Ask HN: After SICP, what next? | Hacker News Reader

Ask HN: After SICP, what next?

78 pointsplinkplonk35 comments

One of my friends just finished working through SICP (progress here - http://lawfulsamurai.blogspot.com/search/label/SICP, - the last few sections still have to be updated) and asked me for reccomendations for future work.

I thought of telling him to get really strong on algorithms etc (perhaps by working through Tardos/ Cormen et al, doing all the exercises like he did for SICP), but thought I'd tap the good folk of HN for better (or different) suggestions. Working through CTM for example is an option (another of my friends was doing this http://ctm-himanshu.blogspot.com/ but he didn't quite complete it).

(This has nothing to do with his day job which is java/python/Django. Just something for him to work through when he has a few hours and learn cool things.). Any suggestions gratefully accepted!

Thanks in advance.

View on HN

This is the single best text I've seen on algorithms in the real-world: http://www.amazon.com/Algorithm-Design-Manual-Steve-Skiena/d...

Can't recommend it enough.

Here's the second edition from 2008: http://www.amazon.com/Algorithm-Design-Manual-Steven-Skiena/...

Thanks! I did not know there was the 2nd edition!

I just bought this book and have been reading it for a few days and it really is pretty good. The best thing about it are the numerous practical examples and the "War Stories" that show you how real people determined they really needed to use algorithm xyz.

Introduction to Information Retrieval: http://www.amazon.com/dp/0521865719/

Convex Optimization: http://www.amazon.com/dp/0521833787/

Foundations of Statistical Natural Language Processing: http://www.amazon.com/dp/0262133601/

Read Peter Norvig's review for Foundations of ....http://www.amazon.com/review/R3GSYXSKRU8V17/

I haven't read any of these books, yet, highly recommended by some friends.

Here are video lectures by the author of the convex optimization book (Stephen Boyd):

http://see.stanford.edu/see/courseinfo.aspx?coll=2db7ced4-39...

http://see.stanford.edu/see/courseinfo.aspx?coll=523bbab2-dc...

And video lectures by the author of the NLP book (Christopher D. Manning):

http://see.stanford.edu/see/courseinfo.aspx?coll=63480b48-88...

I recommend the Kleinberg/Tardos textbook: http://www.amazon.com/Algorithm-Design-Jon-Kleinberg/dp/0321...

It is a really enjoyable read and has a nice narrative that I think other algorithm books are lacking. CLR, for instance, just reads like a handbook to me. The goal Kleinberg/Tardos book, OTOH, is to teach you how to design and analyze algorithms. They will actually follow false starts on certain problems and uncover where they break.

Kleinberg is the rebel king!

The best book to follow up SICP is Essentials of Programming Languages, by Friedman, Wand, and Haynes. It uses Scheme to show how the fundamentals of programming languages are built, and both in its form and its content, it seems a direct continuation from SICP. After you've done with it, you'll have a much better understanding of Scheme, functional programming, programming languages, and programming in general.

- Concepts, Techniques and Models of Computer Programming

- Algorithms, a functonal approach

- Real World Haskell

- Paradigms of artificial intelligence programming

Any of these are excellent post-sicp reads.

I don't suppose you have a copy of "Algorithms, a functonal approach" you want to sell huh? The used price is more than the brand new price! Lame.

I bought it new on Amazon.co.uk for a very reasonable price. It was printed on demand, so it took some time, but it was worth it.

Did he enjoy sections 4 (Metalingustic Abstractions) and 5 (Register Machines)?

If so, he'd probably be interested in both these books. They are quite hefty, but seem to be goldmines (so far, I'm only just getting started on them myself).

http://research.microsoft.com/en-us/um/people/simonpj/papers...

and

http://research.microsoft.com/en-us/um/people/simonpj/papers...

Both books cover much of the same ground. The first one is heavier on theory, while the second one reads a LOT like SICP. Currently I'm working through the second one and using the first one when I want more in-depth info on a certain topic.

Lisp In Small Pieces (http://pagesperso-systeme.lip6.fr/Christian.Queinnec/WWW/LiS...) is sort of intermediate between SICP and the Peyton-Jones stuff; it still focuses on Lisp dialects that use dynamic typing and applicative-order evaluation, but it shows a whole bunch of different approaches to exploring that design space and implementing stuff in it.

I also enjoyed reading Pierce's Types and Programming Languages (http://www.cis.upenn.edu/~bcpierce/tapl/), although I was kind of nodding my head and going "yeah, whatever" during some of the proofs.

Both recommendations are little too high for someone just coming out of SICP. I recommend Essentials of Programming Languages as a stepping stone. Pierce's book is probably irrelevant to someone just interested in Lisp; type theory is best approached through an ML angle. And Quinnec's book is not for the faint of heart; you will need to have been well versed in lisp, compiler/runtime implementation techniques and have brushed against denotational semantics (through the Nielson and Nielson book, perhaps.)

[edit: denotational, not operational, semantics; OS can be learned in a fun-filled weekend. The "small-step" variety should appeal to both language hackers and (virtual) machine enthusiasts.]

