| Title: | Lowest Radial Distance Method of Marginal Likelihood Estimation |
| Version: | 0.0.1.0 |
| Description: | Estimates marginal likelihood from a posterior sample using the method described in Wang et al. (2023) <doi:10.1093/sysbio/syad007>, which does not require evaluation of any additional points and requires only the log of the unnormalized posterior density for each sampled parameter vector. |
| License: | MIT + file LICENSE |
| Encoding: | UTF-8 |
| RoxygenNote: | 7.2.3 |
| Suggests: | knitr, rmarkdown, testthat (≥ 3.0), rstan (≥ 2.32), bridgesampling (≥ 1.1) |
| Imports: | stats |
| Maintainer: | Analisa Milkey <analisa.milkey@uconn.edu> |
| Author: | Analisa Milkey 
[aut, cre],
Elena Korte [aut],
Paul O. Lewis
[aut] |
| Config/testthat/edition: | 3 |
| VignetteBuilder: | knitr, rmarkdown |
| Depends: | R (≥ 3.0) |
| LazyData: | true |
| NeedsCompilation: | no |
| Packaged: | 2023-12-15 20:27:47 UTC; analisamilkey |
| Repository: | CRAN |
| Date/Publication: | 2023-12-17 08:00:05 UTC |
Sequence data used in gtrig vignette
Description
Sequence data used in gtrig vignette
Usage
gtrigsamples
Format
gtrigsamples
A data frame with 10,001 rows and 35 columns:
- Iteration
MCMC iteration
- Posterior
Log of the unnormalized posterior density
- Likelihood
Log likelihood
- Prior
Log of the prior density
- alpha
Shape parameter of the (mean=1) Gamma distribution of among-site rate heterogeneity
- edge_length_proportions.1.
Proportion of total tree length used by edge 1
- edge_length_proportions.2.
Proportion of total tree length used by edge 2
- edge_length_proportions.3.
Proportion of total tree length used by edge 3
- edge_length_proportions.4.
Proportion of total tree length used by edge 4
- edge_length_proportions.5.
Proportion of total tree length used by edge 5
- edge_length_proportions.6.
Proportion of total tree length used by edge 6
- edge_length_proportions.7.
Proportion of total tree length used by edge 7
- edgelens.1.
Edge length 1
- edgelens.2.
Edge length 2
- edgelens.3.
Edge length 3
- edgelens.4.
Edge length 4
- edgelens.5.
Edge length 5
- edgelens.6.
Edge length 6
- edgelens.7.
Edge length 7
- er.1.
Exchangeability parameter for A to C
- er.2.
Exchangeability parameter for A to G
- er.3.
Exchangeability parameter for A to T
- er.4.
Exchangeability parameter for C to G
- er.5.
Exchangeability parameter for C to T
- er.6.
Exchangeability parameter for G to T
- pi.1.
Nucleotide relative frequency for A
- pi.2.
Nucleotide relative frequency for C
- pi.3.
Nucleotide relative frequency for G
- pi.4.
Nucleotide relative frequency for t
- pinvar
Proportion of invariable sites
- site_rates.1.
Rate for site category 1
- site_rates.2.
Rate for site category 1
- site_rates.3.
Rate for site category 1
- site_rates.4.
Rate for site category 1
- tree_length
Tree length (sum of all edge lengths) in substitutions per site
Source
The program RevBayes (version 1.2.1) was used to obtain a sample from the Bayesian posterior distribution for 5 green plant rbcL sequences under a GTR+I+G model.
Sequence data used in k80 vignette
Description
Sequence data used in k80 vignette
Usage
k80samples
Format
k80samples
A data frame with 10,000 rows and 4 columns:
- iter
Iteration
- log.kernel
Log unnormalized posterior
- edgelen
Edge length in substitutions per site
- kappa
Transition transversion rate ratio
Source
Calculate a sum on log scale
Description
Calculates the (natural) log of a sum without leaving the log scale by factoring out the largest element.
Usage
lorad_calc_log_sum(logx)
Arguments
logx |
Numeric vector in which elements are on log scale |
Value
The log of the sum of the (exponentiated) elements supplied in logx
Calculates the LoRaD estimate of the marginal likelihood
Description
Provided with a data frame containing sampled paraneter vectors and a dictionary relating column names to parameter types, returns a named character vector containing the following quantities:
logML (the estimated log marginal likelihood)
nsamples (number of samples)
nparams (length of each parameter vector)
training_frac (fraction of samples used for training)
tsamples (number of samples used for training)
esamples (number of sampled used for etimation)
coverage (nominal fraction of the estimation sampled used)
esamplesused (number of estimation samples actually used for estimation)
realized_coverage (actual fraction of estimation sample used)
rmax (lowest radial distance: defines boundary of working parameter space)
log_delta (volume under the unnormalized posterior inside working parameter space)
Usage
lorad_estimate(params, colspec, training_frac, training_mode, coverage)
Arguments
params |
Data frame in which rows are sample points and columns are parameters, except that last column holds the log posterior kernel |
colspec |
Named character vector associating column names in params with column specifications |
training_frac |
Number between 0 and 1 specifying the training fraction |
training_mode |
One of random, left, or right, specifying how training fraction is chosen |
coverage |
Number between 0 and 1 specifying fraction of training sample used to compute working parameter space |
Value
Named character vector of length 11.
Examples
normals <- rnorm(1000000,0,10)
prob_normals <- dnorm(normals,0,10,log=TRUE)
proportions <- rbeta(1000000,1,2)
prob_proportions <- dbeta(proportions,1,2,log=TRUE)
lengths <- rgamma(1000000, 10, 1)
prob_lengths <- dgamma(lengths,10,1,log=TRUE)
paramsdf <- data.frame(
normals,prob_normals,
proportions, prob_proportions,
lengths, prob_lengths)
columnkey <- c(
"normals"="unconstrained",
"prob_normals"="posterior",
"proportions"="proportion",
"prob_proportions"="posterior",
"lengths"="positive",
"prob_lengths"="posterior")
results <- lorad_estimate(paramsdf, columnkey, 0.5, 'random', 0.1)
lorad_summary(results)
Transforms unconstrained parameters to have the same location and scale
Description
Standardizes parameters that have already been transformed (if necessary) to have unconstrained support. Standardization involves subtracting the sample mean and dividing by the sample standard deviation. Assumes that the log posterior kernel (i.e. the log of the unnormalized posterior) is the last column in the supplied data frame.
Usage
lorad_standardize(df, coverage)
Arguments
df |
Data frame containing a column for each model parameter sampled and a final column of log posterior kernel values |
coverage |
Fraction of the training sample used to compute working parameter space |
Value
List containing the log-Jacobian of the standardization transformation, the inverse square root matrix, a vector of column means, and rmax (radial distance to furthest point in working parameter space)
Transforms training sample using training sample means and standard deviations
Description
Transforms training sample using training sample means and standard deviations
Usage
lorad_standardize_estimation_sample(standardinfo, y)
Arguments
standardinfo |
List containing the log Jacobian of the standardization transformation, the inverse square root matrix, the column means, and rmax (the radial distance representing the edge of the working parameter space) |
y |
Data frame containing a column for each transformed model parameter in the estimation sample, with last column being the log kernel values |
Value
A new data frame consisting of the standardized estimation sample with log kernel in last column
Summarize output from lorad_estimate()
Description
Summarize output from lorad_estimate()
Usage
lorad_summary(results)
Arguments
results |
Named character vector returned from |
Value
String containing a summary of the supplied results object
Examples
normals <- rnorm(1000000,0,10)
prob_normals <- dnorm(normals,0,10,log=TRUE)
paramsdf <- data.frame(normals,prob_normals)
columnkey <- c("normals"="unconstrained", "prob_normals"="posterior")
results <- lorad_estimate(paramsdf, columnkey, 0.5, 'left', 0.1)
lorad_summary(results)
Log (or log-ratio) transform parameters having constrained support
Description
Log-transforms parameters with support (0,infinity), log-ratio transforms K-dimensional parameters with support a (K-1)-simplex, logit transforms parameters with support [0,1], and leaves unchanged parameters with unconstrained support (-infinity, infinity).
Usage
lorad_transform(params, colspec)
Arguments
params |
Data frame containing a column for each model parameter sampled as well as one or more columns that, when summed, constitute the log joint posterior kernel |
colspec |
Named character vector matching each column name in params with a column specification |
Value
A new data frame comprising transformed parameter values with a final column holding the log joint posterior kernel