On 6/12/2012 10:16 AM, Raul Miller wrote:
> On Tue, Jun 12, 2012 at 10:01 AM, Henry Rich<henryhr...@nc.rr.com>  wrote:
>> On 6/11/2012 9:34 PM, Raul Miller wrote:
>>   >     -/ .* performs only one multiply.
>>
>> Keep in mind that -/ .*  is nothing like DET.  It uses rules of linear
>> algebra to calculate a result in far less than O(!n) time.
>
> I was actually using rephrasings such as
>     -/ .(*&([smoutput))
> and
>     -/ .(([smoutput)@:*)
> and
>     -/ .(([smoutput)@*)
> to count how many times multiplication was occurring.
>
> So I am not talking about any special code used for -/ .* -- I have
> not investigated that -- but instead am using -/ .* as an illustration
> of how . builds monadic verb definitions in the general case.
>
> Anyways, . terminates earlier than DET.
>

That may be.  But no way do you do a determinant with just one multiply 
- it takes 0 for a 1x1, 2 for a 2x2.


The definition of DET is so quirky that it makes sense only when you 
have subtraction/ . multiplication   for suitably defined operations on 
the space of interest.  Even +/ . *  runs in factorial time.  For the 
case of finding the characteristic polynomial, why not just use the verb 
definitions I gave earlier, rather than requiring a new DET?

Henry Rich
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to