Hacker News new | past | comments | ask | show | jobs | submit login
Essentials of Metaheuristics (2015) [pdf] (gmu.edu)
43 points by mindcrime 4 months ago | hide | past | web | favorite | 7 comments



Parent page with metadata: https://cs.gmu.edu/~sean/book/metaheuristics/

(Please don't link to PDFs where a sensible parent html page exists. We can click from html to pdf but not vice-versa.)


Wow, my book got posted to HN.


I feel like your book gets posted to HN every other year. My girlfriend is using it for one of her graduate OR classes at GMU.

Edit: I think they may be using a different metaheuristics text.


I bought a physical copy of this book through Lulu about a year ago; it presents some really neat algorithms and ideas. Unfortunately I haven't had the opportunity to apply any of them to any real problems. That said, they seem like the kind of algorithms that, by having them in your back-pocket, would allow you to solve seemingly unsolvable problems.

I've been thinking about applying some of the algorithms to procedural content generation for a toy game that I've been working on. Specifically tweaking knobs and evaluating quality.


If you find this area fascinating, and I do, I can't recommend this cousera course enough: https://www.coursera.org/learn/discrete-optimization


+1 Fantastic course.


> That said, they seem like the kind of algorithms that, by having them in your back-pocket, would allow you to solve seemingly unsolvable problems.

Hmm... not sure. "Solving seemingly unsolvable problems" might be overselling these algorithms a tad.

Metaheuristics are a collection of algorithms for solving various types of optimization/constraint-satisfaction problems via a process of educated trial-and-error (which is essentially what heuristics are). Typically they are not mathematically sophisticated but they get you answers that are good enough for most practical purposes.

However, it's hard to say one metaheuristic algorithm is better than another. There's a huge element of luck involved.

p.s. Ironically, they can succeed where sophisticated optimization algorithms fail, but there's no guarantee. They are definitely very practical though.

The analogy in my head is this: I think of a sophisticated algorithm as one MIT-trained engineer trying to solve a problem (where the goals are clear) in a methodical way, whereas metaheuristics as maybe 1000 Joe Blows meandering around the problem in some sort of orchestrated way. The chances are good that one of them Joe Blows gets lucky.




Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: