I should have mentioned that my example works ok in J901 in Windows as well as J701 on this tablet. Mike
Sent from my iPad > On 25 Oct 2020, at 20:27, Mike Day <[email protected]> wrote: > > Experimenting with Euler Problem 731, I should get: > %3(] * <.@^)2<.@^>: i.3. NB. 1/(b^(c^k).c^k) in maths notation > 0.0555556 0.00308642 1.9052e_5 > %3x(] * <.@^)2x<.@^>: i.3. NB. Try extended nos > 1r18 1r324 1r52488 > > +/\%3x(] * <.@^)2x<.@^>: i.3. NB. Cum sum > 1r18 19r324 3079r52488 > > 23j20":,.+/\%3x(] * <.@^)2x<.@^>: i.3. Looking for repeated pattern .... > 0.05555555555555555556 > 0.05864197530864197531 > 0.05866102728242645938 > > Only using k=1,3 here for concise display. > > This is the sort of thing I see here on the iPad, and also in J901 > However, in J902 beta i, I get (can’t email from Windows just now, so faking > it!): > > +/\%3x(] * <.@^)2x<.@^>: i.3. > 0 0 0 > > My mistake!? > > Thanks, > > Mike > > > Sent from my iPad > >> On 25 Oct 2020, at 16:42, Henry Rich <[email protected]> wrote: >> >> Modern J uses a virtual block for the result of (, y), so now (+/@, y) and >> (+/ , y) run at the same speed. A different example is needed. >> >> Henry Rich >> >>> On 10/25/2020 12:01 PM, Brian Schott wrote: >>> Jan-Pieter, >>> >>> First, thanks for pointing out the link. >>> But I am having some problems. >>> >>> [*******my comment 0) here is likely inappropriate in light of Henry's >>> reply in this thread >>> 0) Are you suggesting that some wording be added to that link? It looks to >>> me as if you are but where? >>> I say where because there are at least 3 different concepts treated on that >>> page: SC, EIP, and AIP. >>> ] >>> >>> >>> 1) I am finding problems on that page that I cannot explain. For example, >>> a) on one line discussing EIP the following is stated twice, when it is >>> clear that the two should be different `(V0@[ V1 V2)` . >>> b) [this comment may also be explained by Henry's reply but I don't see how] >>> I cannot confirm the improvements suggested by the example there. >>> Perhaps using j807 is the problem. See my session below. >>> 9!:52'' >>> 1 >>> a =: 1000 1000 ?@$ 0 NB. it is unfortunate that the +/ result >>> suggests that a contains integers, not reals >>> +/ , a NB. perhaps 1000 1000 >>> could be reduced to 25 25? >>> 499960 >>> +/@, a >>> 499960 >>> ts =: 6!:2 , 7!:2@] >>> ts '+/ , a' NB. notice the small space usage here >>> 0.001987 1408 >>> ts '+/@, a' NB. the time reduction is so small as to be >>> "within statistical error" >>> 0.001965 1536 >>> JVERSION >>> Engine: j807/j64/darwin >>> Release-c: commercial/2019-02-24T10:50:40 >>> Library: 8.07.26 >>> Platform: Darwin 64 >>> Installer: J807 install >>> InstallPath: /users/brian/j64-807 >>> Contact: www.jsoftware.com >>> >>> On Sun, Oct 25, 2020 at 8:42 AM Jan-Pieter Jacobs < >>> [email protected]> wrote: >>> >>>> I guess this is worth mentioning here: >>>> >>>> https://code.jsoftware.com/wiki/Vocabulary/SpecialCombinations#Assignments_In_Place_.28AIP.29 >>>> >>>> Assignments are done in place if you use a specific form recognised by the >>>> interpreter. >>>> >>>> If you do not assign the result, or use it in a longer sentence, it will >>>> not be recognised, copying over the entire array, rather than updating a >>>> single element. >>>> >>>> Best regards, >>>> >>>> Jan-Pieter >>>> >>>> >>>> -- >>> (B=) >>> ---------------------------------------------------------------------- >>> For information about J forums see http://www.jsoftware.com/forums.htm >> >> >> -- >> This email has been checked for viruses by AVG. >> https://www.avg.com >> >> ---------------------------------------------------------------------- >> For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
