Sorting algorithm visualisation (nice)
sortvis.org
sortvis.org
Also, every now and again I run into something I haven't seen before and which is of value. I use this, then, as a means of adding value by cross-referencing, but also of gaining value to me by a directed random search.
The traces of mergesort, quicksort and so on are quit distinctive.
I could not find my old program, so I guess I will have to write a new one. Should be easy enough.
If you are, say, analysing run-times, you wont get very far with this choice of presentation.
This situation is all over the undergrad academia. For a math major, numerical method algorithms described in advanced notation is ok, as long as you are fluent in it. But when I take CS courses and see the same presentation, I get very confused. For god's sake, these people are going to spend more time on Novell manuals, than on the graph theory. I could have learnt all these algorithms, if I didn't need to chew through all that theoretical packaging and generalization. Instead I spend all my time to dig through precise definitions just to get one or two concepts, when I could have gotten them all, if representation was good enough.
I don't think that CS undergrads don't need precision offered by scientific approach. I think there comes a time when you feel the need for precision. But this time won't come when you only know quicksort in java. You have to gain a broad specter of examples before you can set off to learn about big O analysis. Because this kind of analysis is about comparison. Learning is ineffective when you compare one algorithm with a blurred idea of another.
I might sound like a confused student, but what I'm trying to say is that there should be one unified, simple and consistent source of algorithms in understandable languages. I have the large algorithm encyclopedia from Springer Verlag, and it polluted with all uncertainties of the modern CS studies. It shows us the perspective fields of study, but it's too abundant to explain simple ideas. Same thing is about Wikipedia. It tries to be both precise (because experts write articles) and simple (because people like me read them) - but this is like running in opposite directions simultaneously.
Somewhat related: I am quite fond of Chris Okasaki's "Purely Functional Data Structures" and "Concrete Math" by R. Graham, D. Knuth, and O. Patashnik.
But I have almost the exact opposite qualm about most books from what you air: I can't have enough precision. I agree that there can be too many technical details, but there can't be too much precision.
If CS undergrads can't take it, they deserve to suffer. There's always vocational training for people who want a focus on applications. Try to become e.g. a Fachinformatiker (http://de.wikipedia.org/wiki/Fachinformatiker).
P.S. I guess I am a bit too harsh. I agree with most of your points.
> I could have learnt all these algorithms, if I didn't need to chew through all that theoretical packaging and generalization. Instead I spend all my time to dig through precise definitions just to get one or two concepts, when I could have gotten them all, if representation was good enough.
Precise definitions can help your understanding. I hope you will some day get a professor who uses them to good effect.
> Learning is ineffective when you compare one algorithm with a blurred idea of another.
Isn't this a good argument to learn many algorithms with precision?
> I might sound like a confused student, but what I'm trying to say is that there should be one unified, simple and consistent source of algorithms in understandable languages.
Go, write one, please!
> I have the large algorithm encyclopedia from Springer Verlag, and it polluted with all uncertainties of the modern CS studies.
Are you talking about "The Algorithm Design Manual" from Skiena? I co-worker brought it to the office recently. I can't bring myself to like it. It seems there are too many references and not enough meat. I have fonder memories of Sedgewick's "Algorithms". But that may be nostalgia. (Get the version in C or the original Pascal.)
Feel free to ask me about algorithms or data structures, my email address is in my profile. I enjoy talking about algorithms way too much for my own good.
Pretty patterns do indeed emerge from these visualizations, and it's definitely a nice piece of work from the author/artist. But let's not pretend it's a useful algorithm teaching tool.