Pith. sign in

REVIEW 1 major objections 5 minor 45 references

Hyperspatial replica exchange lifts molecules into extra dimensions, claiming to sample transitions that parallel tempering cannot reach.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 04:48 UTC pith:ZO5KV5WP

load-bearing objection Interesting method, but the alanine-dipeptide benchmark rests on a likely false claim about ff14SB's improper dihedrals; the PT baseline is also underpowered. The double-well test and separable-replica construction are worth taking seriously. the 1 major comments →

arxiv 2607.22417 v1 pith:ZO5KV5WP submitted 2026-07-24 physics.comp-ph cond-mat.softcond-mat.stat-mechphysics.chem-ph

Hyperspatial Sampling: Circumventing Free-Energy Barriers via Replica Exchange with Extra Dimensions

Henrik Christiansen , Matheus Ferraz , Takashi Maruyama , Francesco Alesiani This is my paper
classification physics.comp-ph cond-mat.softcond-mat.stat-mechphysics.chem-ph
keywords hyperspatial replica exchangeenhanced samplingextra spatial dimensionsreplica exchange molecular dynamicsparallel temperingfree-energy barriersalanine dipeptideexplicit solvent
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out to show that a classical simulation can escape free-energy barriers — including steric barriers that would require atoms to overlap in ordinary space — by being carried out in a space with extra spatial dimensions, with a ladder of replicas whose exchange is driven by a geometric progression of a harmonic penalty on those extra coordinates. The central claim, demonstrated on a double-well model and alanine dipeptide in explicit water, is that this hyperspatial replica exchange reaches near-uniform acceptance rates with the same number of replicas that leave parallel tempering stalled, cutting replica round-trip times by about a factor of eight. At the physical replica the method samples both backbone basins and both chiral enantiomers of alanine dipeptide, transitions the authors say standard parallel tempering cannot reach on the same timescale. If the claim holds, the paper offers an unbiased enhanced-sampling tool that needs far fewer replicas for solvated biological systems than temperature replica exchange.

Core claim

The central claim is that adding extra spatial dimensions per atom and coupling a suite of replicas through a harmonic penalty on those coordinates yields a canonical sampler with an unbiased physical replica. The force field is evaluated with distances and angles measured in the full higher-dimensional space, so as the penalty weakens, geometric paths open through the extra dimensions that circumvent barriers including steric overlaps. In the alanine benchmark with 16 replicas, the authors report per-pair acceptance of 17–38%, a mean round-trip time of 324 ps versus 2625 ps for parallel tempering, and visits to both backbone basins and both L and D enantiomeric forms, with the signed tetrah

What carries the argument

The central mechanism is the hyperspatial Hamiltonian, formed by the original force field evaluated with distances and angles measured in a higher-dimensional space plus a harmonic penalty on the extra coordinates. A ladder of replicas uses geometrically spaced penalty strengths, so the exchange acceptance between neighbors depends only on the penalty-energy fluctuations and can be made nearly uniform. A separate separable replica, which decouples the physical Hamiltonian from the penalty, preserves detailed balance and gives an unbiased target distribution. The fast harmonic penalty is integrated exactly within a multiple-time-scale scheme, and restricting the penalty to solute atoms keeps

Load-bearing premise

The load-bearing premise is that the higher-dimensional force field is the true physical force field: every bonded term that encodes chirality (notably improper-dihedral terms) must be faithfully included when distances and angles are measured in higher-dimensional space, and the paper never states how impropers are handled — if they are omitted or miscomputed, the alanine L/D sampling is sampling a modified potential.

