is used to resolve the low-strain-rate discrepancy near the glass transition. The single-species model underestimates the Newtonian viscosity and misses the upturn in \(\eta_N\) near \(T_g\), whereas in GSTZ the rare slow species with large \(\Delta_i\) contribute exponentially small but cumulatively large relaxation times \(\propto e^{\Delta_i/\chi}\), leading to Vogel-Fulcher-like divergence of viscosity, stretched-exponential relaxation, and Stokes-Einstein violations [1203.1965]. The central approximation is that the distribution \(p(\Delta)\) is sufficiently broad and that species remain independent in their stress responses except for the shared \(\chi\)-field.

## 4. Coupled shear–volumetric GSTZ in thermal undeformed glasses

A further generalization is developed for thermally driven structural relaxation in undeformed glass. There, local kinematics are obtained by fitting an affine deformation-gradient tensor \(H_i\) to neighbor displacements. If particle \(i\) is at \(X_i(t)\) at time \(t\) and at \(x_i(t+\Delta t)\) at time \(t+\Delta t\), with relative vectors
\[
\Delta X_{ij}=X_j(t)-X_i(t),\qquad
\Delta x_{ij}=x_j(t+\Delta t)-x_i(t+\Delta t),
\]
then \(H_i\) minimizes
\[
D_i^2(H)=\sum_{j\in N_i}\|\Delta x_{ij}-H\cdot \Delta X_{ij}\|^2,
\]
giving
\[
H_i=A_i\cdot B_i^{-1},
\]
with
\[
A_i^{\delta\alpha}=\sum_j \Delta x_{ij}^\delta \Delta X_{ij}^\alpha,\qquad
B_i^{\alpha\beta}=\sum_j \Delta X_{ij}^\alpha \Delta X_{ij}^\beta.
\]
The associated non-affine displacement measure is
\[
D^2_{{\rm min}(i)}=(1/N_i)\sum_{j\in N_i}\|\Delta x_{ij}-H_i\Delta X_{ij}\|^2,
\]
and the local strain tensor is defined by
\[
H_i=I+\epsilon_i,\qquad \epsilon_i=H_i-I
\]
[2509.06319].

The strain is then decomposed into volumetric and deviatoric parts. In \(d\) dimensions,
\[
\epsilon_{\rm vol}=\frac{1}{d}\,{\rm tr}\,\epsilon,
\]
and in 2D,
\[
\epsilon_{\rm vol}=\tfrac12(\epsilon_{xx}+\epsilon_{yy}).
\]
The deviatoric tensor is
\[
e_{\alpha\beta}=\epsilon_{\alpha\beta}-\epsilon_{\rm vol}\delta_{\alpha\beta},
\]
with 2D components
\[
e_{xx}=\tfrac12(\epsilon_{xx}-\epsilon_{yy}),\quad
e_{yy}=-e_{xx},\quad
e_{xy}=\epsilon_{xy},\quad e_{yx}=\epsilon_{yx}.
\]
The norms are
\[
\|\epsilon\|=\sqrt{\epsilon_{xx}^2+\epsilon_{yy}^2+\epsilon_{xy}^2+\epsilon_{yx}^2},
\]
\[
\|e\|=\sqrt{e_{xx}^2+e_{yy}^2+e_{xy}^2+e_{yx}^2},
\]
\[
|\epsilon_{\rm vol}|=\bigl|\tfrac12(\epsilon_{xx}+\epsilon_{yy})\bigr|.
\]
A local volume-change indicator is provided by \(\det H_i\simeq A'/A\), so \(\det H_i<1\) indicates local compression and \(\det H_i>1\) local dilation [2509.06319].

Within this framework, a zone is defined by elevated local strains, both \(\|e_i\|\) and \(|\epsilon_{{\rm vol},i}|\). Its elastic energy density is
\[
u_{\rm el}=\tfrac12 G\|e\|^2 + \tfrac12 K(\epsilon_{\rm vol})^2,
\]
and the activation barrier for local structural relaxation is reduced by strain. In the simplest Arrhenius form,
\[
{\rm rate}_i \sim \exp\!\Bigl[-(\Delta E_0-\alpha G\|e_i\|^2-\beta K\epsilon_{{\rm vol},i}^2)/(k_B T)\Bigr].
\]
Zones for which the combined shear and volumetric strain drives the effective barrier to zero undergo plastic rearrangements under thermal fluctuations [2509.06319]. The local free energy may be expanded as
\[
F(e,\epsilon_{\rm vol})=F_0+\tfrac12 G\|e\|^2+\tfrac12 K\epsilon_{\rm vol}^2 -TS+\ldots,
\]
where \(S\) is a softness entropy-like measure [2509.06319].

