jackknife-after-bootstrap: jackknife-after-bootstrap analysis

In hadron: Analysis Framework for Monte Carlo Simulation Data in Physics

jackknife-after-bootstrap analysis

Description

jackknife-after-bootstrap (JAB) analysis for errors of errors of correlation functions of class cf.

We apply the jackknife-after-bootstrap method as proposed by Efron (1992) for iid data and extended by Lahiri (2002) for dependent data. Blocks of bootstrap samples are deleted for a jackknife analysis. The jackknife replicates are computed from the bootstrap samples in which the corresponding block of blocks is missing.

We use here the moving blocked bootstrap (MBB) which uses overlapping blocks. The estimate of standard error of the bootstrap error is computed using formula (2.3) from Lahiri, 2002:

(m(N-m)^-1)M^-1 ∑_{i=1}^M (ttilde_n^(i)-that_n)^2

with

(N that_n - (N-m) that_n^(i)).

Here, that_n is the MBB estimate (in our case of standard deviation) and that_n^(i) is the i-th jackknife replication of it.

Arguments

cf	An object of class cf generated by bootstrap.cf with sim="fixed".
m	m denotes the number of (overlapping) blocks to delete for the JAB analysis.

Value

Returns an object of class cf with an element jack.boot.se, which is the JAB estimate of standard error of the standard error.

Author(s)

Carsten Urbach [e-mail]

References

S.N. Lahiri, "On the jackknife-after-bootstrap method for dependent data and its consistency properties", Econometric Theory, 2002, Vol. 18, 79-98

See Also

bootstrap.cf, cf, jackknife.cf

hadron documentation built on Sept. 9, 2022, 5:06 p.m.