What would settle it

Run the same hyperspatial-replica-exchange ladder on alanine dipeptide with improper-dihedral energies computed in the full higher-dimensional space instead of the three-dimensional projection and monitor the signed tetrahedral volume at Cα; if the L/D population ratio no longer approaches 1:1, or per-pair acceptance collapses to parallel-tempering levels, the central ergodicity claim over the true force field is falsified.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • For the benchmark system, 16 hyperspatial-replica-exchange replicas produce roughly uniform acceptance near the 23% optimum, while 16 parallel-tempering replicas stall at 4–5%, yielding about an eightfold reduction in mean round-trip time.
  • The physical replica visits both backbone basins and both enantiomers of alanine dipeptide, a required ergodicity test that the paper says parallel tempering fails on the same timescale.
  • Because the exchange criterion between non-separable replicas depends only on the penalty-energy difference, the geometric penalty schedule yields acceptance that is nearly independent of the absolute penalty scale.
  • Restricting the extra-dimensional penalty to a subset of atoms makes the approach practical for explicitly solvated biomolecules and naturally supports other partitions, such as a ligand exploring extra dimensions while a protein remains confined.
  • The method is orthogonal to temperature exchange and collective-variable biasing, so it can be combined with parallel tempering or metadynamics.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The paper fixes the number of extra dimensions at one per atom; a natural extension is to vary that number and map how replica count and barrier-crossing speed scale, which would indicate whether the approach generalizes to systems with more complex topologies.
  • Because the enhanced replicas populate configurations that are nonphysical in three dimensions, quantitative free-energy estimates would require careful reweighting; the paper's projection onto the physical subspace suggests the method's clearest value may be as an ergodic exploration tool rather than a direct free-energy calculator, a distinction the authors do not draw.
  • A stress test would apply the same geometric penalty ladder to a molecule with two competing steric barriers of very different widths, to see whether the ladder remains near-uniform or needs retuning, since the acceptance heuristic assumes roughly Gaussian penalty-energy fluctuations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 5 minor

Summary. The paper introduces HS-REX, a replica-exchange method in which replicas share the physical Hamiltonian but differ in a harmonic penalty μ on additional spatial coordinates x+ for a subset of atoms; one separable replica at large μ is used to recover the target distribution. The authors derive exchange criteria, propose a geometric μ-ladder based on harmonic equipartition, and test the method on a 1D double-well model and on alanine dipeptide in TIP3P water with 16 replicas. They report near-uniform acceptance, roughly 8× shorter round-trip times than PT, and sampling of backbone isomerization and both chiral forms, and conclude that HS-REX achieves enhanced ergodic sampling over PT at equal replica count.

Significance. If the alanine benchmark is correct, the method is a novel and potentially important enhanced-sampling tool: it is orthogonal to temperature or collective-variable methods, restricts the extra-dimensional penalty to the solute, requires far fewer replicas for explicit solvent than temperature REX, and is supported by an available code base and standard detailed-balance exchange criteria. The double-well demonstration is clean, and Eqs. (5), (8), and (9) are internally consistent. However, the headline claim of unbiased sampling of ff14SB's chiral states depends on an implementation detail—the treatment of improper-dihedral terms—that is neither specified nor correctly characterized. Until that is resolved, the significance of the alanine result is conditional.

major comments (1)
  1. [Section IV, paragraph beginning 'Under a classical, non-reactive force field'; End Matter, 'Molecular dynamics: Alanine-] The alanine-dipeptide result rests on the claim that ff14SB's bonded terms do not distinguish L- and D-form geometries, making the two forms exact degenerates. This is not correct for AMBER ff14SB, which includes improper-dihedral terms for protein residues, including at the Cα chiral center; such terms depend on the sign of the improper angle and penalize one handedness. The End Matter states that distances and angles use the full D-dimensional space and that backbone dihedrals are computed from the three-dimensional projection, but it never states how improper dihedrals are handled. If impropers are computed from the 3D projection, the chiral barrier is not relieved by x+; if they are omitted, the simulation is of a modified force field, not ff14SB, and the L↔D interconversion is a trivial consequence of that modification. Either way, the observed chirality flips at the physical replic
minor comments (5)
  1. [End Matter, 'Molecular dynamics: Alanine-Dipeptide'] For reproducibility, please provide the full list of 16 μ values, the PT temperature list, total production time per replica, number of independent repeats, and the protocol for all dihedral terms (proper and improper) in the hyperspatial extension. The current text specifies only the geometric range μ=2000–0.5 and the PT range 298–600 K.
  2. [Fig. 2(d) and Fig. 3(a) captions] The method section states that the target distribution is obtained from the separable Hamiltonian replica, but the figure captions label the physical replica as 'μ=2000'. Please state explicitly whether the plotted physical replica is the separable replica, to avoid ambiguity about whether the free-energy surfaces are from the target distribution.
  3. [Abstract and Section IV, 'Replica-exchange diagnostics'] The phrase 'near-uniform exchange acceptance rates' is stronger than the data: per-pair acceptance in Fig. 4(a) ranges 17–38%, and the double-well non-separable pairs show 36–45%. Suggest saying 'approximately uniform' or reporting the observed range.
  4. [End Matter, 'Penalty Parameter μ Schedule'] The harmonic-equipartition derivation applies to the separable Hamiltonian; the paper correctly labels it a heuristic for non-separable replicas at small μ. This caveat should also appear in the main text where the geometric ladder is introduced, since the near-uniform acceptance claim relies on it.
  5. [Throughout] Minor grammatical and typographical issues: 'on an random even/odd schedule', 'adjacent temperature states overlap do not overlap sufficiently', and 'Since we, for biological systems, are primarily interested...' should be corrected.

