summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--figs/fig_correlation-times.gplot2
-rw-r--r--figs/fig_harmonic-susceptibilities.gplot6
-rw-r--r--monte-carlo.bib19
-rw-r--r--monte-carlo.tex101
4 files changed, 82 insertions, 46 deletions
diff --git a/figs/fig_correlation-times.gplot b/figs/fig_correlation-times.gplot
index 874fa79..88abe42 100644
--- a/figs/fig_correlation-times.gplot
+++ b/figs/fig_correlation-times.gplot
@@ -73,7 +73,7 @@ unset colorbox
set nokey
set logscale xy
-scaledylabel = '"$\\tau\\avg{s_{\\text{\\sc 1c}}}L^{-z}$" offset 2,0'
+scaledylabel = '"$\\tau\\avg{s_{\\text{\\sc 1c}}}L^{-Dz}$" offset 2,0'
# plotting the first row
set format x ""
diff --git a/figs/fig_harmonic-susceptibilities.gplot b/figs/fig_harmonic-susceptibilities.gplot
index 6129723..cd68de5 100644
--- a/figs/fig_harmonic-susceptibilities.gplot
+++ b/figs/fig_harmonic-susceptibilities.gplot
@@ -5,7 +5,7 @@ set output "fig_harmonic-susceptibilities.tex"
data4 = "data/harmonic-susceptibilities/order-4.dat"
data6 = "data/harmonic-susceptibilities/order-6.dat"
-set palette defined (0 'blue', 0.33 'green', 0.66 'yellow', 1 'red')
+set palette defined (0 'purple', 0.3 'blue', 0.6 'green', 1 'red')
stats data6 using "H"
set cbrange[STATS_min:STATS_max]
set logscale xycb
@@ -23,6 +23,8 @@ unset xlabel
set ylabel '$\chi$' offset 3
set format x ''
+set title "fourfold symmetric" offset -9, -3
+
plot data4 using "L":"X":"dX":"H" with yerrorbars pt 0 lw 2 palette,\
data4 using "L":"X":"H" with lines palette
@@ -32,6 +34,8 @@ set cblabel '(Top) $h_4$ or (bottom) $h_6$' offset 0,6.5
set cbtics format '$10^{%T}$' offset 0,2.8
set format x
+set title "sixfold symmetric" offset -9, -3
+
plot data6 using "L":"X":"dX":"H" with yerrorbars pt 0 lw 2 palette,\
data6 using "L":"X":"H" with lines palette
diff --git a/monte-carlo.bib b/monte-carlo.bib
index 6a0447e..f018281 100644
--- a/monte-carlo.bib
+++ b/monte-carlo.bib
@@ -595,7 +595,7 @@ random field Ising model and finally of quantum spin glasses.},
author = {Martin-Mayor, V. and Seoane, B. and Yllanes, D.},
month = aug,
year = {2011},
- keywords = {Barriers, Effective potential, Monte Carlo methods},
+ keywords = {Effective potential, Monte Carlo methods, Barriers},
pages = {554--596},
file = {Martin-Mayor et al. - 2011 - Tethered Monte Carlo Managing Rugged Free-Energy .pdf:/home/pants/.zotero/data/storage/HEICZ4EE/Martin-Mayor et al. - 2011 - Tethered Monte Carlo Managing Rugged Free-Energy .pdf:application/pdf}
}
@@ -670,4 +670,21 @@ random field Ising model and finally of quantum spin glasses.},
urldate = {2018-09-25},
author = {Bierbaum, Matthew K.},
note = {Source: https://github.com/mattbierbaum/ising.js\vphantom{\{}\}}
+}
+
+@article{bortz_new_1975,
+ title = {A new algorithm for {Monte} {Carlo} simulation of {Ising} spin systems},
+ volume = {17},
+ issn = {0021-9991},
+ url = {http://www.sciencedirect.com/science/article/pii/0021999175900601},
+ doi = {10.1016/0021-9991(75)90060-1},
+ abstract = {We describe a new algorithm for Monte Carlo simulation of Ising spin systems and present results of a study comparing the speed of the new technique to that of a standard technique applied to a square lattice of 6400 spins evolving via single spin flips. We find that at temperatures T {\textless} Tc, the critical temperature, the new technique is faster than the standard technique, being ten times faster at T = 0.588 Tc. We expect that the new technique will be especially valuable in Monte Carlo simulation of the time evolution of binary alloy systems. The new algorithm is essentially a reorganization of the standard algorithm. It accounts for the a priori probability of changing spins before, rather than after, choosing the spin or spins to change.},
