Description Usage Arguments Value Examples
View source: R/stitching_mixture_method.R
SGLR_CI
is used to compute R functions which compute bounds of GLR-like, stitching and discrete mixture confidence sequences for general sub-psi class designed to a finite target time interval. For the additive sub-psi class, we recommend to use SGLR_CI_additive
for more efficient computations.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | SGLR_CI(
alpha,
nmax = NULL,
nmin = 1L,
d = NULL,
m_upper = 1000L,
breg = function(mu_1, mu_0) { (mu_1 - mu_0)^2/2 },
breg_pos_inv = function(d, mu_0) { mu_0 + sqrt(2 * d) },
breg_neg_inv = function(d, mu_0) { mu_0 - sqrt(2 * d) },
breg_derv = function(z, mu_0) { z - mu_0 },
mu_lower = NULL,
mu_upper = NULL,
grid_by = NULL,
n_0 = 1L
)
|
alpha |
An upper bound on the boundary crossing probability (positive numeric in |
nmax |
Upper bound of the target time interval. If both |
nmin |
Lower bound of the target time interval. (default = 1L) |
d |
Bregman divergence between the null and alternative spaces. Either |
m_upper |
Upper bound on the grid-search for m (default = 1e+3L). |
breg |
R function of D(μ_1, μ_0) which takes |
breg_pos_inv |
R function of inverse of the mapping z \mapsto D(z, μ_0):= d on z > μ_0 which takes |
breg_neg_inv |
R function of inverse of the mapping z \mapsto D(z, μ_0):= d on z < μ_0 which takes |
breg_derv |
R function of \nabla_z D(z, μ_0) which takes |
mu_lower |
Lower bound of the mean parameter space (default = NULL). |
mu_upper |
Upper bound of the mean parameter space (default = NULL). |
grid_by |
The size of grid-window of mean space. Default is |
n_0 |
Lower bound of the sample size on which test statistics and CI will be computed (default = 1L). |
A list of R functions for GLR-like, stitching and discrete mixture bound which takes sample mean x_bar
and sample size n
as the input and return the anytime-valid confidence interval at n
. The list also contains related quantities to compute these bounds. See ADD_CITE for detailed explanations of these quantities.
R function for the GLR-like bound.
R function for the discrete mixture bound.
R function to generate log of GLR-like statistic minus the threshold.
R function to generate log of GLR-like statistic minus the threshold.
alpha valud used to construct GLR-like and discrete mixture bound functions
The boundary value for initial nmin
and nmax
.
The eta value used to construct underlying martingales
The number of LR-like martingales used to construct bounds for n ≥q n_{\min} part.
Minimum and maximum of the mean parater space.
The size of grid-window of mean space.
1 2 |
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.