Package: apcalc Version: 2.12.3.3-1 Severity: normal Hi, I noticed something I cannot understand.
I wanted to compute a base 10 logarithm with increasing precision, but it seems that the precision does not increase beyond some sort of limit value: $ calc C-style arbitrary precision calculator (version 2.12.3.3) Calc is open software. For license details type: help copyright [Type "exit" to exit, or "help" for help.] ; display(70) 20 ; for (n = 4 ; n <= 70 ; n += 4) { epsilon(10^(-n)) ; print n, log(10^0.5) ; } 4 0.5 8 ~0.4999967645270563710588030921566924346391036219925301832710848605923738 12 ~0.4999967630057326500477720837314340310952835924606966038391383653261530 16 ~0.4999967630057425518978545991487883262399027186658559888821332406844437 20 ~0.4999967630057425701424476678537305654477547120646226005385216711543473 24 ~0.4999967630057425701420116390167636584730304872752540606271171718926431 28 ~0.4999967630057425701420115205854251715452097628767480239729001997741683 32 ~0.4999967630057425701420115206037435941978632849821940415182836793190307 36 ~0.4999967630057425701420115206037445079475375662294797185789976548249805 40 ~0.4999967630057425701420115206037445078609398071744983931208199426734995 44 ~0.4999967630057425701420115206037445078609184139668201163901676365847303 48 ~0.4999967630057425701420115206037445078609184158403543820029531833579432 52 ~0.4999967630057425701420115206037445078609184158405222357335186311126553 56 ~0.4999967630057425701420115206037445078609184158405222305307044210892035 60 ~0.4999967630057425701420115206037445078609184158405222305292556240771302 64 ~0.4999967630057425701420115206037445078609184158405222305292556025796925 68 ~0.4999967630057425701420115206037445078609184158405222305292556025726483 Awkwardly enough, it seems that the exact value is computed, when epsilon is set to 10^-4, but, when I set epsilon to 10^-8 or less, the absolute error (with respect to the exact value of 0.5) does not seem to lower significantly. Indeed, the absolute error seem to be stay almost unchanged: ; for (n = 4 ; n <= 70 ; n += 4) { epsilon(10^(-n)) ; print n, 0.5 - log(10^0.5) ; } 4 0 8 ~0.0000032354729436289411969078433075653608963780074698167289151394076262 12 ~0.0000032369942673499522279162685659689047164075393033961608616346738470 16 ~0.0000032369942574481021454008512116737600972813341440111178667593155563 20 ~0.0000032369942574298575523321462694345522452879353773994614783288456527 24 ~0.0000032369942574298579883609832363415269695127247459393728828281073569 28 ~0.0000032369942574298579884794145748284547902371232519760270998002258317 32 ~0.0000032369942574298579884793962564058021367150178059584817163206809693 36 ~0.0000032369942574298579884793962554920524624337705202814210023451750195 40 ~0.0000032369942574298579884793962554921390601928255016068791800573265005 44 ~0.0000032369942574298579884793962554921390815860331798836098323634152697 48 ~0.0000032369942574298579884793962554921390815841596456179970468166420568 52 ~0.0000032369942574298579884793962554921390815841594777642664813688873447 56 ~0.0000032369942574298579884793962554921390815841594777694692955789107965 60 ~0.0000032369942574298579884793962554921390815841594777694707443759228698 64 ~0.0000032369942574298579884793962554921390815841594777694707443974203075 68 ~0.0000032369942574298579884793962554921390815841594777694707443974273517 ; I seem to get reasonable results with, say, the computation of sqrt(2), where the absolute error really decreases exponentially (as I think it should). Why does the computation of log(10^0.5) fail to increase in precision? What did I fail to understand? Could you please explain if I missed something in the documentation (I tried to read various "help" documents) or if this is an actual bug? Thanks for your time. -- System Information: Debian Release: squeeze/sid APT prefers testing APT policy: (800, 'testing'), (500, 'unstable') Architecture: amd64 (x86_64) Kernel: Linux 2.6.32-5-amd64 (SMP w/2 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/bash Versions of packages apcalc depends on: ii apcalc-common 2.12.3.3-1 Arbitrary precision calculator (co ii libc6 2.11.2-6+squeeze1 Embedded GNU C Library: Shared lib ii libreadline5 5.2-7 GNU readline and history libraries apcalc recommends no packages. apcalc suggests no packages. -- no debconf information -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org