ratetest.pow: 'ratetest' power function

In olli0601/abc.star: abc.star

Description

Computes the power function of the two-sided, one-sample gamma test for testing if simulated and observed summary values occur at similar rates. This test is applicable when the observed and simulated summary values are Exponentially distributed, or when Exponentiality cannot be rejected. The testing problem is described in terms of the scale parameter of the Exponential distribution, the reciprocal of the rate parameter.

Usage

ratetest.pow(rho, c.l, c.u, m, norm = 1, support = c(0, Inf), log = FALSE)

Arguments

rho	Vector of error quantiles
c.l	Lower boundary point of the critical region (equivalent to the lower ABC tolerance epsilon^-)
c.u	Upper boundary point of the critical region (equivalent to the upper ABC tolerance epsilon^+)
m	Number of simulated values
norm	Normalization constant for the truncated power function
support	Support of the truncated power function
log	If TRUE, the power function is returned on the log scale.

Note

The power function can be truncated to support and then standardized with norm. If one of these is set, the other must be provided too.

See Also

ratetest.calibrate, ratetest.pow.norm

Examples

n.of.y <- 40

# compute ABC tolerances
cali	<- ratetest.calibrate(tau.l=1/2, tau.u=2, n.of.y=n.of.y, what='CR', alpha=0.01)
# compute the power for the range (0.1, 3)
rho  	<- seq(0.1, 3, len=1024)
tmp		<- data.frame(rho=rho, power=ratetest.pow(rho, cali['c.l'], cali['c.u'], m=n.of.y))

library(ggplot2)
p 		<- ggplot(tmp,aes(x=rho,y=power)) + geom_line() + labs(y='Power\n(ABC acceptance probability)')
print(p)

olli0601/abc.star documentation built on May 24, 2019, 12:53 p.m.