adelie.adelie_core.state.StateCSSCov32#

class adelie.adelie_core.state.StateCSSCov32#

Core state class for CSS via covariance method.

Methods

`__init__`(args, *kwargs)	Overloaded function.
`solve`(self)	Solves the state-specific problem.

Attributes

`L_T`	Lower Cholesky factor of \(\Sigma_{TT}\) where \(T\) is the current subset.
`S_resid`	Residual covariance matrix \(\Sigma - \Sigma_{\cdot T} \Sigma_{TT}^{\dagger} \Sigma_{T \cdot}\) where \(T\) is the current subset.
`benchmark_L_T`	Benchmark time for updating the Cholesky factor after adding a feature.
`benchmark_L_U`	Benchmark time for updating the Cholesky factor after removing a feature.
`benchmark_S_resid`	Benchmark time for updating the residual covariance matrix after removing a feature.
`benchmark_init`	Benchmark time for the initialization.
`benchmark_resid_fwd`	Benchmark time for updating the residual covariance matrix after adding a feature.
`benchmark_scores`	Benchmark time for computing the scores.
`subset`	Selected subset.

__init__(*args, **kwargs)#

Overloaded function.

__init__(self: adelie.adelie_core.state.StateCSSCov32, S: numpy.ndarray[numpy.float32[m, n], flags.f_contiguous], subset_size: int, subset: numpy.ndarray[numpy.int64[1, n]], method: str, loss: str, max_iters: int, n_threads: int) -> None
__init__(self: adelie.adelie_core.state.StateCSSCov32, arg0: adelie.adelie_core.state.StateCSSCov32) -> None

solve(self: adelie.adelie_core.state.StateCSSCov32) → dict#: Solves the state-specific problem.

L_T#: Lower Cholesky factor of \(\Sigma_{TT}\) where \(T\) is the current subset.

S_resid#: Residual covariance matrix \(\Sigma - \Sigma_{\cdot T} \Sigma_{TT}^{\dagger} \Sigma_{T \cdot}\) where \(T\) is the current subset.

benchmark_L_T#: Benchmark time for updating the Cholesky factor after adding a feature.

benchmark_L_U#: Benchmark time for updating the Cholesky factor after removing a feature.

benchmark_S_resid#: Benchmark time for updating the residual covariance matrix after removing a feature.

benchmark_init#: Benchmark time for the initialization.

benchmark_resid_fwd#: Benchmark time for updating the residual covariance matrix after adding a feature.

benchmark_scores#: Benchmark time for computing the scores.

subset#: Selected subset.