Replica symmetry breaking and activated hopping in the constrained Boltzmann-Gibbs effective-potential theory of the glass transition for three-dimensional simple liquids

Cardenas, Franz and Parisi (J. Chem. Phys. 110, 1726 (1999)) computed the Franz-Parisi effective potential of the three-dimensional hard-sphere liquid - the free energy of a 'real' replica constrained to a fixed overlap with a quenched reference configuration - within the replicated hypernetted chain (HNC) approximation at the replica-symmetric level. They found a dynamical critical density $\rho_{cr} \approx 1.14$ (at coupling $\varepsilon_{cr} \approx 0.305$), a first-order line of coexistence between low- and high-overlap phases, a static glass transition at $\rho_c = 1.17$, and a Kauzmann density $\rho_s \approx 1.203$ above which the replica-symmetric solution yields an unphysical negative configurational entropy. The work explicitly leaves open two issues: (i) implementing replica symmetry breaking (RSB) within the effective-potential HNC construction, needed both for the intermediate-overlap region and for the high-density regime $\rho > \rho_s$; and (ii) the microscopic mechanism and rates of the thermally activated hopping that restores ergodicity in finite dimensions, and how it smooths the sharp mean-field features at $\rho_{cr}$ and $\rho_c$. The RSB effective-potential construction has not been completed within three-dimensional liquid-state integral equations (replicated HNC is documented to fail quantitatively in the glass phase; the coherent RSB thermodynamics of hard-sphere glasses was instead achieved via the small-cage expansion and the exact infinite-dimensional solution), and no quantitative theory of hopping corrections to these specific features exists. Completing the RSB and finite-dimensional program remains an open problem.

Background

When a liquid is cooled (or, for hard spheres, compressed) fast enough to avoid crystallization, its structural relaxation time grows by many orders of magnitude and eventually exceeds the experimental or simulation observation window; the material then behaves as a solid on that window, and one speaks of the glass transition at an operational temperature $T_g$ (or density). This operational transition depends on the cooling rate or observation time, and whether it is the finite-dimensional echo of a genuine underlying thermodynamic transition is a central open question of condensed-matter physics. A key diagnostic is the configurational entropy $S_c$: the logarithm of the number of distinct amorphous packings (metastable basins) available to the system at a given state point. Equilibrium extrapolations suggest that $S_c$ vanishes at a finite 'Kauzmann' temperature or density, which is one of the main facts motivating thermodynamic theories of the glass transition.

In mean-field models such as the $p$-spin glass (a solvable model of spins coupled through random $p$-body interactions) the glass transition has a two-step structure. A dynamical transition (of the kind described by mode-coupling theory, MCT, a microscopic theory that predicts an ideal ergodicity-breaking transition from the static liquid structure) occurs where the equilibrium measure splits into an exponentially large number of long-lived metastable basins and ergodicity is broken. At a lower temperature a thermodynamic ('Kauzmann' or 'ideal') transition occurs where the complexity - the configurational entropy density $\Sigma = S_c/N$ - vanishes, and the correct equilibrium description requires one-step replica symmetry breaking (RSB), the spin-glass technique in which the Gibbs measure is decomposed into pure states with a Parisi breaking parameter fixing the state weight. Random first-order transition (RFOT) theory proposes that structural glasses in finite dimensions follow this scenario, with basins replaced by a mosaic of domains whose finite coherence length allows activated rearrangements to restore ergodicity.

