GN.Rmd

---
title: "Introduction to Scientific Programming and Simulation 2016"
author: "Gerhard Nachtmann"
date: "`r Sys.Date()`"
output: html_document
---

## Catching up with R: Base R

The code from our catch-up-session can be found in file `2016-10-17.R`,
the transcript including in- and output is here: `2016-10-17.Rt`.

### Data structures

atomic vector (corresponding `NA`)

  + logical (`NA`)
  + integer (`NA_integer_`)
  + double (numeric) (`NA_real_`)
  + character (`NA_character_`)

typeof

list

```{r list}
vector(mode = "list", length = 3)

```


data.frame

trees


`NULL`, `NA`, `NaN`

round

paste

seq_along(letters)

seq_len(4)

gl

http://adv-r.had.co.nz/Data-structures.html

### Subsetting / Indexing

`subset()`

`[]`

`%in%`

### functions

vectorization

imperial2metric

### control structures

if

ifelse vectorized


### loops

Explicit

  + `for`
  + `while`
  

Implicit

`lapply` and friends, which are much faster

inch2cm

paste

```{r loops}
s1 <- function(x){
  res <- 0
  for(i in seq_along(x)){
    res <- res + x[i]
  }
  res
}

s2 <- function(x){
  res <- 0
  i <- 1
  while(i <= length(x)){
    res <- res + x[i]
    i <- i + 1
  }
  res
}

cs1 <- function(x){
  res <- 0
  resv <- vector(mode  = "numeric", length = length(x))
  for(i in seq_along(x)){
    res <- res + x[i]
    resv[i] <- res
  }
  resv
}


```



### Your turn: own function in loop


### tidy data

http://vita.had.co.nz/papers/tidy-data.html

http://blog.rstudio.org/2014/07/22/introducing-tidyr/


## Reproducible research

The source from our reproducibility session can be found in file `Report.Rmd`.
To create an HTML document as output, just use `make html` from the command line
or use `rmarkdown::render('Report.Rmd', 'html_document')` within your R-session.
For the latter ensure, that `Report.Rmd` is in your working directory. Otherwise
you could use the path to this file within `render`, or set the working directory
to the directory, where the source file is stored.
As usual, the transcript including in- and output is here: `2016-10-20.Rt`.

Reader: http://kbroman.org/steps2rr/

Make reports using [knitr](http://yihui.name/knitr).

Formatting with [RMarkdown](http://rmarkdown.rstudio.com/).

Write code and text in one document to produce final output including graphics.

Features:

+ Lists
+ emphasize font
    + **bold** (\*\*)
    + *italics* (\*)
    + ~~strikethrough~~ (\~\~)
+ Links

Links:

https://nicercode.github.io/guides/reports/

https://github.com/adam-p/markdown-here/wiki/Markdown-Cheatsheet

R-Chunks

+ `echo`: show code
+ `eval`: evaluate code
+ `fig`: show figure
+ insert chunk in RStudio: Ctrl+Alt+i
+ inline chunk in backticks with `r`

`purl("document.Rmd")` extracts R-code

Include `ggplot` e.g.
```{r eval = FALSE}
ggplot(iris, aes(x = Sepal.Length, y = Petal.Length, col = Petal.Width)) +
  geom_point() + geom_smooth(method = "lm") + facet_wrap(~Species)
```


## R-packages

[Book by Hadley](http://r-pkgs.had.co.nz/)

library(devtools)

Package structure

### Documentation using roxygen.

https://cran.r-project.org/web/packages/roxygen2/vignettes/roxygen2.html

library(roxygen2)

Tools - Project options - Build tools

Ctrl+Shift+Alt+R

In cpp file:
```
//' @export
// [[Rcpp::export]]
```

In any R file

```
#' @importFrom Rcpp evalCpp
#' @useDynLib <packagename>
```

The R-Package can be built within RStudio or using the commandline (e.g. to implement it in a Makefile):

```{r, engine="bash", RCMD, eval=FALSE}
R CMD build firstpackage
R CMD check firstpackage_0.1.0.tar.gz
```
If you want to make it online available for other users, create a
github repository, then add remote origin to local repository:

```{r, engine="bash", gitremote, eval=FALSE}
git remote add origin https://github.com/nachti/firstpackage.git
```