Circularity Check

0 steps flagged

No significant circularity: central derivation is self-contained; the chirality/improper issue is a force-field correctness question, not circularity.

full rationale

The paper's claimed derivation chain is self-contained. The hyperspatial Hamiltonian in Eqs. (2)-(3) is defined directly from the physical Hamiltonian plus a harmonic penalty; the separable replica in Eq. (4) is constructed to preserve detailed balance; the exchange criterion in Eq. (5)/(8) is the standard Metropolis rule; and the geometric μ-ladder in Eq. (12) is motivated by a harmonic-equipartition estimate in the End Matter, which the paper explicitly labels as heuristic for non-separable replicas. No fitted parameter is later renamed as a prediction: the acceptance rates, round-trip-time distributions, and chirality-occupancy observations in Figs. 2-4 are independently measured simulation outputs, not quantities imposed by the choice of μ or by the derivation. The only self-citation, DIMOS [26], is cited as the simulation code base and does not carry the central theoretical claim. The alanine chirality demonstration depends on the factual assertion that ff14SB's bonded terms do not distinguish L and D geometries; whether that is true, and whether the DIMOS extension correctly includes improper dihedrals, is a force-field fidelity question rather than a circular reduction. The paper itself flags the harmonic-ladder derivation as 'heuristics rather than exact results' (End Matter), further confirming that no central result is being passed off as a first-principles derivation from its own inputs.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 1 invented entities

The method's parameter choices (μ endpoints, N, d+) are hand-selected rather than fitted to data. The main domain assumptions are the validity of the D-dimensional force-field extension and the claimed degeneracy of L/D forms. No new physical entities are postulated; the extra dimension is a numerical construct.

free parameters (5)
  • μ_max = 2000
    Largest penalty coefficient; confines the physical-like replica near the 3D subspace. Chosen by hand to give acceptable exchange with the adjacent replica.
  • μ_min = 0.5
    Smallest penalty; chosen so that relevant barriers are crossed in the alanine-dipeptide system. Affects the effective dimensionality explored.
  • Number of replicas N = 16
    Chosen to span μ_max to μ_min and to match the PT setup. The acceptance and round-trip statistics depend directly on this choice.
  • Extra dimension count d+ = 1
    One additional spatial coordinate per solvated solute atom. The paper notes that the influence of d+ is left for future work.
  • Exchange interval = 100 MD steps
    Replica-exchange attempts every 100 steps; an implementation detail that affects acceptance correlations and wall-clock efficiency.
axioms (5)
  • standard math Replica-exchange detailed balance (Eq. 5) is valid for the product of replica Hamiltonians.
    The method relies on the standard replica-exchange acceptance criterion to sample the joint ensemble; this is textbook statistical mechanics.
  • domain assumption The hyperspatial force field, defined by measuring distances and angles in R^D and backbone dihedrals in the 3D projection, is a valid, consistent Hamiltonian that reduces to the physical Hamiltonian at x+=0.
    The D-dimensional extension of ff14SB is described but not independently verified, and the treatment of improper dihedrals is not specified. This is load-bearing for the alanine benchmark.
  • domain assumption In the ff14SB implementation used here, no bonded term distinguishes L- from D-form geometries, so the two chiral forms are exact degenerates.
    Stated in the chirality paragraph; if false, the observed chirality switching is sampling states of different energy and the benchmark's meaning changes.
  • domain assumption A geometric progression in μ yields approximately uniform exchange acceptance.
    Motivated by harmonic equipartition in the End Matter, but the authors explicitly call the small-μ expressions heuristics rather than exact results.
  • domain assumption For the tested systems, continuous paths in D dimensions connect metastable basins at an energy cost accessible near μ_min=0.5.
    This is the core physical premise of the method: that extra dimensions open low-energy routes around barriers. It is demonstrated empirically on two systems, not proven generally.
invented entities (1)
  • Extra spatial coordinate x+ per atom no independent evidence
    purpose: Allows atoms to pass each other geometrically and thereby circumvent steric and free-energy barriers in the original 3D space.
    A computational device, not a claimed physical dimension. No external falsifiable handle is provided, but none is needed because the method is validated by recovering the target ensemble at the separable replica.

