adelie.adelie_core.state.StatePinball32#

class adelie.adelie_core.state.StatePinball32#

Core state class for pinball least squares.

Methods

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

Attributes

`active_set`	Active set buffer.
`active_set_size`	Active set size.
`benchmark_fit_active`	Benchmark time for fitting on the active set.
`benchmark_fit_screen`	Benchmark time for fitting on the screen set.
`benchmark_gradient`	Benchmark time for computing the gradient.
`benchmark_viols_sort`	Benchmark time for sorting the violations.
`beta`	Coefficient vector.
`dbg_active_set`	List of the active sets at each outer loop.
`dbg_beta`	List of the coefficient vectors at each outer loop.
`dbg_iter`	List of the number of iterations at each outer loop.
`dbg_loss`	List of the losses at each outer loop.
`grad`	Internal buffer that is implementation-defined.
`is_active`	Boolean buffer to indicate the active variables.
`is_screen`	Boolean buffer to indicate the screen variables.
`iters`	Number of coordinate descent iterations.
`loss`	Loss \(\frac{1}{2} \\|S^{-\frac{1}{2}} v - S^{\frac{1}{2}} A^\top \beta\\|_2^2\).
`resid`	Residual \(v - S A^\top \beta\).
`screen_AS`	\(A S\).
`screen_ASAT_diag`	Diagonal of \(A S A^\top\).
`screen_set`	Screen set buffer.
`screen_set_size`	Screen set size.

__init__(*args, **kwargs)#

Overloaded function.

__init__(self: adelie.adelie_core.state.StatePinball32, A: adelie.adelie_core.matrix.MatrixConstraintBase32, y_var: float, S: numpy.ndarray[numpy.float32[m, n], flags.f_contiguous], penalty_neg: numpy.ndarray[numpy.float32[1, n]], penalty_pos: numpy.ndarray[numpy.float32[1, n]], kappa: int, max_iters: int, tol: float, screen_set_size: int, screen_set: numpy.ndarray[numpy.int64[1, n], flags.writeable], is_screen: numpy.ndarray[bool[1, n], flags.writeable], screen_ASAT_diag: numpy.ndarray[numpy.float32[1, n], flags.writeable], screen_AS: numpy.ndarray[numpy.float32[m, n], flags.writeable, flags.c_contiguous], active_set_size: int, active_set: numpy.ndarray[numpy.int64[1, n], flags.writeable], is_active: numpy.ndarray[bool[1, n], flags.writeable], beta: numpy.ndarray[numpy.float32[1, n], flags.writeable], resid: numpy.ndarray[numpy.float32[1, n], flags.writeable], grad: numpy.ndarray[numpy.float32[1, n], flags.writeable], loss: float) -> None
__init__(self: adelie.adelie_core.state.StatePinball32, arg0: adelie.adelie_core.state.StatePinball32) -> None

solve(self: adelie.adelie_core.state.StatePinball32) → dict#: Solves the state-specific problem.

loss#: Loss \(\frac{1}{2} \|S^{-\frac{1}{2}} v - S^{\frac{1}{2}} A^\top \beta\|_2^2\).