banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Accelerating atomic-level protein simulations by flat-histogram techniques
Rent this article for
View: Figures


Image of FIG. 1.
FIG. 1.

Schematic illustration of the simulated ensemble. (a) The definition of the function γ(E) (see Eq. (1)), which determines the microstate probability distribution, P ν∝1/γ(E ν). (b) The probability distribution of energy, P(E) (see Eq. (2)), along with the canonical energy distribution at , .

Image of FIG. 2.
FIG. 2.

Snapshots from the simulations showing typical low-energy conformations for the system of 8 GIIFNEQ peptides. The structures share an overall β-sandwich topology, but differ in the organization of the strands. The energies are (a) E ≈ −21.6, (b) E ≈ −21.2, and (c) E ≈ −20.2.

Image of FIG. 3.
FIG. 3.

MC time evolution of the energy in typical runs with three different methods for the system of 8 GFIINEQ peptides. The right panel shows histograms of energies visited in the respective runs. Each run required ∼470 core hours on a 2.26 GHz Nehalem processor. For each of the three methods, 16 independent runs of this length were generated. (a) Wang-Landau simulation, (b) simulation of the 1/γ ensemble (see Fig. 1), and (c) canonical-ensemble simulation at .

Image of FIG. 4.
FIG. 4.

Probability distribution of E for the system of 8 GFIINEQ peptides at , as obtained from simulations of the 1/γ (blue) and canonical (red) ensembles, respectively. Dashed lines indicate statistical 1σ errors.

Image of FIG. 5.
FIG. 5.

Run-to-run variation of the heat capacity , calculated at , in simulations of the canonical (red) and 1/γ (blue) ensembles. The single-run estimates, , are normalized by the mean, . The length of the runs is the same with both methods.

Image of FIG. 6.
FIG. 6.

Average error in the estimated density of states against average number of steps required for convergence, in toy model Wang-Landau simulations with the histogram-based (Eq. (4)) and tunneling (Eq. (5)) criteria. Different data points correspond to different α and τ, respectively.

Image of FIG. 7.
FIG. 7.

Mean (+) and standard deviation (⋄) of the number of steps required for convergence, , in toy model simulations with two different criteria for when to change the Wang-Landau parameter f. (a) The histogram criterion, Eq. (4). (b) The tunneling criterion, Eq. (5).


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Accelerating atomic-level protein simulations by flat-histogram techniques