Reading Group?

Ian Bicking ian at ianbicking.org
Sat Mar 28 01:28:13 UTC 2020


Classical AI (not neural networks/machine learning) could also be an
interesting topic. It's really lots of different algorithms, often
involving search and heuristics. AIMA is mostly like this:
http://aima.cs.berkeley.edu/ – it feels kind of like SICP.


On Fri, Mar 27, 2020 at 6:13 PM Joe Nelson <joe at begriffs.com> wrote:
>
> Brian H wrote:
> > I was wondering if there would be any interest in a reading group for
> > the book "The Art of Computer Programming" starting with volume 1.
>
> That series is overwhelming, you'd need to scale down the scope. For
> instance the exercises are rated on a logarithmic scale of difficulty:
>
>         00 Immediate
>         10 Simple (one minute)
>         20 Medium (quarter hour)
>         30 Moderately hard
>         40 Term project
>         50 Research problem
>
> They are also marked:
>
>         ▶  Recommended
>         M  Mathematically oriented
>         HM Requiring "higher math"
>
> Speaking for myself, I'd probably pick problems that are less than or
> equal to 20 in difficulty, and probably only the recommended ones.
> There's only so much time in a day to learn things, and there are other
> topics that I feel are more exciting.
>
> TAOCP uses a hypothetical computer architecture called MIX. Check out
> the wikipedia page, it's weird.
>
>         When programmed in binary, each byte has 6 bits (values range from 0
>         to 63). In decimal, each byte has 2 decimal digits (values range
>         from 0 to 99). Bytes are grouped into words of five bytes plus a
>         sign.
>
> The integers are sign-magnitude rather than twos' complement, and there
> are only nine registers with special purposes. You program the machine
> with an assembly language called MIXAL.
>
> Subsequently Knuth switched to an improved 64-bit RISC architecture
> called MMIX. I think the goal behind both of these architectures was to
> make the book timeless, and not a vendor-specific trade publication.
> However, I personally don't want to commit any of that stuff to memory.
> I'd rather use a medium level language like C and reason using the C
> abstract machine model.
>
> That's why if I were learning algorithms I would probably go through
> Sedgewick's book "Algorithms in C." I understand it may not match your
> goals because it doesn't go very deep into correctness proofs or
> mathematical analysis.



-- 
Ian Bicking  |  http://ianbicking.org


More information about the Friends mailing list