---
title: "Print methods in the permutations package"
author: "Robin K. S. Hankin"
output: html_vignette
bibliography: permutations.bib
vignette: >
  %\VignetteIndexEntry{print methods}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r out.width='20%', out.extra='style="float:right; padding:10px"',echo=FALSE}
knitr::include_graphics(system.file("help/figures/permutations.png", package = "permutations"))
```


```{r set-options, echo = FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>", dev = "png", fig.width = 7, fig.height = 3.5, message = FALSE, warning = FALSE)
options(width = 80, tibble.width = Inf)
suppressMessages(suppressWarnings(require("permutations")))
```

To cite the permutations package in publications, please use
@hankin2020.  The `permutations` package has a print method which
includes a number of user-configurable options which are illustrated
and motivated here.

Permutations have two natural representations: word form and cycle
form.  Internally, the package coerces a permutation from one form to
another depending on what operations one does with it.
Group-theoretic products and inverses are carried out more easily in
word form, while powers are more easily done using cycle form.

Partly as a result of a perceptive comment from a journal reviewer,
the package coerces to cycle form for printing as this is generally
more comprehensible than word form.  However, the package print method
is extensively customizable.  This document covers some of the
options.  Low-level print functionality includes `print_word()` and
`print_cycle()` which can be used explicitly to print in either word
form or cycle form as desired:

```{r}
set.seed(0)
x <- rperm(r = 9)
print_word(x)
print_cycle(x)
```

It is a matter of taste which one is preferable.  For the example
above, the majority of the symbols are moved: elements that map to
themselves are shown with a dot in the word form, and these are in a
minority.  However, the difference between word form and cycle form
becomes more pronounced if only a small number of elements move:

```{r}
x <- rperm(r = 9, moved = 3)
print_word(x)
print_cycle(x)
```

Above we see the cycle form is arguably more compact.

## Print word as cycle

In use, sometimes we work with objects in word form and sometimes
objects in cycle form.  However, the standard workflow would be to use
R's default system, which is to print values to the console:

```{r}
x <- rperm(n = 2)
y <- as.cycle(x)
unclass(x) # shows internal representation (x is in word form)
unclass(y) # shows internal representation (x is in cycle form)
x # default: print to console
y # default: print to console
```

Above we see that, by default, objects in word form are coerced to
their cycle form for echo-printing.  However, it is possible to
suppress this coercion by setting option `print_word_as_cycle`:

```{r}
options(print_word_as_cycle = FALSE)
x
y
```

Above we see that the coercion is not performed on word object `x`:
the print method shows the internal representation of the object.
This behaviour can be confusing because the same permutation can have
different print styles, so we reinstate the default:

```{r}
options(print_word_as_cycle = TRUE) # restore default
```

Most people seem to prefer printing in cycle form.

###  Commas

Above, we saw the print method for permutations of $\left[9\right]$.
If we have 10 or more elements then we need a comma.  This is added by
default:

```{r}
x <- rperm(r = 15, moved = 4)
print_cycle(x)
```


Above, the comma has been added to aid readability.  However, it is
possible to override this behaviour by setting option `comma`.  The
default value of `NULL` means to add commas iff $r>9$ but Boolean
values are respected:

```{r}
options("comma" = TRUE)
rperm(3, r = 9) # commas printed irregardless
```

Above we see the commas making the output somewhat prolix, but not a
disaster.  However, if we are permuting $>9$ objects commas are
necessary to interpret strings like `124` which might be `1,2,4` or
`12,4` or `1,24` or `124`:

```{r}
options("comma" = FALSE)
x <- rperm(3, r = 20)
x # commas not printed irregardless
options("comma" = NULL) # restore default
x # default for comparison
```

Above, the absence of commas is somewhat confusing, default restored.


### Symbols

The `permutations` package considers permutations of a finite set.  It
is _very_ convenient to identify the finite set with integers
$1,2,\ldots,n$ [chiefly because products and inverses are easy if one
can use R's 1-based vector indexing: products are just `a*b=b[a]` and
the inverse of `W` is just `W[W] <- seq_along(W)`].  Because of this,
the default print method echoes the internal representation of a
permutation.

```{r}
set.seed(0)
x <- rperm(n = 3)
dput(x)
x
```

We see the print method echoing the internal representation of the set
$\left[n\right]=\left\lbrace 1,2,\ldots,n\right\rbrace$.  However, it
is possible to use a different set, using option `perm_set`:


```{r}
options(perm_set = letters)
rperm()
```

Above we see cyclic representation of permutations of the letters a-z.
However, the use of commas is somewhat problematic, and by default the
print method uses the $>9$ criterion for including a comma:

```{r}
(xx <- rperm(n = 2, r = 26))
```

Above we see that commas are included (because $26>9$) but it probably
looks better without them:


```{r}
options(comma = FALSE)
xx
```

One side-effect of using a finite set of symbols is that the print
method might run out of symbols:

```{r}
(z <- rperm(n = 2, r = 50))
```

Above we see `NA` printed where the symbol's index exceeds 26 (also
the absence of commas does not help).  Remember that the print method
does not change the object itself, so this might not be an issue.  It
is possible to use symbols that have more than one character:

```{r}
options(perm_set = state.abb)
options(comma = TRUE)
z
```

## Cycles in length order 


Consider the following:

```{r rgivshap}
options(perm_set = NULL) # revert to numbers
options(comma = FALSE) # supress comma
x <- rgivenshape(30, 2:4)
x
```

We see 30 random permutations with shape
$\left(\cdot\,\cdot\right)\left(\cdot\cdot\cdot\right)\left(\cdot\cdot\cdot\right)$.
However, because function `nicify_cyclist()` sorts each cycle so that
the smallest element is first, then sorts the cycles by first element,
it is not obvious that all the permutations above have the same shape.
The print method is sensitive to experimental option
`print_in_length_order` (via function `as.character_cyclist()`).  If
`TRUE`, permutations cycle form will be printed but with the cycles in
increasing length order:

  
```{r, label=showlenord}
options("print_in_length_order" = TRUE)
x
```


## Wrapup

Above we used state abbreviations (a builtin R dataset) and also direct
the print method to use commas for readability.  However, it is best
to reset to the default, as the option persists between vignettes:

```{r,label=reset_print_defaults}
options(perm_set = NULL)
options(comma = NULL)
options(print_word_as_cycle = FALSE)
options("print_in_length_order" = FALSE)
```

## References {-}