+ number = {1},
+ urldate = {2018-10-04},
+ journal = {Journal of Computational Physics},
+ author = {Bortz, A. B. and Kalos, M. H. and Lebowitz, J. L.},
+ month = jan,
+ year = {1975},
+ pages = {10--18},
+ file = {ScienceDirect Full Text PDF:/home/pants/.zotero/data/storage/MZB8RWET/Bortz et al. - 1975 - A new algorithm for Monte Carlo simulation of Isin.pdf:application/pdf;ScienceDirect Snapshot:/home/pants/.zotero/data/storage/5LZNTCRU/0021999175900601.html:text/html}
} \ No newline at end of file
diff --git a/monte-carlo.tex b/monte-carlo.tex
index 578ae14..cf45bcd 100644
--- a/monte-carlo.tex
+++ b/monte-carlo.tex
@@ -109,7 +109,7 @@ of the spin--spin coupling in a generic class of spin systems,
than the introduction of a `ghost spin,\!' our representation relies on
introducing a `ghost transformation.\!'
-\section{Generalized Clusters Without a Field}
+\section{Clusters Without a Field}
We will pose the problem in a general way, but several specific examples can
be found in Table~\ref{table:models} for concreteness. Let $G=(V,E)$ be a graph, where the set of vertices $V=\{1,\ldots,N\}$
@@ -201,9 +201,11 @@ in the following way.
\item Pick a random site $m_0$ and add it to the stack.
\item Select a rotation $r\in R_2$ distributed by $f(r\mid m_0,\set s)$.
- Typically $f$ is uniform on $R_2$, but it is sufficient for preserving
+ Often $f$ is taken as uniform on $R_2$, but it is sufficient for preserving
detailed balance that $f$ be any function of the seed site $m_0$ and
- $Z(s,r\cdot s)$ for all $s\in\set s$, which is what we will assume here.
+ $Z(s,r\cdot s)$ for all $s\in\set s$. The flexibility offered by the
+ choice of distribution will be useful in situations where the state space
+ is infinite.
\item While the stack isn't empty,
\begin{enumerate}
\item pop site $m$ from the stack.
@@ -379,7 +381,7 @@ authors in an existing interactive Ising simulator at
\texttt{https://mattbierbaum.github.io/ising.js}
\cite{bierbaum_ising.js_nodate}.
-\subsection{The $\mathrm O(n)$ model}
+\subsection{The XY and other $\mathrm O(n)$ models}
\label{sec:examples:on}
In the $\mathrm O(n)$ model spins are described by vectors on the
@@ -422,23 +424,26 @@ limit of large $L$ the expected square angle between neighbors being
\[
\avg{\theta^2}\simeq\frac{(n-1)T}{D+H/2}.
\]
-Fig.~\ref{fig:generator_times} shows the effect of making such a choice on
-autocorrelation times for a critical \threedee $\mathrm O(2)$ model. At small
-fields both methods perform the same as zero field Wolff. Intermediate field
-values see efficiency gains for both methods. At large field the uniform
-sampling method sees correlation times grow rapidly, while for the sampling
-method described here the correlation time crosses over to a constant. A
-similar behavior holds for the critical $\mathrm O(3)$ model, though in that
-case the constant value the correlation time approaches at large field is
-larger than that at zero field (see Fig.~\ref{fig:correlation_time-collapse}).
-More detailed discussion on correlation times and these numeric experiments
-can be found in section \ref{sec:performance}.
