***** LANDAU WANG ALGORITHM FOR THE ALPHA-3-STATES MODEL **************
There are two Wang-Landau programs and both present bugs/problems descibed in 'issues'. There is also a python program ('dos_mean.py'), which calculates the density of states using combinatorics.
The first Wang-Landau code is not able to visit all the energy bins of the histogram for a number of agents N > 14. Nevertheless it delivers results that are in agreement with the exact solution (calculated by 'dos_mean.py') for N=14. The comparison can be observed in 'Images'. The second Wang-Landau code output (which works for any N) can also be seen in 'Images' for N=14 and for N=100.
CONTENTS of the 'alpha3states_wl' folder: --------------------------------
-
'Code' (subfolder): It contains two versions of the Wang-Landau algorithm the first one uses an N-dimensional array with the explicit state of each spin, the second one uses a 3 integers tuple with the number of agents in each state. It also contains a third program ('dos_mean.py') in python, which calculates the density of states using combinatorics. 'main.f90' 'module.f90' 'input.dat' (output name: nom, size: N (must be even), random seed: iseed, initial modifier: ln_f, alpha value: alpha, number of MC attempts: MCtot, flatness tolerance: tol, modifier threshold: eps, numerical tolerance: eps2)) 'r1279.o', 'ran2.o', secs.o' (random generator files)
'main_2.f90' 'module_2.f90' 'input_2.dat' (identical to input.dat)
'dos_mean.py'
-
'Images' (subfolder): 'ln_g_N10.png', 'ln_g_N12.png', 'ln_g_N14.png' (Plots of the ln(states density(E)) .vs. E) 'wl_N14.png' (A comparison of the results obtained for N=14 with the three programs) 'wl_N100.png' (A comparison of the results obtained for N=100 with the 'dos_mean.py' and the second Wang-Landau program) 'exe_times.png' (Execution times for the different programs for diffrerent values of N and alpha = 0.2)
-
'Scripts' 'ln_g.gnu' (For the output visualization of the first Wang-Landau code) Change the filename and 'N = 10' for the desired size 'ln_g_N14_a0.2.gnu' and 'ln_g_N100_a0.2.gnu' (For the output visualization of the second Wang-Landau code) Each one is specific for a size N and it selects the appropiate yrange so the figure can be observed.
'dos_ln_g_N14_a0.2.gnu' and 'dos_ln_g_N100_a0.2.gnu' (For the output visualization of the python code)
COMPILING INSTRUCTIONS (from local folder 'Code'): ------------------------------------------------
gfortran -g -fcheck=all -Wall -fbacktrace module.f90 main.f90 r1279.o ran2.o secs.o -o main.exe ./main.exe (execution) (Same for main_2.f90)
The outputs are .csv files with 2 columns. The first one are the energy bins, and the second one correspond to the logarithm of the states density corresponding to the given energy.