pith-pipeline@v1.3.0-alltime-deepseek · 10371 in / 21477 out tokens · 265064 ms · 2026-08-01T04:48:29.166751+00:00 · methodology

0 comments
read the original abstract

Simulating systems with rugged free-energy landscapes remains a central challenge in computational physics and chemistry. We introduce hyperspatial replica exchange (HS-REX), an enhanced sampling method in which the physical system is artificially extended by additional spatial dimensions. In higher dimensions, free-energy barriers can be circumvented through paths that are geometrically inaccessible in the original space. Restricting the penalty to only solute atoms dramatically reduces the number of replicas required for solvated systems compared to standard temperature replica exchange, a feature especially relevant for biological applications. As proof of concept, we demonstrate the method on a double-well model system and on alanine dipeptide in explicit water as benchmark system. In the latter case, HS-REX achieves enhanced conformational sampling of not only the slow backbone dihedral angles, but also both chiral configurations of the molecule, which are sterically inaccessible to standard sampling in three dimensions. This demonstrates enhanced ergodic sampling over conventional temperature replica exchange.

Figures

Figures reproduced from arXiv: 2607.22417 by Francesco Alesiani, Henrik Christiansen, Matheus Ferraz, Takashi Maruyama.

Figure 1
Figure 1. Figure 1: FIG. 1. (a) Target one-dimensional potential [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Backbone free-energy surfaces ∆ [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Signed C [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Replica-exchange diagnostics for alanine dipeptide [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references

  1. [1]

    Janke,Rugged Free Energy Landscapes: Common Computational Approaches to Spin Glasses, Structural Glasses and Biological Macromolecules, Lecture Notes in Physics, Vol

    W. Janke,Rugged Free Energy Landscapes: Common Computational Approaches to Spin Glasses, Structural Glasses and Biological Macromolecules, Lecture Notes in Physics, Vol. 736 (Springer, Berlin, 2007)

    2007

  2. [2]

    Okamoto, Generalized-ensemble algorithms: en- hanced sampling techniques for Monte Carlo and molecu- lar dynamics simulations, J

    Y. Okamoto, Generalized-ensemble algorithms: en- hanced sampling techniques for Monte Carlo and molecu- lar dynamics simulations, J. Mol. Graph. Model.22, 425 (2004)

    2004

  3. [3]

    Mitsutake, Y

    A. Mitsutake, Y. Mori, and Y. Okamoto, Enhanced sam- pling algorithms, Biomol. Simul. , 153 (2012)

    2012

  4. [4]

    G. M. Torrie and J. P. Valleau, Nonphysical sampling distributions in Monte Carlo free-energy estimation: Um- brella sampling, J. Comput. Phys.23, 187 (1977)

    1977

  5. [5]

    B. A. Berg and T. Neuhaus, Multicanonical ensemble: A new approach to simulate first-order phase transitions, Phys. Rev. Lett.68, 9 (1992)

    1992

  6. [6]

    Wang and D

    F. Wang and D. P. Landau, Efficient, multiple-range ran- dom walk algorithm to calculate the density of states, Phy. Rev. Lett.86, 2050 (2001)

    2050

  7. [7]

    Laio and M

    A. Laio and M. Parrinello, Escaping free-energy minima, Proc. Nati. Acad. Sci.99, 12562 (2002)

    2002

  8. [8]