+We take $\sigma=\sqrt{\avg{\theta^2}}/2$. Fig.~\ref{fig:generator_times} shows
+the effect of making such a choice on autocorrelation times for a critical
+\threedee \textsc{xy} ($\mathrm O(2)$) model. At small fields both methods
+perform the same as zero field Wolff. Intermediate field values see
+efficiency gains for both methods. At large field the uniform sampling method
+sees correlation times grow rapidly, while for the sampling method described
+here the correlation time crosses over to a constant. A similar behavior holds
+for the critical $\mathrm O(3)$ model, though in that case the constant value
+the correlation time approaches at large field is larger than its minimum
+value (see Fig.~\ref{fig:correlation_time-collapse}). This behavior isn't
+particularly worrisome, since the very large field regime corresponds to
+correlation lengths smaller than the lattice spacing and is well-described by
+other algorithms. More detailed discussion on correlation times and these
+numeric experiments can be found in section \ref{sec:performance}.
\begin{figure}
\include{fig_generator-times}
\caption{
The scaled autocorrelation time of the energy $\H$ for the Wolff algorithm
- on a $32\times32\times32$ O(2) model at its critical temperature as a
+ on a $32\times32\times32$ \textsc{xy} model at its critical temperature as a
function of applied vector field magnitude $|H|$. Red points correspond to
reflections sampled uniformly, while the green points represent
reflections sampled as described in section \ref{sec:examples:on}.
@@ -446,12 +451,7 @@ can be found in section \ref{sec:performance}.
\label{fig:generator_times}
\end{figure}
-While this approach ameliorates the inefficiency at large field, it is likely
-not the best solution. Here we have set the scale of transformation based on
-the average difference between nearest neighbors, but cluster methods succeed
-because they tend to produce changes on the order of the system's correlation
-length. A more nuanced analysis that samples reflections with producing this
-behavior in mind may perform much better.
+
\subsection{The Potts model}
@@ -530,13 +530,24 @@ one should expect its natural extension in the presence of a field to scale
roughly like $h^{-z\nu/\beta\delta}$ and collapse appropriately as a function
of $hL^{\beta\delta/\nu}$.
-We measured the autocorrelation time of the energy $\H$ for a variety of
+We measured the autocorrelation time $\tau$ of the energy $\H$ for a variety of
models at critical temperature with many system sizes and canonical fields
-(see Table~\ref{table:models} with $h=\beta H$) using standard methods
-\cite{ossola_dynamic_2004}. The resulting scaling behavior, plotted in
+(see Table~\ref{table:models} with $h=\beta H$) using standard methods for
+obtaining the value and uncertainty from timeseries
+\cite{ossola_dynamic_2004}. Since the computational effort expended in each
+step of the algorithm depends linearly on the size of the associated cluster,
+these values are then scaled by the average cluster size per site
+$\avg{s_{\text{\sc 1c}}}/L^D$ to produce something proportional to machine
+time. The resulting scaling behavior, plotted in
Fig.~\ref{fig:correlation_time-collapse}, is indeed consistent with an
extension to finite field of the behavior at zero field, with an eventual
-finite-size crossover to constant autocorrelation time.
+finite-size crossover to constant autocorrelation time. This crossover isn't
+always kind to the efficiency, e.g., in the $\mathrm O(3)$ model, but in the
+large-field regime where the crossover happens the correlation length is on
+the scale of the lattice spacing and better algorithms exist, like
+Bortz--Kalos--Lebowitz for the Ising model \cite{bortz_new_1975}. Also plotted
+are lines proportional to $h^{-z\nu/\beta\delta}$, which match the behavior of
+the correlation times in the intermediate scaling region.
\begin{figure*}
\include{fig_correlation-times}
@@ -581,21 +592,21 @@ to the scaling functions of the magnetization and susceptibility per site by
\begin{aligned}
\avg{s_{\text{\sc 1c}}}
&=L^{D}\avg{M^2}=\beta\avg\chi+L^{D}\avg{M}^2\\
- &=L^{\gamma/\nu}\big[(hL^{\beta\delta/\nu},ht^{-\beta\delta})^{-\gamma/\beta\delta}\beta \mathcal
- Y(hL^{\beta\delta/\nu,ht^{-\beta\delta}})\\
- &\hspace{1em}+(hL^{\beta\delta/\nu},ht^{-\beta\delta})^{2/\delta}\mathcal
+ &=L^{\gamma/\nu}\big[(hL^{\beta\delta/\nu})^{-\gamma/\beta\delta}\beta \mathcal
+ Y(hL^{\beta\delta/\nu},ht^{-\beta\delta})\\
+ &\hspace{1em}+(hL^{\beta\delta/\nu})^{2/\delta}\mathcal
M(hL^{\beta\delta/\nu},ht^{-\beta\delta})\big].