Franz and Parisi (1997) turned this scenario into a concrete, in-principle measurable order-parameter function for a structural glass. Fix a reference ('quenched') equilibrium configuration $y$ of the liquid and constrain a second ('real') configuration $x$, drawn with Boltzmann weight, to have a fixed overlap $q(x,y)$ with $y$; the overlap is defined through a kernel $w$ as $q(x,y) = N^{-1}\sum_{ij} w(|x_i - y_j|)$, which counts particles of one configuration lying close to particles of the other. The free energy of the constrained system as a function of $q$, averaged over the reference, is the effective potential $V(q)$: the analog of the Landau free energy of a magnet, but for amorphous order. If $V(q)$ develops a secondary minimum at high overlap in addition to the liquid minimum at low $q$, amorphous configurations acquire thermodynamic significance, and the difference in free-energy density between the minima equals $T\Sigma$, giving direct access to the complexity. Computationally the constrained measure is evaluated with replicas: one couples $r+1$ copies of the system - one privileged quenched replica and $r$ 'floated' replicas, taking $r \to 0$ - and computes the free energy of the mixture.

For simple liquids the pair structure can be computed without simulation from the Ornstein-Zernike (OZ) relation between the total correlation function $h(r) = g(r)-1$ (where $g(r)$ is the pair distribution function) and the direct correlation function $c(r)$, supplemented by a closure expressing $c(r)$ in terms of the pair potential. The hypernetted chain (HNC) closure, $g(r) = \exp[-\beta v(r) + h(r) - c(r)]$, sums the chain diagrams and neglects the bridge diagrams; its virtue for the present problem, shown by a virial expansion of the replicated free-energy functional in the source paper's appendix, is that unlike the Percus-Yevick or mean-spherical approximations it embeds the freezing transition rather than suppressing it - a prerequisite for describing glassy order. Applying HNC to the replicated mixture yields the replicated HNC equations: coupled OZ equations for the quenched-liquid correlations $g_{00}$, the cross correlations $g_{10}$, and the floated-replica correlations $g_{11}$, $g_{12}$, with the overlap coupling entering through the kernel $w$.

Cardenas, Franz and Parisi (1999), extending a 1998 letter, solved these replicated HNC equations for the three-dimensional hard-sphere fluid - hard spheres of unit diameter, so that the number density $\rho$ is the only control parameter - with the overlap kernel $w(r) = 1$ for $r < 0.3$ and $0$ beyond, under a replica-symmetric (RS) ansatz for the mixture (the pair functions above, independent of the floated-replica index). They mapped $V(q)$ and the overlap response $q(\varepsilon)$ as functions of the coupling strength $\varepsilon$ multiplying $w$ and of $\rho$, and compared with Monte Carlo simulations of the same overlap-coupled system.

The resulting phase diagram is the mean-field glass scenario realized inside a realistic liquid-state theory. At low density $q(\varepsilon)$ rises smoothly with the coupling. For $\rho \ge \rho_{cr} \approx 1.14$ a high-overlap phase can be induced by the coupling: the point $(\rho_{cr}, \varepsilon_{cr} \approx 0.305)$ is a second-order critical point where the low-overlap minimum of $V$ destabilizes, and for larger $\rho$ a first-order line appears along which the two minima of $V(q)$ coexist. At $\rho_c = 1.17$ the high-overlap phase persists even at vanishing coupling: $q(\varepsilon)$ jumps discontinuously, marking the static glass transition, while the dynamical (ergodicity-breaking) density is identified with $\rho_{cr}$, consistently with MCT. The complexity $\Sigma(\rho)$ extracted from $\Delta V = T\Sigma$ decreases with density and vanishes at a Kauzmann density $\rho_s \approx 1.203$, strikingly close to the random-close-packing density of hard spheres. The Monte Carlo simulations ($N = 256$-$1024$, with thermalization deliberately kept short to avoid crystallization) qualitatively confirm the picture, including a first-order-like jump in $q(\varepsilon)$.

The paper itself states two limitations that define the present problem. (i) Above $\rho_s$ the RS solution yields a negative configurational entropy - 'clearly unacceptable' - and the authors identify the remedy: the quenched replica must be described by the replica formulation of Mézard and Parisi with replica symmetry breaking; they also warn that RSB has to be expected for intermediate $q$, between the two RS minima where the RS ansatz is unstable. (ii) In finite dimensions, metastable states have finite lifetimes: thermally activated hopping over barriers should restore ergodicity between $\rho_c$ and $\rho_s$, smooth the sharp mean-field features into crossovers, and restore the convexity of $V(q)$ in $q$ - but 'at present the physics of these dynamical problems is unclear'.

Later work reshaped the landscape around rather than through this construction. The replicated HNC closure is documented to be quantitatively unreliable in the glass phase (neglected bridge diagrams grow as the inter-replica correlation peaks sharpen), and coherent RSB thermodynamics of hard-sphere glasses was instead achieved via the small-cage expansion of Mézard and Parisi, systematized in the Parisi-Zamponi mean-field theory of hard-sphere glasses and jamming, and exactly in the limit of spatial dimension $d \to \infty$, where state following has been solved with full RSB. In three dimensions the integral-equation line continued at the replica-symmetric level only; related replica field theories study critical fluctuations around that same replica-symmetric saddle, and dynamical extensions of the overlap formalism (dynamical OZ equations with HNC closure) produce mean-field MCT-like equations rather than activated hopping. The effective-potential RSB construction inside a three-dimensional liquid-state closure and a quantitative theory of hopping corrections to the features at $\rho_{cr}$ and $\rho_c$ remain undone; making them precise is the subject of this problem.

Problem Statement

Cardenas, Franz and Parisi (1999) computed the Franz-Parisi effective potential $V(q)$ of the three-dimensional hard-sphere liquid within the replicated hypernetted chain approximation at the replica-symmetric level, obtaining a dynamical critical point at $(\rho_{cr} \approx 1.14, \varepsilon_{cr} \approx 0.305)$ (here $\varepsilon$ denotes the strength of the coupling field multiplying the overlap kernel), a first-order coexistence line of low- and high-overlap phases, a static glass transition at $\rho_c = 1.17$, and a Kauzmann density $\rho_s \approx 1.203$ above which the replica-symmetric solution gives an unphysical negative complexity $\Sigma$ (the configurational-entropy density).

Objective: complete the program this computation initiates - construct and solve the replica-symmetry-broken, finite-dimensional effective-potential theory for simple glass formers, and elucidate the activated dynamics that restores ergodicity beyond mean field.

Requirement 1 (replica symmetry breaking within the effective-potential construction): formulate and solve the effective-potential theory for the three-dimensional hard-sphere system (overlap kernel $w(r) = 1$ for $r < 0.3$, with $r$ measured in units of the hard-sphere diameter) with a replica-symmetry-breaking ansatz for the replicated pair correlations - at minimum one-step RSB, with explicit treatment of the intermediate-overlap region where the source paper warns RSB is also expected - within the replicated HNC closure or a controlled replacement closure whose diagrammatic status and bridge-diagram error are assessed. The solution must cover the high-density regime $\rho > \rho_s \approx 1.203$ with $\Sigma(\rho) \ge 0$ (removing the negative-complexity pathology), and must reduce to the source's replica-symmetric results in the regime where those are valid.

Requirement 2 (activated dynamics): identify the microscopic mechanism of the thermally activated hopping processes that restore ergodicity between $\rho_c$ and $\rho_s$ in finite-dimensional hard-sphere-like systems, derive their rates as functions of density, and determine quantitatively how they smooth or shift the sharp mean-field features at $\rho_{cr}$ and $\rho_c$ - crossovers replacing transitions, and the convexity of $V(q)$ in $q$ restored.

