mgplvm.kernels.stationary module
- class mgplvm.kernels.stationary.Exp(n, distance, d=None, ell=None, scale=None, learn_scale=True, Y=None, eps=1e-06)[source]
Bases:
mgplvm.kernels.stationary.QuadExp- K(x, y)[source]
- Parameters
- xTensor
input tensor of dims (… n x d x mx)
- yTensor
input tensor of dims (… n x d x my)
- Returns
- kxyTensor
exponential kernel with dims (… n x mx x my)
- Return type
Tensor
- training: bool
- class mgplvm.kernels.stationary.Matern(n, distance, d=None, nu=1.5, ell=None, scale=None, learn_scale=True, Y=None, eps=1e-06)[source]
Bases:
mgplvm.kernels.stationary.Stationary- K(x, y)[source]
- Parameters
- xTensor
input tensor of dims (… n x d x mx)
- yTensor
input tensor of dims (… n x d x my)
- Returns
- kxyTensor
matern kernel with dims (… n x mx x my)
- Return type
Tensor
- property msg
- training: bool
- class mgplvm.kernels.stationary.QuadExp(n, distance, d=None, ell=None, scale=None, learn_scale=True, Y=None, eps=1e-06, ell_byneuron=True)[source]
Bases:
mgplvm.kernels.stationary.Stationary- K(x, y)[source]
- Parameters
- xTensor
input tensor of dims (… n x d x mx)
- yTensor
input tensor of dims (… n x d x my)
- Returns
- kxyTensor
quadratic exponential kernel with dims (… n x mx x my)
- Return type
Tensor
- training: bool
- class mgplvm.kernels.stationary.Stationary(n, distance, d=None, ell=None, scale=None, learn_scale=True, Y=None, eps=1e-06, ell_byneuron=True)[source]
Bases:
mgplvm.kernels.kernel.Kernel- diagK(x)[source]
- Parameters
- xTensor
input tensor of dims (… n x d x mx)
- Returns
- diagKTensor
diagonal of kernel K(x,x) with dims (… n x mx )
- Return type
Tensor
- property ell: torch.Tensor
- Return type
Tensor
- property msg
- property prms: Tuple[torch.Tensor, torch.Tensor]
- Return type
Tuple[Tensor,Tensor]
- property scale: torch.Tensor
- Return type
Tensor
- property scale_sqr: torch.Tensor
- Return type
Tensor
- trK(x)[source]
- Parameters
- xTensor
input tensor of dims (… n x d x mx)
- Returns
- trKTensor
trace of kernel K(x,x) with dims (… n)
- Return type
Tensor
- training: bool