\end{aligned}
\]
-We therefore expect that, for the Ising model, $\avg{s_{\text{\sc 1c}}}$
-should go as $(hL^{\beta\delta})^{2/\delta}$ for large argument. We further
-conjecture that this scaling behavior should hold for other models whose
-critical points correspond with the percolation transition of Wolff clusters.
-This behavior is supported by our numeric work along the critical isotherm for
-various Ising, Potts, and $\mathrm O(n)$ models, shown in
-Fig.~\ref{fig:cluster_scaling}. Fields for the Potts and $\mathrm O(n)$ models
-take the form $B(s)=(h/\beta)\sum_m\cos(2\pi(s-m)/q)$ and
-$B(s)=(h/\beta)[1,0,\ldots,0]s$ respectively. As can be seen, the average
+We therefore expect that, for the Ising model, $\avg{s_{\text{\sc
+1c}}}L^{-\gamma/\nu}$ should go as $(hL^{\beta\delta/\nu})^{2/\delta}$ for
+large argument. We further conjecture that this scaling behavior should hold
+for other models whose critical points correspond with the percolation
+transition of Wolff clusters. This behavior is supported by our numeric work
+along the critical isotherm for various Ising, Potts, and $\mathrm O(n)$
+models, shown in Fig.~\ref{fig:cluster_scaling}. Fields for the Potts and
+$\mathrm O(n)$ models take the form $B(s)=(h/\beta)\sum_m\cos(2\pi(s-m)/q)$
+and $B(s)=(h/\beta)[1,0,\ldots,0]s$ respectively. As can be seen, the average
cluster size collapses for each model according to the scaling hypothesis, and
the large-field behavior likewise scales as we expect from the na\"ive Ising
conjecture.
@@ -614,8 +625,8 @@ conjecture.
\section{Applying Nonlinear Fields to the xy Model}
-This far our numeric work has quantified the performance of existing
-techniques. Here, we apply our general framework in a new way:
+Thus far our numeric work has quantified the performance of existing
+techniques. Briefly, we demonstrate our general framework in a new way:
harmonic perturbations to the low-temperature {\sc xy}, or \twodee O(2),
model. We consider fields of the form $B_n(s)=h_n\cos(n\theta(s))$, where
$\theta$ is the angle made between $s$ and the $x$-axis, say. Corrections of
@@ -634,7 +645,7 @@ sufficiently high temperatures below the Kosterlitz--Thouless point
\begin{figure}
\include{fig_harmonic-susceptibilities}
\caption{Susceptibilities as a function of system size for a \twodee O(2)
- model at $T=0.7$ and with (top) fourfold symmetric and (bottom) eightfold
+ model at $T=0.7$ and with (top) fourfold symmetric and (bottom) sixfold
symmetric perturbing fields. Different field strengths are shown in
different colors.
}
@@ -644,7 +655,7 @@ sufficiently high temperatures below the Kosterlitz--Thouless point
We made a basic investigation of this result using our algorithm. Since we ran
the algorithm at fairly high fields we did not choose reflections though the
origin uniformly. Instead, we choose the planes of reflection first by
-rotating our starting spin by $2\pi m/n$ degrees for $m$ uniformly taken from
+rotating our starting spin by $\pi m/n$ for $m$ uniformly taken from
$1,\ldots,n$ and generating a normal to the plane from that direction as
described in Section \ref{sec:examples:on}. The resulting susceptibilities as
a function of system size are shown for various field strengths in
@@ -652,7 +663,11 @@ Fig.~\ref{fig:harmonic-susceptibilities}. In the fourfold case, for each field
strength there is a system size at which the divergence in the susceptibility
is cut off, while for the sixfold case we measured no such cutoff, even up to
strong fields. This conforms to the expected result, that even in a strong field
-the sixfold perturbations preserve the critical behavior.
+the sixfold perturbations preserve the critical behavior. Previous work has
+used Monte Carlo to investigate similar symmetry-breaking fields and used a hybrid
+cluster--metropolis method \cite{ala-nissila_numerical_1994}. To our
+knowledge, no application of a direct cluster method has been applied to this
+problem before now.
\section{Conclusions}