I am not sure this is relevant to your questions, but in my J session below
I loaded the debug addon named "trace" and used it for two different
versions of your code. To run your version of the code which uses the
defined conjunction "c" I had to use "c__" in the trace because otherwise I
got a value error. But I think the result of trace may clarify the parsing
that J performs.
c=: 2 :(':';'[: u v')
(- c - c - c - )i. 4
0 1 2 3
([: - [: - [: - [: - ])i. 4
0 1 2 3
2([: - [: - [: - [: - ])i. 4
0 1 2 3
load'trace'
trace'([: - [: - [: - [: - ])i. 4'
--------------- 5 Trident ----
[:
-
]
[: - ]
--------------- 5 Trident ----
[:
-
[: - ]
[: - [: - ]
--------------- 5 Trident ----
[:
-
[: - [: - ]
[: - [: - [: - ]
--------------- 5 Trident ----
[:
-
[: - [: - [: - ]
[: - [: - [: - [: - ]
--------------- 8 Paren ------
(
[: - [: - [: - [: - ]
)
[: - [: - [: - [: - ]
--------------- 1 Monad ------
i.
4
0 1 2 3
--------------- 0 Monad ------
[: - [: - [: - [: - ]
0 1 2 3
0 1 2 3
==============================
0 1 2 3
trace'(- c__ - c__ - c__ - )i. 4'
--------------- 4 Conj -------
-
2 : '[: u v'
-
[: - -
--------------- 4 Conj -------
[: - -
2 : '[: u v'
-
[: ([: - -) -
--------------- 4 Conj -------
[: ([: - -) -
2 : '[: u v'
-
[: ([: ([: - -) -) -
--------------- 8 Paren ------
(
[: ([: ([: - -) -) -
)
[: ([: ([: - -) -) -
--------------- 1 Monad ------
i.
4
0 1 2 3
--------------- 0 Monad ------
[: ([: ([: - -) -) -
0 1 2 3
0 1 2 3
==============================
0 1 2 3
On Sat, Aug 6, 2016 at 4:02 PM, Erling Hellenäs <[email protected]>
wrote:
> u1(u2(u3(u4(y)))) was the mathematical description of the strawman
> algorithm. We are still in a discussion about tacit J. The expression you
> are opposing is this one:
>
>
> "As far as I can understand the pattern of the strawman, [: u1 [:u2 [: u3
> [: u4 ] , is the only expression in tacit J with the meaning of 4 monadic
> verbs ux in sequence. All other similar expressions have a more complex
> meaning. Correct me if I'm wrong."
>
> Bob Thierrault says:
>
> I believe it could also be expressed as
>
> u1 @: u2 @: u3 @: u4 @: ]"_
>
> Cheers, bob
>
> You say he is right:
>
> "Yes, and the @:]"_ is redundant for any and all verbs u1 u2 u3 and u4.
> And ([: u1 [:u2 u3@:u4) is another example equivalent..."
>
> I think I proved you both wrong with this:
>
> There is a considerable difference in memory allocation indicating
> differences between these expressions. If the first version creates
> three copies, the second seems to create five.
>
>
> ts'([: - [: - [: - [: - ])i. 100000000'
>
> 1.45991 3.22123e9
>
> ts'(- @: - @: - @: - )i. 100000000'
>
> 1.48942 5.36871e9
>
> And this:
>
> Its is some redundant calls to the @: conjunction that makes the
> difference?
>
> ts'(- @: - @: - @: - )i. 100000000'
>
> 1.56681 5.36871e9
>
> c=: 2 :(':';'[: u v')
>
> ts'(- c - c - c - )i. 100000000'
>
> 1.56479 5.36871e9
>
> It is unclear which tacit J expression you mean by u1(u2(u3(u4(y)))).
>
> /Erling
>
>
>
> On 2016-08-05 22:22, Raul Miller wrote:
>
>> If you want the algorithm of u1(u2(u3(u4(y)))) and not any expression
>> with the same result, then you should be using u1(u2(u3(u4(y)))) and
>> not some other expression.
>>
>> That said, you might want to wrap it in verb def '' (with the
>> expression going inside the quotes) so that you can have a verb
>> definition.
>>
>> Thanks,
>>
>>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
--
(B=) <-----my sig
Brian Schott
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
