Diagonal#

class hmclab.MassMatrices.Diagonal(diagonal: numpy.ndarray, rng: Optional[numpy.random._generator.Generator] = None)[source]#

Bases: hmclab.MassMatrices._AbstractMassMatrix

The diagonal mass matrix.

This mass matrix or metric does only performs scaling on each dimension separately. It is optimal when all parameters of your target distribution are expected to be independent (not have trade-offs) but still varying scales of disperion / variance.

__init__(diagonal: numpy.ndarray, rng: Optional[numpy.random._generator.Generator] = None)[source]#: Constructor for diagonal mass matrices.

Methods

`__init__`	Constructor for diagonal mass matrices.
`create_default`
`full_name`
`generate_momentum`
`kinetic_energy`	param momentum
`kinetic_energy_gradient`	param momentum

Attributes

`dimensions`
`matrix`
`name`
`rng`

kinetic_energy(momentum: numpy.ndarray) → float[source]#

Parameters: momentum –

kinetic_energy_gradient(momentum: numpy.ndarray) → numpy.ndarray[source]#

Parameters: momentum –

generate_momentum() → numpy.ndarray[source]#