In this construction, the athermal shear-driven limit is recovered when \(T\to0\) and \(\epsilon_{\rm vol}=0\), whereas in the thermal undeformed case \(\sigma=0\) and both \(e\) and \(\epsilon_{\rm vol}\) build up spontaneously through noise [2509.06319]. The work emphasizes that undeformed glass under thermal fluctuations differs qualitatively from shear-driven response: “while the shear deformation response is dominated by volume preserving deviatoric strain, changes in local density must be considered to model response of undeformed glass under thermal noise” [2509.06319].

## 5. Computational identification and prediction of GSTZ

The thermal undeformed-glass formalism provides an explicit analysis pipeline for locating GSTZ in simulation data. Avalanches are detected by monitoring mean-square displacement versus time or frame index and identifying sudden jumps above a threshold, then cross-checking with discontinuities in shear stress \(\sigma_{xy}\) and the Hessian’s lowest eigenvalue \(\lambda_{\min}\) [2509.06319]. For each frame, one computes \(H_i\), \(\epsilon_i\), and \(D^2_{{\rm min}(i)}\), then visualizes the scalar and vector fields of \(D^2_{\min}\), \(\|e\|\), \(|\epsilon_{\rm vol}|\), and \(\det H\). Zones of elevated values are taken as putative GSTZs [2509.06319]. Persistent homology via the Vietoris-Rips complex is used to compute the Betti number \(\beta_1\); avalanches coincide with sharp drops in \(\beta_1\), indicating network rearrangements [2509.06319]. Softness can be estimated from the depth of the Ramakrishnan-Yussouff caging potential through
\[
\beta \phi(\Delta r=0)= -2\pi \rho \int_{r_{\min}}^{r_{\max}} r\,c(r)\,g(r)\,dr,\qquad
S=\frac{1}{|\beta \phi(0)|},
\]
and Pearson or Spearman correlations among \(S\), \(D^2_{\min}\), \(\|e\|\), \(\epsilon_{\rm vol}\), \(\det H\), and mobility \(m\) are reported to cluster GSTZs robustly [2509.06319].

A complementary prediction framework is developed through quasistatic Gaussian Phase Packets (GPPs) for 2D silica glass at finite temperature [2507.13960]. There the canonical density in phase space is approximated as a Gaussian,
\[
f(z)=\frac{1}{Z}\exp\!\bigl[-\tfrac12 (z-\bar z)^T \Sigma^{-1}(z-\bar z)\bigr],
\]
with an independent-atom approximation
\[
f(z)=\prod_i f_i(q_i,p_i),
\]
\[
f_i(q_i,p_i)=\frac{1}{Z_i}\exp\!\Bigl[-\tfrac12(q_i-\bar q_i)^T (\Sigma_i^{qq})^{-1}(q_i-\bar q_i)-\tfrac12(m_i k_B T)^{-1}|p_i|^2\Bigr].
\]
The Helmholtz free energy is
\[
F=\int f(z)\,[U(q)+k_B T\ln f(z)]\,dz,
\]
which becomes
\[
F[\{\bar q_i,\Sigma_i\}]
=\sum_i \Bigl\{\langle U_i(q)\rangle_{f_i}
+ k_B T\,\tfrac12 \ln[(2\pi e)^d \det \Sigma_i^{qq}] \Bigr\}.
\]
Stationarity requires force balance and
\[
k_B T(\Sigma_i^{qq})^{-1}_{\alpha\beta}
+\langle f_i^\alpha \delta q_i^\beta + f_i^\beta \delta q_i^\alpha\rangle_{f_i}=0,
\]
which determines the directional covariance \(\Sigma_i^{\alpha\beta}\) as a function of \(T\) and applied strain \(\epsilon\) [2507.13960].

