Graduate Students: Start a Tea Time at Your institution

At @jhubiostat, we have a tea time every week the day after our weekly seminar.  This is a tradition that started before my time (which is a bit of time) with @kwbroman.  It is one of my favorite events in our department.

We have faculty and students come, have tea and baked goods (baked by them!) and discuss things.  The formal purpose is to discuss the seminar from the day before.  Such things discussed are methods, slide choice, table presentation (if any), too many formulas, speed, applied dataset of stat method (or lack thereof).  Most discussions are positive and fun, and each person gives a good point of view.

Many times though, we discuss the seminar for a short period of time, and then it becomes a more social discussion divulging into: cool articles, Twitter points, discussion of our department’s social media, etc.  

So my overall message to other graduate students: start a tea time.

Here are some qualities that are good for a room choice:

  1. Be a place where people normally walk by.  No one wants to walk a mile to grab some tea.
  2. Have a window in the room – or at least open door.  Many times people walk by and look in and think – “Hey, what’s going on?”
  3. Publicize it on list-servs for your institution
  4. Ask your department for a small amount of funds to buy tea/big kettle.  It’s worth it.
  5. Bring delicious treats.

Overall, new (and more veteran) students get to meet interact with faculty that they would not otherwise.  Also – faculty get to present some topics to students without the formality of a faculty meeting or a departmental list.   It boosts morale and you usually can connect to people in much more interesting ways than just being in an office/on a computer.

Viewing PDFs on cluster nodes

We have a cluster.  We are recommended to use cluster nodes instead of the portal computer to do any computation or anything above light work (i.e. text editing).  I recently wanted to view pdf files (output from an analysis) on a node.  Some of these can be big files, and so I don’t want to do that on our portal computer.  

I downloaded the linux xpdf binary from (, under pre-compiled binaries), untarred it, then installed it into a subdirectory of my home directory (we do not have root access or access to /usr/* or /opt/* so I installed it to ~/programs/xpdfbin-linux-3.03/.  Then editing my ~/.bash_profile, I added 


and had the ability to do 

xpdf mypdf.pdf

to view a pdf on a cluster node!  NOTE – some systems may fail using the bin64 case, as did our cluster.  

ALSO – now I can view R Vignettes directly on the cluster instead of toggling back to local.

Hope that helps.

SizeUp and Mavericks

So after recommendation of @acfrazee, I started using SizeUp, and it changed my life.  But after upgrading Mavericks, I couldn’t get SizeUp 1.5 to work.  In order to get it to work, you need to go to System Preferences -> Security and Privacy -> Accessibility -> (unlock if necessary) and then make SizeUp checked.  This is different than on Mountain Lion for settings.  But after upgrading to Sizeup 1.6, all worked fine. Just FYI.

Mavericks and RStudio

Mavericks and RStudio

I just upgraded to OSX Mavericks and RStudio Preview. I like the RStudio Presentation mode and saw that they added a new R-HTML type document (along with a pres document). I’m a big fan of Shiny, but thought that the live preview in the plots page was pretty slick.

Now, with Mavericks, my X11 went crazy. I had just installed XQuartz when upgrading to Mountain Lion, but when Knitting, I was having it come up with X11 problems, and X11 windows would not load when I was working on my cluster interactive sessions. I Googled around for an X11 solution. After re-installing XQuartz, it’s all good on the X11 front.

I also found that RStudio is slow. After installing the “new” RStudio, it pretty much exactly became a Glimmer or Spark Session where you’re interacting with all menus from RStudio and nothing from the Mac toolbar (which I knew going into downloading). It’s not lagging as it was with the previous RStudio version, so if you upgraded to Mavericks, download immediately. If not, and use R regularly (especially for things like making posts with knitr!), I’d hold off on the Mavericks update until RStudio fixes the issues.