Knuth's Art of Computer Programming, V 4B, has gone into print • Hacker News Board

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#181

Post by aamargulies » Tue, Oct 04, 2022, 9:39 PM UTC

Many years ago I needed to understand hashing for an important work project. I sat down and read Knuth's chapter on hashing. It was very, very good and it helped to greatly accelerate my project work.

However.

What I also learned is how impossibly large a topic even just hashing is. A section of Knuth's encyclopedia isn't sufficient space to cover it thoroughly (for some definition of thoroughly) and further research showed how much progress had been made in the subject since Knuth published his work.

This isn't a complaint about Knuth, again, I'm grateful for how much his writing accelerated my understanding. But further research underlined how impossible a task he's undertaken.

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#182

Post by cromwellian » Tue, Oct 04, 2022, 9:41 PM UTC

A few years ago I had the pleasure of meeting Knuth in his home, my wife was doing some photography for him. He took me to his room and showed me his setup, he was researching Soduku algorithms at the time. His hands were blisteringly quick, moving between EMacs panes, triggering evaluations and printing of results, as fast as any 20 year old. In his 80s , he hasn’t seemed to suffer any mental decline.

I started talking to him about some of the latest AI research and he mentioned the papers authors, he had already read them! Not only is he still very productive at 84, but he has not ossified into one particular subject but he continues to keep abreast of other related fields.

He also played his huge pipe organ for us, he was gracious, humble, down to earth, truly a treasure. I only wish he could live another hundred years, selfishly so I could see Volumes 5, 6, and 7 of Art of Computer Programming finished.

I should not with some sadness that COVID hit not long after and that year we lost a number of other luminaries in the field like Jon Conway.

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#183

Post by reidjs » Tue, Oct 04, 2022, 9:43 PM UTC

xwolfi wrote: post #177

svat wrote: post #131

Note that, to get an idea of the contents, you can see the previously published fascicles, and earlier drafts online (pre-fascicles): • Mathematical Preliminaries Redux: https://cs.stanford.edu/~knuth/fasc5a.ps.gz • 7.2.2 Introduction to Backtracking: https://cs.stanford.edu/~knuth/fasc5b.ps.gz • 7.2.2.1 Dancing Links: https://cs.stanford.edu/~knuth/fasc5c.ps.gz • 7.2.2.2 Satisfiability: https://cs.stanford.edu/~knut…

[flagged]

He is writing programs to help him publish a volume for one of the most important series on computer programming. we have different definitions of the word productive.

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#184

Post by linschn » Tue, Oct 04, 2022, 9:48 PM UTC

lupire wrote: post #148

linschn wrote: post #46

Earlier quoted context omitted.

A colleague of mine had invented a technique for factoring composite numbers, that he thought would be quite fast. I thought a little about it and saw that it would degenerate into sat solving. I checked taocp and indeed my colleague's idea was in there. Also a faster algorithm was presented in the relevant section. I lended the book to him and we saved around a month of work.

There is quite a lot of money waiting for you if you ever figure out how to factor composites quickly. (Any old bum can factor primes quickly.)

We had a very specific use case, where the input was a known composite, and some properties of the factors were known.

Also, while no fast algorithms are known (and may never be depending on if p=np) some are faster than others depending on the properties of your inputs, and choosing the right one can mean saving a lot of time and effort.

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#185

Post by jupp0r » Tue, Oct 04, 2022, 9:58 PM UTC

rietta wrote: post #69

Very cool. I own the originals and the 4th but sadly I have to admit that even though I am a reasonably accomplished developer with a couple of CS degrees I have a hard time understanding this work and have not yet really read it. I feel like I am publicly admitting failure just posting this. I still have time to make sure these are not just bookshelf decorations though.

I always viewed them as more of a reference where I could read up how to find and implement somewhat specialized algorithms. I really lack the discipline/interest to read those from front to back, so you are not alone. Incredible books though!

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#186

Post by aasasd » Tue, Oct 04, 2022, 10:05 PM UTC

Since the pipe organ will be inevitably mentioned in the comments, I'm gonna plug the Youtube channel ‘Look Mum No Computer’, whose gear-crazy author in now apparently in the final stages of installing and patching up an organ in his seemingly interminable basement, after moving it from someone else's house: https://youtube.com/shorts/XGL3rskcQD0

Said organ was built by its late owner not just in the house, but pretty much into the walls of the house, and as a result the thing had to be ripped out of there rather grotesquely, and then rewired again at the new place: https://youtube.com/watch?v=8PwwRR8deHk

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#187

Post by gameman144 » Tue, Oct 04, 2022, 10:07 PM UTC

xwolfi wrote: post #177

svat wrote: post #131

Note that, to get an idea of the contents, you can see the previously published fascicles, and earlier drafts online (pre-fascicles): • Mathematical Preliminaries Redux: https://cs.stanford.edu/~knuth/fasc5a.ps.gz • 7.2.2 Introduction to Backtracking: https://cs.stanford.edu/~knuth/fasc5b.ps.gz • 7.2.2.1 Dancing Links: https://cs.stanford.edu/~knuth/fasc5c.ps.gz • 7.2.2.2 Satisfiability: https://cs.stanford.edu/~knut…

[flagged]

"Imagine being Noah Webster at work, writing 'thousands of definitions' of words that already exist, and being paid for the privilege!"

There is value in providing reference texts. There is value in providing sample usages of programs to the unfamiliar. There is value in demonstrating alternative means to reach the same goal.

Donald Knuth's work of cataloguing and examining the different branches and leaves of computer science would be valuable even if he weren't also a world-class computer scientist in his own right (which he just happens to be).

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#188

Post by fragmede » Tue, Oct 04, 2022, 10:21 PM UTC

moron4hire wrote: post #143

lupire wrote: post #135

Earlier quoted context omitted.

You read a math book slowly, so you don't need whitespace for help skimming. Extra whitespace would make the huge books even huger. Beautiful formatting here refers to the text , not the whitespace.

Except everyone who talks about how to read the books successfully--including the books themselves--talk about jumping around in them in a way that requires a lot of skimming . I don't find the text in any other book I own to be insufficient, or even noticeably that different from Knuth's books (with the obvious exception of some fly-by-night print-on-demand stuff purchased off Amazon in recent years). It's an assert…

Paper's having supply chain issues, same as everything else. https://www.npr.org/2021/10/04/1043145212/supply-chain-issue...

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#189

Post by pksebben » Tue, Oct 04, 2022, 10:31 PM UTC

cromwellian wrote: post #182

A few years ago I had the pleasure of meeting Knuth in his home, my wife was doing some photography for him. He took me to his room and showed me his setup, he was researching Soduku algorithms at the time. His hands were blisteringly quick, moving between EMacs panes, triggering evaluations and printing of results, as fast as any 20 year old. In his 80s , he hasn’t seemed to suffer any mental decline. I started talk…

may we all age so gracefully. perhaps part of the secret is to work like Don.

Re: Knuth's Art of Computer Programming, V 4B, has gone into print

#190

Post by Brystephor » Tue, Oct 04, 2022, 10:39 PM UTC

So who is the target audience for this type of work? I mean if I want to learn more about something such as combinatorial algorithms, this doesn't seem like it'd be the right place to begin. I want to learn more about the things that I don't know, but at the same time, learning absolutely everything about a single category is typically more depth than I require.