    Doucet, N

    A. Doucet, N. De Freitas, and N. Gordon, An introduc- tion to sequential monte carlo methods, inSequential Monte Carlo methods in practice(Springer, 2001) pp. 3–14

    2001

  9. [9]

    Hukushima and Y

    K. Hukushima and Y. Iba, Population annealing and its application to a spin glass, AIP Conf. Proc.690, 200 (2003)

    2003

  10. [10]

    Machta, Population annealing with weighted averages: A monte carlo method for rough free-energy landscapes, Phys

    J. Machta, Population annealing with weighted averages: A monte carlo method for rough free-energy landscapes, Phys. Rev. E82, 026704 (2010)

    2010

  11. [11]

    Christiansen, M

    H. Christiansen, M. Weigel, and W. Janke, Accelerating molecular dynamics simulations with population anneal- ing, Phys. Rev. Lett.122, 060602 (2019)

    2019

  12. [12]

    R. H. Swendsen and J.-S. Wang, Replica Monte Carlo simulation of spin-glasses, Phys. Rev. Lett.57, 2607 (1986)

    1986

  13. [13]

    C. J. Geyer, Markov chain Monte Carlo maximum likeli- hood, inComputing Science and Statistics: Proceedings of the 23rd Symposium on the Interface(Interface Foun- dation of North America, 1991) pp. 156–163

    1991

  14. [14]

    Hukushima and K

    K. Hukushima and K. Nemoto, Exchange Monte Carlo method and application to spin glass simulations, J. Phys. Soc. Jpn.65, 1604 (1996)

    1996

  15. [15]

    Sugita and Y

    Y. Sugita and Y. Okamoto, Replica-exchange molecular dynamics method for protein folding, Chem. Phys. Lett. 314, 141 (1999)

    1999

  16. [16]

    D. J. Earl and M. W. Deem, Parallel tempering: Theory, applications, and new perspectives, Phys. Chem. Chem. Phys.7, 3910 (2005)

    2005

  17. [17]

    U. H. E. Hansmann, Parallel tempering algorithm for conformational studies of biological molecules, Chem. Phys. Lett.281, 140 (1997)

    1997

  18. [18]

    Fukunishi, O

    H. Fukunishi, O. Watanabe, and S. Takada, On the Hamiltonian replica exchange method for efficient sam- 6 pling of biomolecular systems: Application to protein structure prediction, J. Chem. Phys.116, 9058 (2002)

    2002

  19. [19]

    Sugita, A

    Y. Sugita, A. Kitao, and Y. Okamoto, Multidimensional replica-exchange method for free-energy calculations, J. Chem. Phys.113, 6042 (2000)

    2000

  20. [20]

    C. J. Pickard, Hyperspatial optimization of structures, Phys. Rev. B99, 054102 (2019)

    2019

  21. [21]

    D. A. Kofke, On the acceptance probability of replica- exchange Monte Carlo trials, J. Chem. Phys.117, 6911 (2002)

    2002

  22. [22]

    Kone and D

    A. Kone and D. A. Kofke, Selection of temperature inter- vals for parallel-tempering simulations, J. Chem. Phys. 122, 206101 (2005)

    2005

  23. [23]

    Barducci, G

    A. Barducci, G. Bussi, and M. Parrinello, Well-tempered metadynamics: a smoothly converging and tunable free- energy method, Phys. Rev. Lett.100, 020603 (2008)

    2008

  24. [24]

    M. P. Allen and D. J. Tildesley,Computer Simulation of Liquids, 2nd ed. (Oxford University Press, Oxford, 2017)

    2017

  25. [25]

    Leimkuhler and C

    B. Leimkuhler and C. Matthews, Efficient molecular dynamics using geodesic integration and solvent–solute splitting, Proc. R. Soc. A472, 20160138 (2016)

    2016

  26. [26]

    Christiansen, T

    H. Christiansen, T. Maruyama, F. Errica, V. Zaverkin, M. Takamoto, and F. Alesiani, Fast, modular, and differentiable framework for machine learning-enhanced molecular simulations, J. Chem. Phys.163, 182501 (2025)

    2025

  27. [27]

    Code available at github.com/nec-research/DIMOS

  28. [28]

    J. A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, K. E. Hauser, and C. Simmerling, ff14SB: Improving the accuracy of protein side chain and backbone parameters from ff99SB, J. Chem. Theory Comput.11, 3696 (2015)

