Walker: Integrating the Dirichlet SDE
This example runs Walker to integrate the Dirichlet SDE (see DiffEq/Dirichlet.h) using constant coefficients. For more details on the Dirichlet SDE, see https:/
Control file
This example runs the setup that was used in the paper Bakosi, Ristorcelli, A Stochastic Diffusion Process for the Dirichlet Distribution, Int. J. Stoch. Anal., 2013.
# vim: filetype=sh: # This is a comment # Keywords are case-sensitive title "Dirichlet for the IJSA paper" walker term 140.0 # Max time dt 0.05 # Time step size npar 10000 # Number of particles ttyi 1000 # TTY output interval rngs mkl_mrg32k3a seed 0 end end dirichlet # Select Dirichlet SDE depvar y init zero coeff const ncomp 2 # = K = N-1 b 0.1 1.5 end S 0.625 0.4 end kappa 0.0125 0.3 end rng mkl_mrg32k3a end statistics <Y1> <Y2> <y1y1> <y2y2> <y1y2> end end
Example run on a single CPU
Main/walker -v -c ../../tmp/dir.q
Results
Left – time evolution of the means and the means of the invariant distribution, right – time evolution of the components of the covariance matrix and those of the invariant.
Gnuplot commands to reproduce the above plots:
plot "stat.txt" u 2:3 w l t "<Y1>", "stat.txt" u 2:4 w l t "<Y2>", 0.5 lt 1, 0.2 lt 2 plot "stat.txt" u 2:5 w l t "<y1y1>", "stat.txt" u 2:6 w l t "<y1y2>", "stat.txt" u 2:7 w l t "<y2y2>", 1.0/44.0 lt 1, -1.0/110 lt 2, 4.0/275.0 lt 3