That seems powerful. asserting unique or sorted attributes to nouns may make
some internal verbs faster as well.
Assertions about verbs would usually be part of the definition.
On Wednesday, July 4, 2018, 9:00:57 p.m. EDT, Henry Rich
wrote:
Rather than a new table adverb, I would
ive with 2 combinations #
SPHERES
Date: Fri, 29 Jun 2018 12:34:09 -0400
From: Raul Miller
To: Programming forum
Subject: Re: [Jprogramming] Proposal for a new table adverb
Message-ID:
Content-Type: text/plain; charset="UTF-8"
Here's a tentative implementation of part o
Rather than a new table adverb, I would prefer a conjunction that
asserted properties of a verb/noun: commutative, left-associative, and
right-associative come to mind for verbs, Hermitian and antiHermitian
for nouns, and there may be others. Then the existing adverbs like [x]
u/\ y and x u/ y
Here's a tentative implementation of part of what I think you have described:
commute=:3 :0
assert.2=#$y
assert.=/$y
assert. 0={.u b. 0
mask=. <:/~i.#y
inds=. (>. |:)mask*i.$y
inds{u mask #&, y
)
That said, I don't think this would work for you - this winds up
testing each item indepe
I have an equality test. If a=b then b=a . (In my case "find overlapping
spheres", in other words the equality test may be non-trivial.) The new
adverb would compute only the upper triangular set of cases and thereby
save about half the time.
NB. assuming the main diagonal could be interesti
