import numpy as np
import matplotlib.pyplot as plt
from Var_analytical import *


rdir='./data/'


clabels=[r'$K_{perm}$',r'$K_{diss}$',r'$K_{average}$',r'$K_{1/3}$','analitycal Gaussian cov']
cases=[r'$Lognormal \ \sigma^{2}_{\log(k)} = 0.1$',r'$Lognormal \ \sigma^{2}_{\log(k)} = 7$', r'$Binary p = 0.2; k+/k- = 10^4$']

scales=np.array([4,8,16,32,64,128])
variances=[0.1,7,13.572859162824695]
x=scales/512.0
lcs=[16,16,8]
va=VarLgauss(16/2.45398,scales,3)

for i in range(3):



	kpost=np.zeros((len(scales),4))
	for scale in range(len(scales)):
		kpost[scale,0]=np.nanvar(np.log(np.load(rdir+str(i)+'/kperm/'+str(scales[scale])+'.npy')))/variances[i]
		kpost[scale,1]=np.nanvar(np.log(np.load(rdir+str(i)+'/KpostProcess/Kd'+str(scales[scale])+'.npy')))/variances[i]
		kpost[scale,2]=np.nanvar(np.log(np.load(rdir+str(i)+'/KpostProcess/Kv'+str(scales[scale])+'.npy')))/variances[i]
		kpost[scale,3]=np.nanvar(np.log(np.load(rdir+str(i)+'/KpostProcess/Kpo'+str(scales[scale])+'.npy')))/variances[i]
	plt.loglog(x,(x**3)*kpost[:,0],label=clabels[0],marker='x')
	plt.loglog(x,(x**3)*kpost[:,1],label=clabels[1],marker='s')
	plt.loglog(x,(x**3)*kpost[:,2],label=clabels[2],marker='^')
	plt.loglog(x,(x**3)*kpost[:,3],label=clabels[3],marker='o')
	if i==0 or i==1:
		plt.loglog(x,(x**3)*va,label=clabels[4],marker='',linestyle='--')

	plt.vlines(lcs[i]/512.0,((x**3)*kpost[:,0]).min(),((x**3)*kpost[:,0]).max(),label=r'$lc = $'+str(lcs[i]))
	plt.xlabel(r'$\lambda / L$')
	plt.ylabel(r'$(\lambda / L)^3 \sigma^{2}_{\log(K_{eff})} / \sigma^{2}_{\log(k)}$')
	plt.legend()
	plt.grid()
	plt.title(cases[i]) 
	plt.tight_layout()
	plt.savefig(rdir+str(i)+'/Kpost_var.png')
	plt.close()