The string approach seems more efficient in the long run, I guess. The human eye makes no difference between the string and the corresponding number, so why would I?
I prefer Hui's solution with a slight amendment on f and preprocessing the input number: f=: ; @: (*:&.".&.>) f^:(i.10) ": 9 R.E. Boss > -----Oorspronkelijk bericht----- > Van: Programming <[email protected]> > Namens Ben Gorte > Verzonden: dinsdag 14 mei 2019 03:40 > Aan: [email protected] > Onderwerp: Re: [Jprogramming] Seqfan problem > > I can make it a few characters shorter and avoid the logarithm, but the main > idea is still the same: > > > g =: (#.~10x^(1+>&9))@:*:@:(10&#.^:_1) > > ,.g^:(<10) 9 > > 9 > > 81 > > 641 > > 36161 > > 9361361 > > 8193619361 > > 641819361819361 > > 36161641819361641819361 > > 93613613616164181936136161641819361 > > 8193619361936136136161641819361936136136161641819361 > > > I think the difference between R.E.'s/my idea at one side, and Mike's or > Roger's at the other, is that we actually compute the numbers of the > sequence, whereas they 'only' generate the strings. Or is that futile? > > > Greetings > > Ben > > On Tue, May 14, 2019 at 12:07 AM Roger Hui <[email protected]> > wrote: > > > f=: ; @: (*:&.".&.>) @: ": > > > > f 9 > > 81 > > f f 9 > > 641 > > f f f 9 > > 36161 > > > > f^:(i.10) 9 > > |domain error > > | f^:(i.10)9 > > > > This last runs afoul because of f^:0]9 which has a numeric result and > > can not be mixed with the other f^:n]9. It can be fixed by > > preapplication of > > ": or by starting with '9', or by not doing f^:0 of course. > > > > f^:(i.10) ": 9 > > 9 > > 81 > > 641 > > 36161 > > 9361361 > > 8193619361 > > 641819361819361 > > 36161641819361641819361 > > 93613613616164181936136161641819361 > > 8193619361936136136161641819361936136136161641819361 > > > > On Mon, May 13, 2019 at 2:48 AM R.E. Boss <[email protected]> > wrote: > > > > > Was wrong subject. My mailing is getting rusty too. > > > > > > > > > R.E. Boss > > > > > > > > > > -----Oorspronkelijk bericht----- > > > > Van: Programming <[email protected]> > > > > Namens R.E. Boss > > > > Verzonden: maandag 13 mei 2019 11:07 > > > > Aan: [email protected] > > > > Onderwerp: Re: [Jprogramming] JELM - Extreme Learning Machine in J > > > > > > > > It took me more time than expected to produce the sequence of > > > > numbers > > in > > > > J: > > > > > > > > (from Seqfan Mailing list - http://list.seqfan.eu/) > > > > > Beginning with a two-digit square, successively replace each > > > > > digit with its square, then repeat. > > > > > > > > > > For example, > > > > > 9 > > > > > 81 > > > > > 641 > > > > > 36161 > > > > > 9361361 > > > > > 8193619361 > > > > > 641819361819361 > > > > > 36161641819361641819361 > > > > > 93613613616164181936136161641819361 > > > > > > > > ((]#.~ 10x^10>:@<.@^.]) *:@(10&#.^:_1))^:(i.10)9 > > > > 9 81 641 36161 9361361 8193619361 641819361819361 > > > > 36161641819361641819361 93613613616164181936136161641819361 > > > > 8193619361936136136161641819361936136136161641819361 > > > > > > > > Curious about other, more elegant solutions, since my J is getting > > rusty. > > > > > > > > > > > > R.E. Boss > > > > ------------------------------------------------------------------ > > > > ---- For information about J forums see > > > > http://www.jsoftware.com/forums.htm > > > -------------------------------------------------------------------- > > > -- For information about J forums see > > > http://www.jsoftware.com/forums.htm > > > > > ---------------------------------------------------------------------- > > For information about J forums see > http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
