In [1]:

    

import VFF
%matplotlib inline
import numpy as np
from matplotlib import pyplot as plt


    

In [55]:

    

X = np.random.rand(100, 2)
Y = np.sin(3 * X[:, :1]) * np.sin(5 * X[:, 1:]) + np.random.randn(X.shape[0], 1) * 0.1
# Y = np.sin(8 * X[:, 1:])


    

In [56]:

    

plt.scatter(X[:, 0], X[:, 1], 20, Y[:, 0])
plt.colorbar()


    

    
Out[56]:





<matplotlib.colorbar.Colorbar at 0x1216f5e80>






    

In [64]:

    

m = VFF.gpr.GPRKron(X, Y, np.arange(18), np.array([-1, -1]), np.array([2, 2]), [GPflow.kernels.Matern12(1), GPflow.kernels.Matern12(1)])
m.likelihood.variance = 0.1


    

In [65]:

    

k = GPflow.kernels.Matern32(1) * GPflow.kernels.Matern32(1, active_dims=[1]) + GPflow.kernels.White(1)
m0 = GPflow.gpr.GPR(X, Y, kern=k)
m0.likelihood.variance = 0.1


    

In [66]:

    

m.optimize()
m0.optimize()


    

    
Out[66]:





      fun: -45.779183835929516
 hess_inv: <6x6 LbfgsInvHessProduct with dtype=float64>
      jac: array([ -4.16419685e-06,   6.66518484e-06,  -1.31283273e-05,
         6.66518483e-06,   1.92558733e-07,   2.75248998e-06])
  message: b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'
     nfev: 18
      nit: 14
   status: 0
  success: True
        x: array([-0.5816852 , -0.65854351, -0.84389603, -0.65854351, -7.34731781,
       -4.67885514])

In [67]:

    

m


    

    
Out[67]:




Name
values
prior
constraint
model.kerns.item0.lengthscales
[ 0.61889174]
None
+ve
model.kerns.item0.variance
[ 0.26679497]
None
+ve
model.kerns.item1.lengthscales
[ 0.40846014]
None
+ve
model.kerns.item1.variance
[ 0.26679497]
None
+ve
model.likelihood.variance
[ 0.01991182]
None
+ve

In [68]:

    

m0


    

    
Out[68]:




Name
values
prior
constraint
name.kern.prod.matern32_1.lengthscales
[ 0.44401712]
None
+ve
name.kern.prod.matern32_1.variance
[ 0.41713422]
None
+ve
name.kern.prod.matern32_2.lengthscales
[ 0.35769771]
None
+ve
name.kern.prod.matern32_2.variance
[ 0.41713422]
None
+ve
name.kern.white.variance
[ 0.00064511]
None
+ve
name.likelihood.variance
[ 0.00924776]
None
+ve

In [69]:

    

Xtest = np.random.rand(100,2)
mu, var = m0.predict_f(Xtest)
plt.scatter(Xtest[:, 0], Xtest[:, 1], 20, mu[:, 0], vmin=-1, vmax=1)
plt.colorbar()


    

    
Out[69]:





<matplotlib.colorbar.Colorbar at 0x1255879e8>






    

In [70]:

    

mu, var = m.predict_f(Xtest)
plt.scatter(Xtest[:, 0], Xtest[:, 1], 20, mu[:, 0], vmin=-1, vmax=1)
plt.colorbar()


    

    
Out[70]:





<matplotlib.colorbar.Colorbar at 0x128308b00>






    

In [ ]:

    




    

In [ ]: