I thought it would be fun to test my R skills by doing Project Euler. Modern programming languages have rather full-featured libraries, but it’s still nice to learn (or re-learn) things from first principles, or see what others come up with. You can see the results in this GitHub repository: euler-r.

The lessons I learned so far are:

• Project Euler is great for stretching your programming muscles. Specifically, it forces you to really understand data structures and classic algorithms by implementing them. You also learn the strengths of the language you’re using (at the cost of also exposing its weaknesses).

• Libraries are wonderful: many Project Euler problems involve handling large integers, big matrices, or dates. It’s tedious to re-invent low-level algorithms for common tasks. This point complements the point above.

• A good way to learn is to write everything 3 times: first to write code that obstensively works, second to get it to work correctly, third to get it to perform reasonably well.

• Brute-force solutions have their place. Computers are fast now, and sometimes code legibility is worth the extra CPU cycles (ie. that old adage about premature optimization).

• On the other hand, since many Project Euler problems are rooted in math, there exist elegant (or at least more efficient) solutions if you’re familiar with, for example, binomial theorem or the properties of prime numbers. This leads me to my last point:

• We stand on the shoulder of giants: the body of knowledge in math is vast. It’s humbling to retrace the steps of pioneers.

After doing a few problems in R, I reimplimented them in Python: euler-python. I’m enjoying seeing the subtle (and not so subtle) ways the two languages differ. I like R but found its TIMTOWTDI approach to be bewildering at times (eg. do we really need 3 different object systems ?). Lessons learned for Python so far:

• List comprehensions are nice syntactic sugar.

• The with statement is nice, especially in long pipelines where you otherwise would have to do a lot of cleaning up.

• As was the case in R (or any language), it’s good to be familiar with your libraries. For project Euler in particular, math is very helpful.

Update: As I grew more comfortable with Python, more code snippets were added to python-snippets.