have you defined DP ?
On Saturday, October 10, 2020, 04:39:53 p.m. EDT, Don Guinn <[email protected]> wrote: Playing with Roger's page for calculating extended precision transcendental functions, "Essays/Extended_Precision_Functions", I found an error in Beta J, not his definitions. There appears to be an error in the handling of $: as to the rank of monadic "ln". Probably other functions in the essay have the same problem. In trying to better isolate the problem I had a J crash and a stack error. Hopefully the miscalculation and the stack error are sufficient to isolate the error; however, I can send the J crash if necessary. In J6 it treats the modac rank of ln as 0, but it is unclear what the monadic rank of ln is. This is the original problem: ln=: DP&$: : (4 : 0) " 0 assert. 0<y m=. <.0.5+2^.y t=. (<:%>:) (x:!.0 y)%2x^m if. x<-:#":t do. t=. (1+x) round t end. ln2=. 2*+/1r3 (^%]) 1+2*i.>.0.5*(%3)^.0.5*0.1^x+>.10^.1>.m lnr=. 2*+/t (^%]) 1+2*i.>.0.5*(|t)^.0.5*0.1^x lnr + m * ln2 ) 0j20":ln 2 3 0.69314718055994530942 43.00000000000000000000 Obviously not the ln of 3. This gives the correct answer in J8.07 but is also wrong in J9.01 . This results in a stack error: add=:(4 : 0)"0 echo 'Running add ',":x,y x+y ) p=:DP&$: : (4 : 0) " 0 x add y ) p 2 3 |stack error: p | p 2 3 |[-9] JVERSION Engine: j902/j64avx2/windows Beta-h: commercial/2020-10-03T19:50:26 Library: 9.02.06 Qt IDE: 1.8.7/5.12.7(5.12.7) Platform: Win 64 Installer: J902 install InstallPath: c:/j902 Contact: www.jsoftware.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
