Kukacka, J., Kristoufek, L. (2021), Does parameterization affect the complexity of agent-based models?, Journal of Economic Behavior & Organization, 192, pp. 324-356, DOI.
This codebase is using the R language, and it is initialized as a Git.
It is authored by Jiri Kukacka. Question can be directed to [email protected].
-
simulation_gh07_cd.R
master file #1. It provides R code for an illustrative replication of the results for the Gaunersdorfer and Hommes (2007) model: subsection 5.2.2. Computational setup is trivialized for personal computers. The original setup for a cluster computation to replicate results in the paper is also included. Run this file first. -
heatmaps_gh07_cd.R
master file #2. It produces the presented heat map graphics: Fig. 4, panels (c) and (d) and Fig. B.11, panel (d). It loads the output ofsimulation_gh07_cd.R
. -
src
includes additional source scripts:-
gh07_cd.R
function generating the model output. This part of the code was downloaded in MATLAB language and translated to R from the agentFin 0.1 documentation web page [accessed 2019-10-28] by Prof. Blake LeBaron (Brandeis University, USA). Only inevitable modifications were made to link the model script to the multifractality analysis. Credit: Patrick Herb, Blake LeBaron, Axel Szmulewiez. -
gh07_cd_MF.R
function supporting evaluation of independent for-loop iterations in parallel to estimate multifractality measures. -
MF_DFA.R
set of functions needed for estimation of the multifractal spectrum, finally returning$\Delta H$ and$\Delta \alpha$ of estimated generalized Hurst exponents$H(q)$ for a given range of $q$s based on MF-DFA. -
heatmap.2
support function forheatmaps_gh07_cd.R
, downloaded from https://rdrr.io/cran/gplots/src/R/heatmap.2.R [accessed 2019-06-24].
-
-
replication_of_results_gh07_cd
includes the files for replication of Fig. 4, panels (c) and (d) and Fig. B.11, panel (d):-
gh07_cd.Rda
data set for replication of the results in subsection 5.2.2. It must be copy-pasted to the main root to be loaded byheatmaps_gh07_cd.R
to produce the following graphical outputs (for verification purposes):-
heatmap_gh07_c_mfst.pdf
actual Fig. 4, panel (c). -
heatmap_gh07_d_conf.pdf
actual Fig. 4, panel (d). -
heatmap_gh07_diff.pdf
actual Fig. B.11, panel (d).
-
-
-
multifractal detrended fluctuation analysis follows the setting in subsection 2.2.
-
general Monte Carlo setup follows subsection 3.1.
-
simulation_gh07_cd
further follows the simulation setup in subsection 3.2. and the model benchmark setup in subsection 4.2.2. -
heatmaps_gh07_cd.R
follows the multifractality evaluation logic in subsection 3.3.
- install all by
Pkg.add(["doParallel", "foreach", "iterators", "paralell", "gplots", "gtools"])
doParallel
foreach
iterators
paralell
gplots
gtools
- general simulation setup:
-
runs
number of independent Monte Carlo runs for each experiment. -
burn
number of burn-in periods to discard. -
obs
number of periods desired. -
metrics
actual number of estimates generated ingh07_cd_MF.R
.
-
- parameterization:
-
k_val
benchmark parameterization for$\psi$ , associated with the vertical axis of heatmaps. -
j_val
benchmark parameterization for$\gamma$ , associated with the horizontal axis of heatmaps. -
kk,kkk
vertical position in the grid. -
jj,jjj
horizontal position in the grid.
-
- multifractal spectrum estimation setup:
-
smin
minimum scale to consider for the scaling law. -
smax
maximum scale to consider for the scaling law. -
sstep
step length for scales between smin and smax. -
qmax
maximum moment$q$ for estimation of the multifractal spectrum. -
qmin
minimum moment$q$ for estimation of the multifractal spectrum. -
qstep
step length for moments between qmin and qmax.
-