Chris is starting a blog in 2017


A quick Google search turns up articles from as early as 2012 insisting that blogging is a dead medium. While this premonition didn’t turn out to be completly true, I think it is fair to say blogging isn’t in its heyday.1 So, why am I starting a blog now?

###Unavailability of low-stakes writing in academia

In my day-to-day life, I do a fair amount of writing, but it’s a very specific type, mostly comprising of technical papers and grant proposals. In both of these contexts, I’m trying to sell an idea. In doing so, this doesn’t leave a lot of room for writing about expository thoughts that aren’t fully fleshed out or are only tangentially relevant. This blog will be my place to go down these rabbit holes.

###Documenting non-research ideas

Currently, whenever I find some exciting mathematical result that slipped through the cracks in my education, I race to alert my officemates and force them to listen as I go through an excruciatingly long explanation on our white board. This blog will (hopefully) spare them of those occurrences!

#Who is this blog written for?

Nobody. Or really, myself. I don’t expect anyone to read or care about anything I post. That said, I wouldn’t be too upset if anyone happens to stumble in here.

#You changed your website again? What heck is Hugo?2

My previous academic personal site was made from a modified an HTML template, such as the beautiful ones found on HTML5Up. I’ve seen a lot of academics go this route recently, and while it was working fine (aside from being a bit annoying to edit), it lacked a critical feature for me to start my blog: a blog.

I had initially considered Jekyll, which I’ve seen a few academics use, but the internet (reddit) contends that Hugo is the modern choice, citing its blazing speed generating websites in 1 ms in stead of 10 ms. This didn’t really matter to me, but a key other ingredient sold me on Hugo: blogdown.

The pipeline with blogdown and Hugo seems ideal for blogs with heavy mathematical and programming content. Posts and pages are written in Rmarkdown (basically just vanilla Markdown with R code evaluation) and then converted to an appropriate format by blogdown, and ultimately pushed to the website using Hugo.

###What sort of fancy things can you do that made all this effort worth it?

One requirement I had was LaTeX support, which is straightforward with Hugo: \[ \mathrm{d}X_t = \mu X_t \mathrm{d} t + \sigma X_t \mathrm{d} W_t. \]

Because the posts are written in Rmarkdown, R code embeds seamlessly, including plots

pie(c(Sky = 78, "Sunny side of pyramid" = 17, "Shady side of pyramid" = 5),
    init.angle = 315, col = c("deepskyblue", "yellow", "yellow3"), border = FALSE)

While possible to make R’s plots look pretty, I was hoping to make the plots interactive in some way, which it turns out is exactly what htmlwidgets and the usfeul R package widgetframes provides. With htmlwidgets, I can now use a bunch of nice R libraries like plotly and leaflet.

For example, here is an interactive Plotly plot


g <- gapminder
pp <- plot_ly(x = g$gdpPercap, y = g$lifeExp, size = g$pop, color = g$continent, 
    frame = g$year, text = g$country, hoverinfo = "text", type = "scatter", 
    mode = "markers") %>% layout(xaxis = list(type = "log", title = "$ GDP/capita"), 
    yaxis = list(title = "life expect"))


and Leaflet


m <- leaflet() %>%
addTiles() %>%  # Add default OpenStreetMap map tiles
addMarkers(lng=174.768, lat=-36.852, popup="The birthplace of R")

These are far cry away in fanciness from posts by the New York Times, but I’m pretty happy with the capabilities. I’m now out of excuses to avoid adding actual content to the blog.

  1. but live-tweeting every thought might be?

  2. actual quote, from my officemate

comments powered by Disqus