We consider the effect of perturbing a single bond on ground states of nearest-neighbor Ising spin glasses, with a Gaussian distribution of the coupling constants, across various two- and three-dimensional lattices and regular random graphs. Our results reveal that the ground states are strikingly fragile with respect to such changes. Altering the strength of only a single bond beyond a critical threshold value leads to a new ground state that differs from the original one by a droplet of flipped spins whose boundary and volume diverge with the system size-an effect that is reminiscent of the more familiar phenomenon of disorder chaos. These elementary fractal-boundary zero-energy droplets and their composites feature robust characteristics and provide the lowest-energy macroscopic spin-glass excitations. Remarkably, within numerical accuracy, the size of such droplets conforms to a universal power-law distribution with exponents that depend on the spatial dimension of the system. Furthermore, the critical coupling strengths adhere to a stretched exponential distribution that is predominantly determined by the local coordination number.
We consider the effect of perturbing a single bond on ground states of nearest-neighbor Ising spin glasses, with a Gaussian distribution of the coupling constants, across various two- and three-dimensional lattices and regular random graphs. Our results reveal that the ground states are strikingly fragile with respect to such changes. Altering the strength of only a single bond beyond a critical threshold value leads to a new ground state that differs from the original one by a droplet of flipped spins whose boundary and volume diverge with the system size-an effect that is reminiscent of the more familiar phenomenon of disorder chaos. These elementary fractal-boundary zero-energy droplets and their composites feature robust characteristics and provide the lowest-energy macroscopic spin-glass excitations. Remarkably, within numerical accuracy, the size of such droplets conforms to a universal power-law distribution with exponents that depend on the spatial dimension of the system. Furthermore, the critical coupling strengths adhere to a stretched exponential distribution that is predominantly determined by the local coordination number.
Using mappings to computer-science problems and by applying sophisticated algorithms, one can study numerically many problems much better compared to applying standard approaches like Monte Carlo simulations. Here, using calculations of ground states of suitable perturbed systems, droplets are obtained in two-dimensional +-J spin glasses, which are in the focus of a currently very lifely debate. Since a sophisticated matching algorithm is applied here, exact ground states of large systems up to L^2=256^2 spins can be generated. Furthermore, no equilibration or extrapolation to T=0 is necessary. Three different +-J models are studied here: a) with open boundary conditions, b) with fixed boundary conditions and c) a diluted system where a fraction p=0.125 of all bonds is zero. For large systems, the droplet energy shows for all three models a power-law behavior E_D L^θ'_D with θ'_D<0. This is different from previous studies of domain walls, where a convergence to a constant non-zero value (θ_dw=0) has been found for such models. After correcting for the non-compactness of the droplets, the results are likely to be compatible with θ_D= -0.29 for all three models. This is in accordance with the Gaussian system where θ_D=-0.287(4) (ν=3.5 via ν=-1/θ_D). Nevertheless, the disorder-averaged spin-spin correlation exponent ηis determined here via the probability to have a non-zero-energy droplet, and η~0.22$ is found for all three models, this being in contrast to the behavior of the model with Gaussian interactions, where exactly η=0.
We study large-scale, low-energy excitations in the Ising spin glass with Gaussian interactions in two dimensions at zero temperature, using an optimization algorithm to determine exact ground states. Periodic boundary conditions are applied. Our results for the fractal dimension of the surface, ds, and stiffness exponent, ${\ensuremath{\theta}}^{\ensuremath{'}},$ for ``droplet'' excitations, are in reasonable agreement with estimates from ``domain wall'' calculations, and so support the predictions of the ``droplet theory.'' Restricting our analysis to small lattices, we do not find an effective value of ${\ensuremath{\theta}}^{\ensuremath{'}}$ close to $\ensuremath{-}0.47$ as has been recently proposed. The effects of averaging over droplets of different sizes are studied and are also found to be too small to give ${\ensuremath{\theta}}^{\ensuremath{'}}\ensuremath{\approx}\ensuremath{-}0.47$ for smaller sizes. Larger corrections to finite-size scaling would be needed in three and four dimensions in order for the numerical data used to support the ``TNT'' scenario to be compatible with the droplet theory prediction that the stiffness exponent is the same for droplets and domain walls.
We analyze exact ground-state energies of two-dimensional Ising spin glasses with either Gaussian or bimodal nearest-neighbor interactions for large system sizes and for three types of boundary conditions: free on both axes, periodic on both axes, and free on one axis and periodic on the other. We find accurate values for bulk-, edge-, and corner-site energies. Fits for the system with Gaussian bonds are excellent for all types of boundary conditions over the whole range of system sizes. In particular, the leading behavior for nonfree boundary conditions is governed by the stiffness exponent $\ensuremath{\theta}\ensuremath{\approx}\ensuremath{-}0.282$ describing the scaling of domain-wall and droplet excitations. For the system with a bimodal distribution of bonds the fit is good for free boundary conditions but worse for other geometries, particularly for periodic-free boundary conditions where there appear to be unorthodox corrections to scaling up to large sizes. Finally, by introducing hard bonds we test explicitly for the Gaussian case the relationship between domain walls and the standard scaling behavior.