Completion boundary: the source explicitly acknowledges two open limitations of its replica-symmetric effective-potential computation - (i) replica symmetry breaking (the Mézard-Parisi formulation of the quenched replica, plus the intermediate-overlap warning) is needed for the high-density regime $\rho > \rho_s$ and for the intermediate-overlap region, and (ii) the physics of the barrier-jumping processes that restore ergodicity between $\rho_c$ and $\rho_s$ is at present unclear, with the convexity of $V(q)$ restored and the sharp mean-field transitions becoming crossovers - and this problem asks to advance these source-stated open issues. The source's Sec. 4 locates the high-density remedy inside the HNC approach (its introduction also notes that at high density one needs to resort to the Mézard-Parisi approach for a coherent theory), so solving the RSB problem within the exact $d \to \infty$ solution, or via the small-cage expansion, does not by itself complete this problem: what is asked for is the three-dimensional theory (or a controlled bridge from the exact $d \to \infty$ results to finite dimension). The closure used may be the replicated HNC closure or a specified replacement, whose diagrammatic status and bridge-diagram error are assessed and whose replica-symmetric limit reproduces the source's results. Requirement 1 thus asks for a self-consistent replica-symmetry-broken (at minimum one-step) solution covering the high-density regime $\rho > \rho_s$ with $\Sigma(\rho) \ge 0$ and reducing to the source's replica-symmetric results where those are valid, and Requirement 2 asks for a microscopic hopping mechanism together with a rate law with explicit density dependence and a quantitative account of the induced shift or rounding of the sharp features at $\rho_{cr}$ and $\rho_c$.

