compTools 2026

 

Cover image of the moodle page of compTools 2026, showing the convergence of a GA to the global minimum of a multimodal function.

 

one of the problems in life is that many things seemingly flow away without leaving any trace, literally slipping among our fingers, as we immediately have to take care of other commitments, other dreams, other loved persons and so on. and, yet, we know that reflection is useful as it leads to understanding and, among other things, consolidates and sharpens what we have done or learnt. think, do, pause and re-think to what you did. or, in a more yogic way, inspire, expire, ignite your fire!

hence, i decided to remember CompTools 2026 with a post, it is the first time i do this even though i taught this material, in one version or another, for more than a dozen years. this delivery, perhaps, was different, and now that exams are over, i can unleash my personal side and write unconventional thoughts that have nothing to do with R and all that jazz.

The course was fun, i'll remember some of you because of the questions you asked in class, the way you attended the lectures but also for some nods and spectacular gazes. i'll also remember some of the discussion in the forum, much deeper and worth-reading than many others seen in the past, with a touch of vibe and an almost existential meaning. thank you. Even though some of the discussions were most likely polished using AI, i believe they were good and, in any case, too insightful to be purely produced by AI. incidentally, this post is 100% AI free, i decided to honor my brain and write myself, resisting the temptation to ask gemini to decorate my writing or improve my vocabulary. i decided to write the old way, only used the dictionary to pick some synonym and be sure i was hitting the right significance.

The workshop was also an extremely interesting experience. you know, i believe that one professor's duty is to force you to work hard, because this is the way we learn. ideally, i should push you to the limit, spurring you to learn as much as it is possible and stretching a bit you limits, pushing you outside of some comfort zone. the problem is that, at the same time, i do not want to overwhelm students and seed despair in their minds, "i'll never be able to do that". the workshops were a trick, primarily, to keep as many as i could "attached" to the course, they were excuses to give you some extra points, provided that you were doing some intermediate works, good, fair or even bad. but one more important thing was to keep you focused. then, some of the tasks were, i hope, stimulating and able to bridge different topics (tying, say, constrOptim with GA and simulation, in a comparison that sheds light on the methods, as well as on the ideas and their computational efficiency). The workshops were also a gorgeous way to force you to have a look at the work some of your peers can produce. You had to look three or four of them and grade, but i hope this was also a way to think "wow, nice! i hadn't though about this and next time i'll do better..." it was good and rewarding to see many of you absorbing so much from the material, mastering a blend of R, "programming", maths, new ideas, new words, and new problems. 

c'mon, was the course really perfect? No way! it was also painful to see, for instance, that only 40 or 45 students attended (out of 130 matriculated ones). but, again, video recordings were perhaps helping the absent ones, the forum was there to flood the world with thoughts and summaries (good or bad...), the sessions allowed everyone to replicate the lectures. i still have the bitter feeling that, in a few cases, some students deserved an even better grade, grant me an amnesty if needed, as grading is an awful job (and Wawrinka once famously said "Next time i'll fail better"). further, i decided that i should not care if some of you just carelessly graded the workshops giving the top mark without even reading or giving zero feedback. in contrast, some of you gave other students detailed remarks and, clearly, read very carefully their work. i decided that i should not care about the one who cheated during the exam and used AI to dupe, because most of you proved during the oral exams to be fully able to explain the tools they used. (as an aside, i'll ask the cheater to be suspended for 6 months, c'est la vie but this is another story...) i just would like to invite the few ones who did the bare minimum to think that they were losing an opportunity, i would like you to be brave enough to think that you can do things that you cannot even immagine right now. if you dream more and study more, you will reach new horizons, not a simple passing grade! 

ok, chaps, time to end this undecipherable post (or, perhaps, it's crystal clear!) i think compTools 2026 was a good one, AI was helpful to many of you and, undoubtedly, raised my and your understanding, and the level of our discussions, the material we produced was abundant, the workshops were fun and wish you to progress on the mesmerizing paths of life and computational sciences. respectfully, paolo

Minimize f <- function(x,y) (x-y+2)^2+(y-2)^2 on the feasible sets depicted here ;-) (using constrOptim, GAs and simulation)