Domain-wall free energy $\ensuremath{\delta}F$, entropy $\ensuremath{\delta}S$, and the correlation function Ctemp of $\ensuremath{\delta}F$ are measured independently in the four-dimensional $\ifmmode\pm\else\textpm\fi{}J$ Edwards-Anderson (EA) Ising spin glass. The stiffness exponent $\ensuremath{\theta}$, the fractal dimension of domain walls ds, and the chaos exponent $\ensuremath{\zeta}$ are extracted from the finite-size scaling analysis of $\ensuremath{\delta}F$, $\ensuremath{\delta}S$, and Ctemp, respectively, well inside the spin-glass phase. The three exponents are confirmed to satisfy the scaling relation $\ensuremath{\zeta}={d}_{s}/2\ensuremath{-}\ensuremath{\theta}$ derived by the droplet theory within our numerical accuracy. We also study bond chaos induced by random variation of bonds, and find that the bond and temperature perturbations yield the universal chaos effects described by a common scaling function and the chaos exponent. These results strongly support the appropriateness of the droplet theory for the description of chaos effect in the EA Ising spin glasses.
For the long-range one-dimensional Ising spin-glass with random couplings decaying as J ( r ) ∝ r − σ , the scaling of the effective coupling defined as the difference between the free energies corresponding to periodic and antiperiodic boundary conditions J R ( N ) ≡ F (P) ( N ) − F (AP) ( N ) ∼ N θ ( σ ) defines the droplet exponent θ ( σ ). Here we study numerically the instability of the renormalization flow of the effective coupling J R ( N ) with respect to magnetic, disorder and temperature perturbations respectively, in order to extract the corresponding chaos exponents ζ H ( σ ), ζ J ( σ ) and ζ T ( σ ) as a function of σ . Our results for ζ T ( σ ) are interpreted in terms of the entropy exponent θ S ( σ ) ≃ 1/3 which governs the scaling of the entropy difference S (P) ( N ) − S (AP) ( N ) ∼ N θ S ( σ ) . We also study the instability of the ground state configuration with respect to perturbations, as measured by the spin overlap between the unperturbed and the perturbed ground states, in order to extract the corresponding chaos exponents and .
By using problem-dependent mappings to computer-science problems and by applying sophisticated algorithms, one can numerically study some important problems much better compared to applying standard approaches, such as Monte Carlo simulations. Here, by using calculations of ground states of suitable perturbed systems, droplets are obtained in two-dimensional $\ifmmode\pm\else\textpm\fi{}J$ spin glasses, which are in the focus of a currently very lively debate. By using a sophisticated matching algorithm, exact ground states of large systems up to L2=2562 spins can be generated. Furthermore, no equilibration or extrapolation to T=0 is necessary. Three different $\ifmmode\pm\else\textpm\fi{}J$ models are studied here: (a) with free boundary conditions, (b) with fixed boundary conditions, and (c) a diluted system wherein a fraction p=0.125 of all bonds is zero. For large systems, the droplet energy shows for all three models a power-law behavior ${E}_{\text{dp}}\ensuremath{\sim}{L}^{{\ensuremath{\theta}}_{\text{dp}}^{\ensuremath{'}}}$ with ${\ensuremath{\theta}}_{\text{dp}}^{\ensuremath{'}}<0$. This is different from the previous studies of domain walls, wherein a convergence to a constant nonzero value $({\ensuremath{\theta}}_{\text{DW}}=0)$ was found. After correcting for the noncompactness of the droplets, the results are likely to be compatible with ${\ensuremath{\theta}}_{\text{dp}}\ensuremath{\approx}\ensuremath{-}0.29$ for all three models. This is in accordance with the Gaussian system wherein ${\ensuremath{\theta}}_{\text{dp}}=\ensuremath{-}0.287(4)$ ($\ensuremath{\nu}\ensuremath{\approx}3.5$ via $\ensuremath{\nu}=\ensuremath{-}1/{\ensuremath{\theta}}_{\text{dp}}$). Nevertheless, the disorder-averaged spin-spin correlation exponent $\ensuremath{\eta}$ is determined here via the probability to have a nonzero-energy droplet, and $\ensuremath{\eta}\ensuremath{\approx}0.22$ is found for all three models, in contrast to the model with the Gaussian interactions, where $\ensuremath{\eta}=0$ is exact.