import numpy as np
from matplotlib import pyplot
import rft1d

eps           = np.finfo(float).eps

def here_anova1(Y, X, X0, Xi, X0i, df):
	Y         = np.matrix(Y)
	### estimate parameters:
	b         = Xi*Y
	eij       = Y - X*b
	R         = eij.T*eij
	### reduced design:
	b0        = X0i*Y
	eij0      = Y - X0*b0
	R0        = eij0.T*eij0
	### compute F statistic:
	F         = ((np.diag(R0)-np.diag(R))/df[0]) / (np.diag(R+eps)/df[1])
	return F

def here_design_matrices(nResponses, nGroups):
	nTotal    = sum(nResponses)
	X         = np.zeros((nTotal,nGroups))
	i0        = 0
	for i,n in enumerate(nResponses):
		X[i0:i0+n,i] = 1
		i0   += n
	X         = np.matrix(X)
	X0        = np.matrix(np.ones(nTotal)).T  #reduced design matrix
	Xi,X0i    = np.linalg.pinv(X), np.linalg.pinv(X0) #pseudo-inverses
	return X,X0,Xi,X0i



#(0) Set parameters:
np.random.seed(123456789)
nResponses  = 6,8,9   #number of responses in each group
nNodes      = 101
FWHM        = 12.0
nIterations = 5000
### derived parameters:
nGroups     = len(nResponses)
nTotal      = sum(nResponses)
df          = nGroups-1, nTotal-nGroups
X,X0,Xi,X0i = here_design_matrices(nResponses, nGroups)


#(1) Generate Gaussian 1D fields, compute test stat, store field maximum:
F           = []
generator   = rft1d.random.Generator1D(nTotal, nNodes, FWHM)
for i in range(nIterations):
	y       = generator.generate_sample()
	f       = here_anova1(y, X, X0, Xi, X0i, df)
	F.append( f.max() )
F           = np.asarray(F)


#(2) Survival functions:
heights     = np.linspace(6, 14, 21)
sf          = np.array(  [ (F>h).mean()  for h in heights]  )
sfE         = rft1d.f.sf(heights, df, nNodes, FWHM)  #theoretical
sf0D        = rft1d.f.sf0d(heights, df) #theoretical (0D)



#(3) Plot results:
pyplot.close('all')
name        = 'Times New Roman'
ax          = pyplot.axes()
ax.plot(heights, sf,   'o',  label='Simulated')
ax.plot(heights, sfE,  '-',  label='Theoretical')
ax.plot(heights, sf0D, 'r-', label='Theoretical (0D)')
ax.set_xlabel(r'$u$', size=20, usetex=True)
ax.set_ylabel(r'$P (F_\mathrm{max} > u)$', size=20, usetex=True)
leg = ax.legend()
pyplot.setp( leg.get_texts(), name=name)
ax.set_title('ANOVA validation (1D)', size=20, name=name)
# pyplot.show()