Research proposal

[pozorvlak]

I'm in the middle of writing up a research proposal for a postdoc position I'm applying for. I've had enough of categorification for a while, and it struck me that doing formal semantics for a simple language in the APL family (J, K, A+, Matlab...) could be both Fun and Interesting. Some poking around on Google and MathSciNet suggests that nobody's

Comments

[andustar]

Haha. I was SO WAITING for that punchline.

(Sorry I'm not anywhere near understanding the content enough to comment on it).

[susannahf]

good start. obviously starts getting a little more, shall we say, informal at the end. I like the addition metaphor. I hadn't come across the last one; now I want to see if Matlab adheres to it.

[pozorvlak]

Metaphor, you say? :-)

pozorvlak@delirium:~$ j601/jconsole
1 + 3
4
1 2 3 + 4 5 6
5 7 9
1 + 4 5 6
5 6 7
i.3
0 1 2
3 3 $ 0
0 0 0
0 0 0
0 0 0
(i.3) + 3 3 $ 0
0 0 0
1 1 1
2 2 2 I don't have a copy of Matlab to hand to check, but I believe it should work in approximately the same way: the line of descent from APL to Matlab isn't as direct as the one to J, K or A+, though. Note that you can also get row-wise addition in J by using the " (rank) conjunction:
(i.3) (+"1) 3 3 $ 0
0 1 2
0 1 2
0 1 2 Accounting for the effects of the rank conjunction is, I think, going to be one of the Interesting Bits :-)

[susannahf]

bad choice of words. example then.
I shall be checking it when I get in this morning, and shall try to remember to report back

[pozorvlak]

...and what was the answer? :-)

[michiexile]

Hey, hey, hey! I like your research programme! Can I join it?

[pozorvlak]

Sure :-)

[half_of_monty]

Looks good. I'd focus your later drafts on the whole `getting two communities to talk to each other' aspect, that generally goes down well.

I'd remove the intended bits about how much you need to learn - you need to sell yourself as an expert. (Mind you I've never written a research proposal for a project that was new to me - but I have read research proposals for work which I happen to know was outside an academic's field of expertese; you would think they were already experts).

Presumably you already have the requirements in terms of length and intended audience in front of you? Greatly changes what you actually write. And, unfortunately, they're generally unique to each job.

Good luck!

[pozorvlak]

The APL community's a bit isolated in general - Guy Steele (major serious CS dude, mostly hailing from the Lisp community) gave a keynote speech at SIGAPL '07 on " What APL Can Teach the World (and vice versa)", in which he apparently claimed that "there are certain important lessons from APL that the rest of the world has still not learned." So, nice that a big wheel agrees with me - I should probably mention that in the proposal, actually. But yeah, it's really frustrating to read stuff about APL and see them struggling with things that the Lispers have been doing for years, or read the functional guys talking about stuff that's been standard APL practice since the mid-Sixties. And it would be nice to break the "category theory == Haskell" myth :-)

[half_of_monty]

Great. Find examples. Include.

[jonjonc]

This sounds a bit like some of Barry Jay's work on shape theory and polymorphism.

[pozorvlak]

Wow! That's... worryingly close to the kind of thing I wanted to do. Though his statement that "initially, it was not at all clear that shape polymorphic algorithms exist" suggests that I'm on the money about the invisibility of APL to academics - APL programming is all about shape polymorphism.

[pozorvlak]

OK, so I've now got the problem of explaining how my planned work is different from his. I'd been planning on starting with APL-style arrays and later generalising to other shapes, and using enriched categories as my basic model, but I like his approach of a shape being a functor (or rather, the domain of a functor, in a Lawvere theory kind of way). But it's so, so frustrating to read his stuff and see that he clearly has no idea just how old and how thoroughly implemented the array part of his work is.

Thanks!

[pozorvlak]

BTW - did we meet at CT2007?