References

1. M. Cardenas, S. Franz, G. Parisi, 'Constrained Boltzmann-Gibbs measures and effective potential for glasses in hypernetted chain approximation and numerical simulations', J. Chem. Phys. 110, 1726 (1999). DOI: 10.1063/1.478028, https://doi.org/10.1063/1.478028 (arXiv:cond-mat/9801155, https://arxiv.org/abs/cond-mat/9801155)
2. M. Cardenas, S. Franz, G. Parisi, 'Glass transition and effective potential in the hypernetted chain approximation', J. Phys. A: Math. Gen. 31, L163 (1998). DOI: 10.1088/0305-4470/31/9/001, https://doi.org/10.1088/0305-4470/31/9/001 (arXiv:cond-mat/9712099, https://arxiv.org/abs/cond-mat/9712099)
3. S. Franz, G. Parisi, 'Phase diagram of coupled glassy systems: a mean-field study', Phys. Rev. Lett. 79, 2486 (1997). DOI: 10.1103/PhysRevLett.79.2486, https://doi.org/10.1103/PhysRevLett.79.2486
4. M. Mézard, G. Parisi, 'A tentative replica study of the glass transition', J. Phys. A: Math. Gen. 29, 6515 (1996). DOI: 10.1088/0305-4470/29/20/009, https://doi.org/10.1088/0305-4470/29/20/009 (arXiv:cond-mat/9602002, https://arxiv.org/abs/cond-mat/9602002)
5. M. Mézard, G. Parisi, 'A first-principle computation of the thermodynamics of glasses', J. Chem. Phys. 111, 1076 (1999). DOI: 10.1063/1.479193, https://doi.org/10.1063/1.479193 (arXiv:cond-mat/9812180, https://arxiv.org/abs/cond-mat/9812180)
6. A. de Candia, 'Effective potential in glass forming liquids', J. Stat. Mech. L02001 (2005). DOI: 10.1088/1742-5468/2005/02/L02001, https://doi.org/10.1088/1742-5468/2005/02/L02001 (arXiv:cond-mat/0402552, https://arxiv.org/abs/cond-mat/0402552)
7. G. Parisi, F. Zamponi, 'Mean-field theory of hard sphere glasses and jamming', Rev. Mod. Phys. 82, 789 (2010). DOI: 10.1103/RevModPhys.82.789, https://doi.org/10.1103/RevModPhys.82.789
8. L. Berthier, H. Jacquin, F. Zamponi, 'Can the jamming transition be described using equilibrium statistical mechanics?', J. Stat. Mech. P01004 (2011). DOI: 10.1088/1742-5468/2011/01/P01004, https://doi.org/10.1088/1742-5468/2011/01/P01004
9. H. Jacquin, L. Berthier, F. Zamponi, 'Microscopic mean-field theory of the jamming transition', Phys. Rev. Lett. 106, 135702 (2011). DOI: 10.1103/PhysRevLett.106.135702, https://doi.org/10.1103/PhysRevLett.106.135702
10. P. Charbonneau, J. Kurchan, G. Parisi, P. Urbani, F. Zamponi, 'Exact theory of dense amorphous hard spheres in high dimension. III. The full replica symmetry breaking solution', J. Stat. Mech. P10009 (2014). DOI: 10.1088/1742-5468/2014/10/P10009, https://doi.org/10.1088/1742-5468/2014/10/P10009
11. C. Rainone, P. Urbani, 'Following the evolution of glassy states under external perturbations: the full replica symmetry breaking solution', J. Stat. Mech. 053302 (2016). DOI: 10.1088/1742-5468/2016/05/053302, https://doi.org/10.1088/1742-5468/2016/05/053302
12. S. Franz, H. Jacquin, G. Parisi, P. Urbani, F. Zamponi, 'Quantitative field theory of the glass transition', Proc. Natl. Acad. Sci. USA 109, 18725 (2012). DOI: 10.1073/pnas.1216578109, https://doi.org/10.1073/pnas.1216578109 (arXiv:1206.2482, https://arxiv.org/abs/1206.2482)
13. S. Franz, H. Jacquin, G. Parisi, P. Urbani, F. Zamponi, 'Static replica approach to critical correlations in glassy systems', J. Chem. Phys. 138, 12A540 (2013). DOI: 10.1063/1.4776213, https://doi.org/10.1063/1.4776213 (arXiv:1210.7304, https://arxiv.org/abs/1210.7304)
14. S. Franz, G. Parisi, F. Ricci-Tersenghi, P. Urbani, 'Quasi equilibrium construction for the long time limit of glassy dynamics', J. Stat. Mech. P10010 (2015). DOI: 10.1088/1742-5468/2015/10/P10010, https://doi.org/10.1088/1742-5468/2015/10/P10010
15. C. Cammarota, B. Seoane, 'First-principles computation of random-pinning glass transition, glass cooperative length scales, and numerical comparisons', Phys. Rev. B 94, 180201(R) (2016). DOI: 10.1103/PhysRevB.94.180201, https://doi.org/10.1103/PhysRevB.94.180201
16. J.-M. Bomont, G. Pastore, J.-P. Hansen, 'Coexistence of low and high overlap phases in a supercooled liquid: an integral equation investigation', J. Chem. Phys. 146, 114504 (2017). DOI: 10.1063/1.4978499, https://doi.org/10.1063/1.4978499
17. Q. Guiselin, L. Berthier, G. Tarjus, 'Statistical mechanics of coupled supercooled liquids in finite dimensions', SciPost Phys. 12, 091 (2022). DOI: 10.21468/SciPostPhys.12.3.091, https://doi.org/10.21468/SciPostPhys.12.3.091
18. P. Vardhan, S. P. Das, 'Complexity calculation for an amorphous metastable solid', J. Non-Cryst. Solids 597, 121744 (2022). DOI: 10.1016/j.jnoncrysol.2022.121744, https://doi.org/10.1016/j.jnoncrysol.2022.121744
19. H. Frusawa, 'Replica field theory for a generalized Franz-Parisi potential of inhomogeneous glassy systems: new closure and the associated self-consistent equation', Entropy 26, 241 (2024). DOI: 10.3390/e26030241, https://doi.org/10.3390/e26030241
20. H. Frusawa, 'Pole analysis of the inter-replica correlation function in a two-replica system as a binary mixture: mean overlap in the cluster glass phase', Axioms 13, 468 (2024). DOI: 10.3390/axioms13070468, https://doi.org/10.3390/axioms13070468
21. G. Biroli, J.-P. Bouchaud, 'The RFOT theory of glasses: recent progress and open issues', C. R. Physique 24, 9 (2024). DOI: 10.5802/crphys.136, https://doi.org/10.5802/crphys.136