    2015

  29. [29]

    M. E. Tuckerman,Statistical Mechanics: Theory and Molecular Simulation(Oxford University Press, Oxford, 2010)

    2010

  30. [30]

    M. E. Tuckerman, B. J. Berne, and G. J. Martyna, Reversible multiple time scale molecular dynamics, J. Chem. Phys.97, 1990 (1992)

    1990

  31. [31]

    Nymeyer, How efficient is replica exchange molecu- lar dynamics? an analytic approach, J

    H. Nymeyer, How efficient is replica exchange molecu- lar dynamics? an analytic approach, J. Chem. Theory Comput.4, 626 (2008)

    2008

  32. [32]

    P. Liu, B. Kim, R. A. Friesner, and B. J. Berne, Replica exchange with solute tempering: A method for sampling biological systems in explicit water, Proc. Nat. Acad. Sci. 102, 13749 (2005)

    2005

  33. [33]

    L. Wang, R. A. Friesner, and B. Berne, Replica exchange with solute scaling: a more efficient version of replica exchange with solute tempering (rest2), J. Phys. Chem. B115, 9431 (2011)

    2011

  34. [34]

    Chipot and A

    C. Chipot and A. Pohorille, eds.,Free Energy Calcula- tions: Theory and Applications in Chemistry and Biology (Springer, Berlin, 2007)

    2007

  35. [35]

    P´ eriole and A

    X. P´ eriole and A. E. Mark, Convergence and sampling ef- ficiency in replica exchange simulations of peptide folding in explicit solvent, J. Chem. Phys.126, 014903 (2007)

    2007

  36. [36]

    Nadler and U

    W. Nadler and U. H. E. Hansmann, Optimized explicit- solvent replica exchange molecular dynamics from scratch, J. Phys. Chem. B112, 10386 (2008)

    2008

  37. [37]

    Rosta and G

    E. Rosta and G. Hummer, Error and efficiency of replica exchange molecular dynamics simulations, J. Chem. Phys.131, 165102 (2009)

    2009

  38. [38]

    Lingenheil, R

    M. Lingenheil, R. Denschlag, G. Mathias, and P. Ta- van, Efficiency of exchange schemes in replica exchange, Chem. Phys. Lett.478, 80 (2009)

    2009

  39. [39]

    Barducci, M

    A. Barducci, M. Bonomi, and M. Parrinello, Metady- namics, Wiley Interdiscip. Rev. Comput. Mol. Sci.1, 826 (2011)

    2011

  40. [40]

    D. A. Case, H. M. Aktulga, K. Belfon, D. S. Cerutti, G. A. Cisneros, V. W. D. Cruzeiro, N. Forouzesh, T. J. Giese, A. W. Gotz, H. Gohlke,et al., Ambertools, Journal of chemical information and modeling63, 6183 (2023)

    2023

  41. [41]

    W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and M. L. Klein, Comparison of simple potential functions for simulating liquid water, J. Chem. Phys.79, 926 (1983)

    1983

  42. [42]

    M. J. Abraham, T. Murtola, R. Schulz, S. P´ all, J. C. Smith, B. Hess, and E. Lindahl, Gromacs: High perfor- mance molecular simulations through multi-level paral- lelism from laptops to supercomputers, SoftwareX1, 19 (2015)

    2015

  43. [43]

    I. G. Tironi, R. Sperb, P. E. Smith, and W. F. van Gun- steren, A generalized reaction field method for molecular dynamics simulations, J. Chem. Phys.102, 5451 (1995)

    1995

  44. [44]

    M. E. Tuckerman, G. J. Martyna, and B. J. Berne, Molec- ular dynamics algorithm for condensed systems with mul- tiple time scales, J. Chem. Phys.93, 1287 (1990)

    1990

  45. [45]

    S. J. Stuart, R. Zhou, and B. J. Berne, Molecular dynam- ics with multiple time scales: The selection of efficient reference system propagators, J. Chem. Phys.105, 1426 (1996). Replica Exchange Criteria The probability of accepting an exchange between replicasiandjfollows from the usual replica-exchange criterion [15, 18] presented in Eq. 5. For exchanges...

    1996