From the covariance, a scalar local susceptibility is defined as
\[
S_i=\sqrt{\det \Sigma_i}
=\sqrt{\Sigma_i^{xx}\Sigma_i^{yy}-(\Sigma_i^{xy})^2}.
\]
The GSTZ detection algorithm is then: compute \(\Sigma_i(0,T)\); form the map \(S_i=\sqrt{\det\Sigma_i}\); identify top-\(k\) atoms with largest \(S_i\); cluster them into small groups of \(\sim 10\)–\(30\) atoms as initial GSTZs; apply a small affine strain increment; resolve the stationarity equations; recompute \(S_i\); and repeat until the first irreversible atomic rearrangement occurs in one of the top-ranked clusters [2507.13960]. Optionally, short Metropolis-GPP sampling can be interleaved to mimic thermal barrier crossing and reproduce strain-rate dependence [2507.13960].

The study reports that in 2D silica glass under uniaxial tension, the atoms with the largest \(S_i(\epsilon,T)\) coincide with the oxygen atoms that break first in large-scale molecular dynamics at strain rates down to \(10^4\,{\rm s}^{-1}\) [2507.13960]. Pure GPP captures thermal expansion and zero-\(T\) stress-strain stiffening but misses temperature softening of yield, whereas Metropolis-GPP reproduces the drop in yield stress with temperature and the shift of the first bond-break to lower \(\sigma/\epsilon\), in quantitative agreement with MD to within \(\pm 5\%\) [2507.13960].

## 6. Unified picture, scaling laws, and recurrent issues

Across these formulations, several recurrent organizing principles appear. First, non-monotonic elastic interactions deplete near-threshold excitations and generate a singular distribution \(P(x)\sim x^\theta\), with \(\theta\) depending on the spatial dimension \(d\) and the decay exponent \(\alpha\) of the kernel \(\mathcal G(\mathbf r)\) [1307.1646]. For any loading protocol, the far-field elastic propagator decays as \(1/r^\alpha\) with angular modulation, and the same argument yields the stability bound above; as \(\alpha\) increases, \(\theta\) decreases toward zero, recovering the depinning-like case \(\theta=0\) for strictly monotonic kernels [1307.1646].

Second, GSTZ descriptions consistently elevate hidden state variables beyond stress alone. In the multi-species theory, the key state variable is the effective temperature \(\chi\), which controls Boltzmann weights \(e^{-\Delta_i/\chi}\) and therefore the density of active defect species [1203.1965]. In the thermal undeformed-glass theory, the local state resides in a multidimensional space of \((e,\epsilon_{\rm vol})\), with softness \(S\) correlating with low curvature in the local free-energy landscape and enhanced thermal activation [2509.06319]. In the GPP framework, the covariance tensor \(\Sigma_i(\epsilon,T)\) encodes both magnitude and orientation of atomic fluctuations, and its principal axes identify directions of easiest motion [2507.13960].

Third, the theory connects local activation to collective avalanche statistics. In the mean-field stability treatment, the avalanche size distribution follows \(\rho(S)\sim S^{-\tau}\) with \(\tau\approx 3/2\), only weakly modified in \(d=2,3\), reflecting a partly mean-field character of the long-range Eshelby kernel [1307.1646]. In mean-field rheology, one finds the Herschel-Bulkley scaling
\[
\sigma-\sigma_c \sim \dot\gamma^{1/2}\qquad (\tau_c>0),
\]
so that \(\beta=2\), while real elastoplastic models have \(\beta>1\), with the precise value depending on microscopic dynamics [1307.1646]. In undeformed thermal glasses, avalanches leave behind softer, lower-barrier landscapes for subsequent events [2509.06319].

A central point of possible confusion is terminological rather than physical. “Generalized Strain Transformation Zones” refers, in different works, to barrier-distributed STZ species [1203.1965], to coupled shear-and-volume transformation zones in thermal amorphous solids [2509.06319], and to finite-temperature anisotropic predictors of rearrangement zones based on covariance tensors [2507.13960]. This suggests that the common content of GSTZ is the systematic relaxation of one or more assumptions of classical STZ theory: single barrier, purely deviatoric strain, zero temperature, or monotonic effective interactions. Under that reading, the various GSTZ frameworks are complementary rather than mutually exclusive.

Source: https://www.emergentmind.com/topics/generalized-strain-transformation-zones-gstz