stheno.model.observations module¶

class stheno.model.observations.AbstractObservations[source]¶

Bases: object

Abstract base class for observations.

Also takes in one or multiple tuples of the described arguments.

Parameters

fdd (fdd.FDD) – FDD of observations.
y (column vector) – Values of observations.

fdd¶

FDD of observations.

Type: fdd.FDD

y¶

Values of observations.

Type: column vector

posterior_kernel(measure, p_i, p_j)[source]¶

Get the posterior kernel between two processes.

Parameters

measure (measure.Measure) – Prior.
p_i (gp.GP) – First process.
p_j (gp.GP) – Second process.

Returns

Posterior kernel between the first and: second process.

Return type

mlkernels.Kernel

posterior_mean(measure, p)[source]¶

Get the posterior kernel of a process.

Parameters

measure (measure.Measure) – Prior.
p (gp.GP) – Process.

Returns

Posterior mean of p.

Return type

mlkernels.Mean

stheno.model.observations.Obs¶

Shorthand for Observations.

alias of stheno.model.observations.Observations

class stheno.model.observations.Observations(*args)[source]¶

Bases: stheno.model.observations.AbstractObservations

Observations.

Takes arguments according to the constructor of measure.AbstractObservations.

Parameters

fdd (fdd.FDD) – FDDs to corresponding to the observations.
y (tensor) – Values of observations.

K_x(measure)[source]¶

Kernel matrix of the data.

Parameters: measure (measure.Measure) – Measure.
Returns: Kernel matrix.
Return type: matrix

posterior_kernel(measure, p_i, p_j)[source]¶

Get the posterior kernel between two processes.

Parameters

measure (measure.Measure) – Prior.
p_i (gp.GP) – First process.
p_j (gp.GP) – Second process.

Returns

Posterior kernel between the first and: second process.

Return type

mlkernels.Kernel

posterior_mean(measure, p)[source]¶

Get the posterior kernel of a process.

Parameters

measure (measure.Measure) – Prior.
p (gp.GP) – Process.

Returns

Posterior mean of p.

Return type

mlkernels.Mean

stheno.model.observations.PseudoObs¶

Shorthand for PseudoObservations.

alias of stheno.model.observations.PseudoObservations

class stheno.model.observations.PseudoObservations[source]¶

Bases: stheno.model.observations.AbstractObservations

Observations through inducing points.

Further takes arguments according to the constructor of measure.AbstractObservations. Can also take in tuples of inducing points.

Parameters: u (fdd.FDD) – Inducing points

A(measure)[source]¶

Parameter of the corrective variance of the kernel of the optimal approximating distribution.

Parameters: measure (measure.Measure) – Measure.
Returns: Corrective variance.
Return type: matrix

K_z(measure)[source]¶

Kernel matrix of the data.

Parameters: measure (measure.Measure) – Measure.
Returns: Kernel matrix.
Return type: matrix

elbo(measure)[source]¶

ELBO.

Parameters: measure (measure.Measure) – Measure.
Returns: ELBO.
Return type: scalar

mu(measure)[source]¶

Mean of optimal approximating distribution.

Parameters: measure (measure.Measure) – Measure.
Returns: Mean.
Return type: matrix

posterior_kernel(measure, p_i, p_j)[source]¶

Get the posterior kernel between two processes.

Parameters

measure (measure.Measure) – Prior.
p_i (gp.GP) – First process.
p_j (gp.GP) – Second process.

Returns

Posterior kernel between the first and: second process.

Return type

mlkernels.Kernel

posterior_mean(measure, p)[source]¶

Get the posterior kernel of a process.

Parameters

measure (measure.Measure) – Prior.
p (gp.GP) – Process.

Returns

Posterior mean of p.

Return type

mlkernels.Mean

stheno.model.observations.SparseObs¶: alias of stheno.model.observations.PseudoObservations

stheno.model.observations.SparseObservations¶: alias of stheno.model.observations.PseudoObservations

stheno.model.observations.combine[source]¶

Combine multiple FDDs or tuples of observations into one.

Parameters: *objs (fdd.FDD or tuple) – FDDs or tuples of observations.
Returns: args combined into one.
Return type: fdd.FDD or tuple