def run_demo():
LG.basicConfig(level=LG.INFO)
#1. simulate data from a linear PCA model
N = 25
K = 5
D = 200
SP.random.seed(1)
S = SP.random.randn(N,K)
W = SP.random.randn(D,K)
Y = SP.dot(W,S.T).T
Y+= 0.5*SP.random.randn(N,D)
#use "standard PCA"
[Spca,Wpca] = gplvm.PCA(Y,K)
#reconstruction
Y_ = SP.dot(Spca,Wpca.T)
if 1:
#use linear kernel
covariance = linear.LinearCFISO(n_dimensions=K)
hyperparams = {'covar': SP.log([1.2])}
if 0:
#use ARD kernel
covariance = se.SqexpCFARD(n_dimensions=K)
hyperparams = {'covar': SP.log([1]*(K+1))}
#initialization of X at arandom
X0 = SP.random.randn(N,K)
X0 = Spca
hyperparams['x'] = X0
#standard Gaussian noise
likelihood = lik.GaussLikISO()
hyperparams['lik'] = SP.log([0.1])
g = gplvm.GPLVM(covar_func=covariance,likelihood=likelihood,x=X0,y=Y,gplvm_dimensions=SP.arange(X0.shape[1]))
#specify optimization bounds:
bounds = {}
bounds['lik'] = SP.array([[-5.,5.]]*D)
hyperparams['x'] = X0
print "running standard gplvm"
[opt_hyperparams,opt_lml2] = opt.opt_hyper(g,hyperparams,gradcheck=False)
print "optimized latent X:"
print opt_hyperparams['x']
K = 3
D = 10
S = SP.random.randn(N, K)
W = SP.random.randn(D, K)
Y = SP.dot(W, S.T).T
Y += 0.5 * SP.random.randn(N, D)
[Spca, Wpca] = PCA(Y, K)
#reconstruction
Y_ = SP.dot(Spca, Wpca.T)
#construct GPLVM model
linear_cf = linear.LinearCFISO(n_dimensions=K)
noise_cf = noise.NoiseCFISO()
mu_cf = fixed.FixedCF(SP.ones([N, N]))
covariance = combinators.SumCF((mu_cf, linear_cf, noise_cf))
# covariance = combinators.SumCF((linear_cf, noise_cf))
#no inputs here (later SNPs)
X = Spca.copy()
#X = SP.random.randn(N,K)
gplvm = GPLVM(covar_func=covariance, x=X, y=Y)
gpr = GP(covar_func=covariance, x=X, y=Y[:, 0])
#construct hyperparams
covar = SP.log([0.1, 1.0, 0.1])
ard = True
if ard:
covar_r = SP.zeros([Kr]) + 0.2 * SP.random.randn(Kr)
covar_c = SP.zeros([Kc]) + 0.2 * SP.random.randn(Kc)
lik = SP.log([0.1])
else:
covar_r = SP.log([0.5])
covar_c = SP.log([0.3])
lik = SP.log([0.1])
if ard:
covariance_c_ = linear.LinearCF(n_dimensions=Kc)
covariance_r_ = linear.LinearCF(n_dimensions=Kr)
else:
covariance_c_ = linear.LinearCFISO(n_dimensions=Kc)
covariance_r_ = linear.LinearCFISO(n_dimensions=Kr)
likelihood_ = LIK.GaussLikISO()
hyperparams = {}
hyperparams['covar_r'] = covar_r
hyperparams['covar_c'] = covar_c
hyperparams['lik'] = lik
hyperparams['x_r'] = X0r
hyperparams['x_c'] = X0c
kgp = kronecker_gplvm.KroneckerGPLVM(covar_func_r=covariance_r_,
covar_func_c=covariance_c_,
likelihood=likelihood_)
kgp.setData(x_r=X0r, x_c=X0c, y=Y)
sim_fa_noise = False
if sim_fa_noise:
#inerpolate noise levels
noise_levels = SP.linspace(0.1, 1.0, Y.shape[1])
Ynoise = noise_levels * random.randn(N, D)
Y += Ynoise
else:
Y += 0.1 * SP.random.randn(N, D)
#use "standard PCA"
[Spca, Wpca] = gplvm.PCA(Y, K)
#reconstruction
Y_ = SP.dot(Spca, Wpca.T)
covariance = linear.LinearCFISO(n_dimensions=K)
hyperparams = {'covar': SP.log([1.2])}
hyperparams_fa = {'covar': SP.log([1.2])}
#factor analysis noise
likelihood_fa = lik.GaussLikARD(n_dimensions=D)
hyperparams_fa['lik'] = SP.log(0.1 * SP.ones(Y.shape[1]))
#standard Gaussian noise
likelihood = lik.GaussLikISO()
hyperparams['lik'] = SP.log([0.1])
#initialization of X at arandom
X0 = SP.random.randn(N, K)
X0 = Spca
hyperparams['x'] = X0
print limix.__file__
import pygp.covar.linear as lin
import pygp.covar.se as se
import pygp.covar.gradcheck as GC
import pygp.covar.combinators as comb
import scipy as SP
import pdb
n_dimensions = 3
X = SP.randn(3, n_dimensions)
params = SP.zeros([0])
if 0:
c1 = limix.CCovLinearISO()
c2 = lin.LinearCFISO(n_dimensions=n_dimensions)
c1.setX(X)
K1 = c1.K()
K2 = c2.K(params, X, X)
dK1 = c1.Kgrad_param(0)
dK2 = c2.Kgrad_theta(params, X, 0)
dKx1 = c1.Kgrad_X(0)
dKx2 = c2.Kgrad_x(params, X, X, 0)
dKx1diag = c1.Kdiag_grad_X(0)
dKx2diag = c2.Kgrad_xdiag(params, X, 0)