How to use LaTeX Code in R Chunk in R-Markdown?

Question

I am currently writing on a report with rmarkdown and therefore I want to create sections inside a r code chunk. I figured out that this is possible with the help of cat() and results="asis". My problem with this solution is, that my R code results and code isn't properly displayed as usual.

For example

---
title: "test"
output: pdf_document
---

```{r, results='asis'}
for (i in 1:10) {
  cat("\\section{Part:", i, "}")
  summary(X)
  $\alpha = `r X[1,i]`$  
}
```

pretty much does the trick, but here there are still two problems:

• the R output for summary() is displayed very strange because I guess it`s interpreted as LaTeX code
• I can't use LaTeX formulas in this enviroment, so if I want every section to end with a equation, which also might use a R variable, this is not possible

Does somebody know a solution for some of these problems, or is there even a workaround to create sections within a loop and to have R code, R output and LaTeX formulas in this section? Or maybe at least one of those things?

I am very thankful for every kind of advice

Solution

You can do what you are after inline without relying as much on code blocks.

As a minimal example.

---
title: "test"
output: pdf_document
---

```{r sect1_prep, include=FALSE}
i <- 1
```

\section{`r paste0("Part: ", i)`}

```{r sect1_body}
summary(mtcars[, i])
```

$\alpha = `r mtcars[1, i]`$

```{r sect2_prep, include=FALSE}
i <- i + 1
```

\section{`r paste0("Part: ", i)`}

```{r sect2_body}
summary(mtcars[, i])
```

$\alpha = `r mtcars[1, i]`$

Produces...

If you really want to have a section factory, you could consider pander.

---
title: "test"
output: pdf_document
---

```{r setup, include=FALSE}
library(pander)
panderOptions('knitr.auto.asis', FALSE)
```

```{r, results='asis', echo=FALSE}

empty <- lapply(1:10, function(x) {

  pandoc.header(paste0("Part: ", x), level = 2)
  pander(summary(mtcars[, x]))
  pander(paste0("$\\alpha = ", mtcars[1, x], "$\n"))

})

```

which produces...

remove summary table format example

---
title: "test"
output: pdf_document
---

```{r setup, include=FALSE}
library(pander)
panderOptions('knitr.auto.asis', FALSE)
```

```{r, results='asis', echo=FALSE}

content <- lapply(1:10, function(x) {

  head <- pandoc.header.return(paste0("Part: ", x), level = 2)
  body1 <- pandoc.verbatim.return(attr(summary(mtcars[, x]), "names"))
  body2 <- pandoc.verbatim.return(summary(mtcars[, x]))
  eqn <- pander_return(paste0("$\\alpha = ", mtcars[1, x], "$"))

  return(list(head = head, body1 = body1, body2 = body2, eqn = eqn))

})

writeLines(unlist(content), sep = "\n\n")
```

Comments

[Quastiat]

Thanks, the first approach is actually what I was trying to avoid, because my code is way longer and i runs from 1 to 13. But the approach with pander seems promising I haven't considered that yet, but I will have a deeper look into it now :) Thanks!

[Quastiat]

Thanks for the tip, that's already an improval, but what doensn't seem to work there is to have a regular R output not beeing displayed in a table right? Like for a linear regression pander always tries to put the content into a table is there a way to trick pander somehow, to get a regular R output?

[Kevin Arseneau]

@Quastiat, you need to read the package documentation for pander. It will output based on the object class, you'll note the equation output simply as text. It is up to you to format as suits your requirements, you could substitute pander for pandoc.verbatim or anything else that works for you if you don't like the standard output.

[Quastiat]

alright thank I will have a look thanks a lot, that really helps me.