If your friend wants to a) get really strong on algorithms, b) put in some hours, and c) actually do the exercises, I'd recommend Knuth. It's worth the effort.

I haven't read Knuth but I recommend http://www.amazon.com/Algorithm-Design-Manual-Steve-Skiena/d...

From what I understand Knuth uses MIX assembly language to imlpement his algorithms. Contrast this to the Skiena book which uses C++. I found this to make it very practical.

Though to be fair, I don't think it offers near the depth of the Art of Computer Programming:)

It's actually a common misconception that Knuth uses MIX/MMIX to implement his algorithms. The vast majority of the algorithms are described in English, and the MMIX implementation is usually only given when there are relevant implementation details to be discussed. MMIX is just an idealized machine language, and it is trivial for a reader to implement the algorithms described in the programmer's language of choice. (This is assuming that the objective is to learn CS, not to have a cookbook of ready-to-run algorithms to pilfer....)

YES! Thank you for this clarification.

I personally enjoy thinking about the implementations, instead of looking at the code itself. Knuth makes that easy (hard?) by not spoon feeding you.

Plus the typography of Knuth's books is so wonderful. It's a pleasure to read.

I'll also add that I usually recommend that newcomers begin with Volume 4, and turn to the earlier ones later. I think that for many people, skipping the preliminaries, and starting with an interesting and relevant topic (combinatorial algorithms) gives more immediate pleasure than beginning with the lower-level underpinnings.

Readers might also consider Knuth's "Concrete Mathematics" as a substitute for the more-or-less equivalent material in Volume 1, especially if a little more help is needed in understanding it.

Sorry if this seems dense but I can't reconcile what your saying and I've not read the book. You say:

> It's actually a common misconception that Knuth uses MIX/MMIX to implement his algorithms.

So you state very clearly that he doesn't use MIX to code his algorithms.

But then you say:

> and the MMIX implementation is usually only given when there are relevant implementation details to be discussed

So you say that he does actually use the MIX language.

Do you mean that most algorithms aren't presented in source form at all and only the few that are actually coded are in the MIX language?

Yes, that is how it is.

For example the first algorithm in the book (found via google):

Algorithm E ( Euclid's algorithm). Given two positive integers m and n, find their greatest common divisor, that is, the largest positive integer that evenly divides both m and n.

E1 [Find remainder.] Divide m by n and let r be the remainder. (We will have 0 ≤ r < n)

E2 [Is it zero?] If r=0, the algorithm terminates: n is the answer.

E3 [Reduce.] Set m← n, n← r, and go back to step E1.

Thanks for the clarification:)

You can run MIX programs using the GNU MIX Development Kit:

http://www.gnu.org/software/mdk/

I'm personally going to work through Programming Language Pragmatics. I want to have a better foundation in language paradigms, plus fill a hole in my education (no, zero, nada, real exposure to compiler design).

I'll also be learning more about Lisp concurrently.

Sadly, none of this has much bearing to my day job, but when I have an hour or two it seems like a nice way to spend it. At that pace, I plan to finish sometime in 2015 . . .

PLP is pretty good. We used it in my university course on compiler design, I quite enjoyed reading it.

It depends on what he wants to learn.

I got more out of How to Design Programs than I did from SCIP. Though to be fair I read it before SCIP:)

http://www.htdp.org/

On the other hand if he wants to learn Lisp then I really enjoyed Lisp in Small Pieces though its not a quick read. It consumed almost all of 2005 for me:)

Artificial Intelligence: A Modern Approach http://aima.cs.berkeley.edu/ -- has lots of broadly useful material and ought to be accessible to anyone who's completed SICP. One of the very best technical books I've read.

The CLR "Introduction to Algorithms" would be one of the canonical answers.

http://en.wikipedia.org/wiki/Introduction_to_Algorithms

It's certainly one worth having and the only book from my computer science education that I still reference frequently.

In fact, I'm tempted to say that it's the last computer science book that you have to read. After that you should be able to work directly from research papers, though a topical book can sometimes be useful in other areas with fairly wide breadth.

He should write a lot of code for fun and try to incorporate what he's learned so far. I understand he codes at his day job, but that's not a place to take risks or to code things 5 different ways "just because".

"Problems On Algorithms" (Parberry, Gasarch): A slim volume containing ~700 exercises on proving various properties of algorithms. And it teaches you most of what is needed. Download it from: http://www.eng.unt.edu/ian/books/free/

"Elements of the Theory of Computation" (Papadimitriou): Classic text. Sure you've heard of Turing machines before, but what do you know about mu-recursive functions? It's very beautiful and if you have a person with whom to read the book and do the exercises, you will gain immensely.

I'm currently working through Practical Foundations of Mathematics: http://www.cs.man.ac.uk/~pt/Practical_Foundations/index.html

The full text is available online, but I recommend you buy a hard copy. Oh and I can vouch that the back cover blurb is totally true.

CTM has my vote.

Go be the smartest person at a Java shop!

I like CTM. http://www.amazon.com/Concepts-Techniques-Models-Computer-Pr...

Reply on news.ycombinator.com