3.5 Output Templates

An output template file consists of text with markup commands and placeholders for the program output. The text is copied into the final output. The backslash <\> character is used as an escape character, removing any special effect of the following character, so that it is treated as text. The markup commands currently have little effect, but may be useful in later versions if alternative output formats are supported. (e.g. tex, html, postscript, ...). The placeholders are replaced by the appropriate program output.

Markup commands start with the <<> character, and end with the next whitespace character. The end of the range of affected text is indicated with the <>> character.

The markup commands currently supported are:

comment

This begins a comment – the text between this and the following <>> character is not included in the output.

table

This begins a table. Table columns are separated with the <&> character, and rows are separated with the <;> character.

include

This inserts another template file into the template. This allows partial templates to be reused in many output formats.

noeffect

This has no effect. It is to allow for conditional markup commands. One condition can give one markup command, while another can give this noeffect command.

Placeholders begin with the <{> character and end with the <}> character. The first word of the placeholder is an expression, giving the value to be included. Currently only single variables are possible, but hopefully support for arithmetic expressions will be made available in later versions. The following variables are currently available:

varsel

A list of the variables selected. Useful if the --variableselect option is used.

LL

The log likelihood at the current point.

converge

Whether or not the algorithm converged.

numsteps

The number of iterations of Newton-Raphson performed.

varnum

The number of the variable (usually in a table, where there will be a row for each variable.

truevarnum

The number of the variable within the original numbering from the input file. (This may be different from the variable number if only some variables are selected.)

varname

Same as truevarnum cases

coeff

The coefficient of the variable. This is the parameter in the model provided to cold.

expcoeff

The exponential of the coefficient. This is the parameter used in some other models.

tstat

The modified t statistic for the variable. This can be used to test the hypothesis that the variable is non-zero.

branchno

The number of the branch.

branchtop

The node above a branch.

branchbot

The node below a branch.

branchlen

The length of a branch.

tree

The inferred tree. Currently only Newick format is supported.

branchlenstd

The standard deviation of the estimated branch length (based on the observed information matrix).

treestd

The standard deviations of all branchlengths on a formatted tree.

mixnum

The number of site classes.

mixclass

The number of the current site class.

mixprob

The mixing probability for the site class.

siteno

The number of the site.

sitelike

The log-likelihood of the given site.

sitepostprob

The posterior probabilities for the given site.

observedinformation

The hessian matrix at the estimate.

observedinformationinverse

The inverse of the modified hessian matrix.

After the initial expression, various modifiers are allowed, affecting the way in which the output is formatted. The options vary, depending on the specific output. For example, there are options controlling the precision of numbers. For output that has only a discrete set of possible outcomes (such as @varnum, or @converge) the options can include a list of choices. This list is of the form value: followed by the text that should be displayed if the value matches, terminated with the <|> character.

Markup commands and placeholders can be nested using braces (<{> and <}>).