If you forked my package to your own Github account, then you have to change
the username to be able to add and commit and push changes:

```{r, engine="bash", gitremotownuser, eval=FALSE}
git remote add origin https://github.com/<youruser>/firstpackage.git
git pull origin master # get updates from upstream
git push origin master # put your changes online
```

The package can be found in its own repository here:
https://github.com/nachti/firstpackage

It can be installed using
```{r instfpkg, eval=FALSE}
devtools::install_github("nachti/firstpackage")
```



### data.table

fread

data.table package extends data.frame especially for large data.
It is quite fast due to optimized code.

https://github.com/Rdatatable/data.table/wiki

Binary search in `data.table`

merge using `[`

The code from our data.table session can be found in file `2016-10-20.R`.
As usual, the transcript including in- and output is here: `2016-10-20.Rt`.


### aggregate

with(x, aggregate(v, by = list(group), sum))

## Batch mode, makefile, version control

R CMD build

R CMD check

Reader: http://kbroman.org/minimal_make/

### Git

https://git-scm.com/book/en/v2/Getting-Started-First-Time-Git-Setup

```{r, engine="bash", gitconf, eval=FALSE}
git config --global user.name "John Doe"
git config --global user.email johndoe@example.com
git config --global core.editor emacs
```

http://r-pkgs.had.co.nz/git.html

First clone the Repository:

```{r, engine="bash", git1, eval=FALSE}
git clone https://github.com/nachti/ISPS16.git
git pull origin master # if something changed
```

### use Makefile

Make documentation using [pandoc](http://pandoc.org/):
```{r, engine="bash", make1, eval=FALSE}
make html # just produce html
make # produces pdf, html, docx
```

Make pdf etc. from Report.Rmd:
```{r, engine="bash", makereport, eval=FALSE}
make FILE=Report
```

### Rscript

Create R Script `i2cm.R` and tell the system, where Rscript is living:

```{r Rscript, eval=FALSE}
#!/usr/bin/Rscript
### first line tells your system where *Rscript* is located
library(firstpackage)
example("inch2cm")
```

Make it executable for everyone using `chmod 755 i2cm.R` and use it in dfferent ways:

```{r, engine="bash", rscriptexamples, eval=FALSE}
Rscript i2cm.R # works even if i2cm.R is not made executable
./i2cm.R # send output to console
R CMD BATCH i2cm.R # send input and output to file i2cm.Rout
nohup ./i2cm.R & # send output to nohup.out
nohup ./i2cm.R > log.out & # send output to log.out
```

If you use `nohup ./i2cm.R > log.out &` to run your script, it's possible to logoff the system while the calculations are done.

### Hints for Mac users

Ensure that pandoc is installed.
Possibly R, pandoc etc. are not exported, so
create the file `.bash_profile ` in your home directory:

```{r, engine="bash", mac, eval=FALSE}
touch ~/.bash_profile 
```

Open it using any texteditor and add the following lines:

```{r, engine="bash", mac_bp, eval=FALSE}
export PATH=$PATH:/Library/Frameworks/R.framework/Resources/bin
export PATH=$PATH:/Applications/RStudio.app/Contents/MacOS/pandoc
```

## shiny

You can find the code of our shiny session in the `shiny` folder in this repository.

### some links

http://shiny.rstudio.com/

http://shiny.rstudio.com/tutorial/

[Shiny tutorial by Dean Attall](http://deanattali.com/blog/building-shiny-apps-tutorial/)

[Shiny tutorial by Paul Hiemstra](http://stcorp.nl/R_course/tutorial_shiny.html)

https://github.com/rstudio/shiny-examples

http://shiny.rstudio.com/gallery/

https://www.r-bloggers.com/building-shiny-apps-an-interactive-tutorial/

[drinkR](https://gallery.shinyapps.io/drinkr/)

[Shiny Cheat Sheet](https://www.rstudio.com/wp-content/uploads/2016/01/shiny-cheatsheet.pdf)

### ui

sidebarPanel

  + selectInput
  + checkboxInput
  + radioButtons
  + sliderInput

### server

R-code for output