[IA-64]BigDecimal#sqrt $B$N;EMM(B

TAKANO_Mitsuhiro · September 30, 2008, 7:54am

e$B$3$s$K$A$Oe(B
e$B9bLn$G$9!#e(B

IA-64e$B$N%a%s%F$r$7$F$$$k$N$G$9$,!"e(BBigDecimal#sqrte$B$N%F%9%H$Ge(BFaile$B$7$F$*$j!“e(B
e$B$=$N860x$,$h$/$o$+$j$^$;$s!#e(B
e$B<1<T$NJ}$N$40U8+$,$$$?$@$1$l$P$H;W$$!”%a!<%k$5$;$F$$$?$@$-$^$9!#e(B

e$B%F%9%H$N%3!<%I$Oe(B

x = BigDecimal.new((2**100).to_s)
assert_equal(1125899906842624, x.sqrt(1))
assert_equal(1125899906842624, x.sqrt(2))
assert_equal(1125899906842624, x.sqrt(3)) # I don’t understand the
meaning of argument…

e$B$H$$$&ItJ,$G!"<B9T;~$N%m%0$O0J2<$NDL$j$G$9!#e(B

http://www.atdot.net/sp/readonly/1coz7k_takano32

Failure:
test_sqrt_bigdecimal(TestBigDecimal)
[/home/takano/ruby/test/bigdecimal/test_bigdecimal.rb:469]:
Expected 1125899906842624, not
#<BigDecimal:60000000061650b0,‘0.1125899906 8426239999 9999999999
999975E16’,45(54)>.

e$B0z?t$NCM$,0[$J$k$N$KF1$8CM$HHf3S$7$F$$$k$N$,$$$1$J$$$N$+$J!)$H9M$(!"e(B
e$BB>$N%W%i%C%H%U%)!<%`$G$b0J2<$N$h$&$J%3!<%I$G<B9T$7$F$_$^$7$?!#e(B

require ‘bigdecimal’
puts BigDecimal.new( (2100).to_s ).sqrt(1)
puts BigDecimal.new( (2100).to_s ).sqrt(2)
puts BigDecimal.new( (2**100).to_s ).sqrt(3)

e$B$^$:$O$o$l$i$,!J!)!Ke(BIA-64e$B$G$9!#e(B
ruby 1.9.0 (2008-09-30 revision 19570) [ia64-linux]

0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16

e$B<!$Ke(B x64 e$B$G$9!#e(B
ruby 1.8.6 (2007-09-24 patchlevel 111) [x86_64-linux]

0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16

e$B:G8e$Ke(B IA-32 e$B$G$9!#e(B
ruby 1.8.7 (2008-08-11 patchlevel 72) [i686-linux]

0.1125899906842624E16
0.1125899906842624E16
0.1125899906842624E16

e$B$$$:$l$b0[$J$kCM$,5"$C$F$$$k$N$G$9$,!"e(B
revision 19633
http://www.rubyist.net/~akr/chkbuild/debian-sarge/ruby-trunk/log/20080930T132601.txt.gz
e$B$G$OC1=c$Ke(BFixnume$B$H$NHf3S$G%Q%9$7$F$$$k$h$&$G$9!#e(B

IA-64e$B$G$O@:EY$,$"$,$C$?$N$G!"Hf3S$9$k7e?t$,B?$/$J$C$F$$$k!)e(B

e$B<+J,$O$“$^$j$3$3$i$X$s$N$3$H$K>$7$/$J$$$N$G!”<1<T$NJ}$N$40U8+$,;G$($l$P$He(B
e$B;W$$!"O"Mm$5$;$F$$$?$@$-$^$7$?!#e(B

e$B$A$g$C$H$*$C$/$&$G!"e(BIA-64e$B0J30$Ne(Btrunke$B$N4D6-$r:n$C$F$$$J$$$N$,?=$7Lu$J$$e(B

e$B$H$3$m$J$N$G$9$,!"?<$_$KF~$k$h$&$G$7$?$i!"4D6-$r:n$C$F$*$3$&$H;W$$$^$9!#e(B

e$B2?$+$*J,$+$j$NJ}$,$$$i$C$7$c$$$^$7$?$i!"$h$m$7$/$*4j$$$7$^$9!#e(B

TAKANO_Mitsuhiro · September 30, 2008, 8:03am

e$B9bLn$G$9!#e(B

e$B$$$:$l$b0[$J$kCM$,5"$C$F$$$k$N$G$9$,!"e(B

e$B$9$_$^$;$s!"e(BIA-64e$B$He(Bx64e$B$OF1$8CM$,JV$C$F$$$^$9!#e(B orz

TAKANO Mitsuhiro e$B$5$s$O=q$-$^$7$?e(B:

TAKANO_Mitsuhiro · September 30, 2008, 8:21am

e$B9bLn$G$9!#e(B

e$B%a!<%j%s%0%j%9%H$KMpEj$9$_$^$;$s!#e(B

require ‘bigdecimal’
p BigDecimal.new( (2100).to_s ).sqrt(1) ==
1125899906842623.99999999999999999975
p BigDecimal.new( (2100).to_s ).sqrt(2) ==
1125899906842623.99999999999999999975
p BigDecimal.new( (2**100).to_s ).sqrt(3) ==
1125899906842623.99999999999999999975

e$B$Ge(B
true
true
true

e$B$H$J$j$^$7$?!#e(B @ IA-64

e$B@:EY$,$"$,$C$?$?$a!"$b$H$Ne(BFixnume$B$G$O0lCW$7$J$+$C$?$h$&$G$9!#e(B
e$B$H$j$"$($:!"%F%9%H$Ne(BFixnume$B$re(BFloate$B$K$7$F@:EY$r$"$2$kI,MW$,$"$k$h$&$G$9$,!“e(B
e$BB>$N%”!<%-%F%/%A%c$G$3$N%3!<%I$rF0$+$9$He(Bfalsee$B$K$J$C$F$j$7$F$h$/$o$+$j$^e(B
e$B$;$s!#e(B orz

TAKANO Mitsuhiro e$B$5$s$O=q$-$^$7$?e(B:

TAKANO_Mitsuhiro · September 30, 2008, 8:50am

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

In message “Re: [ruby-dev:36630] Re: [IA-64]BigDecimal#sqrt
e$B$N;EMMe(B”
on Tue, 30 Sep 2008 15:17:53 +0900, TAKANO Mitsuhiro
[email protected] writes:

e$BAG?M$G?=$7Lu$J$$$s$G$9$,!"$3$l$O%F%9%H$,0-$$$G$7$g$&$M$(!#e(B
e$B=q$-49$($F$/$@$5$kJ}$rJg=8$7$^$9!#e(B

TAKANO_Mitsuhiro · September 30, 2008, 9:51am

e$B9bLn$G$9!#e(B

LL Futuree$B$N$H$-$O$"$j$,$H$&$4$6$$$^!A$9!!e(B

|http://www.ruby-lang.org/ja/man/html/BigDecimal.html
|sqrt
e$B!!e(Bae$B$NM-8z7ee(B n e$B7e$NJ?J}:,e(B(n
e$B$NJ?J}:,$G$O$"$j$^$;$se(B)e$B$r%K%e!<%H%sK!$G7W;;$7e(B
e$B$^$9!#e(B
e$B!!e(Bc = a.sqrt(n)

e$B$3$N%I%-%e%a%s%H$N0UL#$,$h$/$o$+$i$J$+$C$?$N$Ge(B
e$B>/$7e(BBigDecimale$B$N%=!<%9%3!<%I$rFI$s$G$_$^$7$?!#e(B
http://www.atdot.net/sp/readonly/7jyz7k_takano32

e$B$I$&$d$i!"0z?t$G$Oe(Bsqrte$B$NM-8z?t;z$N2<8B$rM?$($k$h$&$K8+$($?$N$Ge(B
e$B%F%9%H$GBg$-$J0z?t$rM?$($k=$@5$r$7$?$ie(BIA-64e$B$G$b%Q%9$7$^$7$?!#e(B
sqrte$B$N0z?t$N;H$$J}$H$7$F$b@5$7$$$H;W$$$^$9!#e(B

e$B$=$b$=$b!"e(B1, 2, 3

e$B$N$h$&$J0z?t$G$O%F%9%H$K$J$C$F$$$J$$$h$&$G$9!#e(B

e$B$J$<$J$i!"e(BBigDecimal_sqrte$B$K$h$k$HM?$($i$l$?0z?t$,%G%U%)%k%H$Ne(B

e$B:GBgM-8z?t;z$h$j>.$5$$$H$-$O%G%U%)%k%H$NCM$,M%@h$5$l$k$?$a$G$9!#e(B

e$B$D$^$j!"e(Bsqrte$B$N0z?t$r$G$C$+$/$9$k$HLa$j$Ne(BBigDecimale$B$NM-8z?t;z$,Bg$-$/$J$j!"e(B
precse$B$NFsHVL$NMWAG$NCM$bBg$-$/$J$k$h$&$G$9!#e(B

e$B$J$s$@$+i%C_$C$]$/$J$C$F$-$?$N$G!“$H$j$”$($:%Q%C%A$rAw$j$^$9!#e(B

Index: test/bigdecimal/test_bigdecimal.rb

— test/bigdecimal/test_bigdecimal.rb (e$B%j%S%8%g%se(B 19633)
+++ test/bigdecimal/test_bigdecimal.rb (e$B:n6H%3%T!<e(B)
@@ -466,9 +466,9 @@
x = BigDecimal.new(“0.09”)
assert_in_delta(0.3, x.sqrt(1), 0.001)
x = BigDecimal.new((2**100).to_s)

assert_equal(1125899906842624, x.sqrt(1))
assert_equal(1125899906842624, x.sqrt(2))
assert_equal(1125899906842624, x.sqrt(3)) # I don’t understand the
meaning of argument…

assert_equal(1125899906842624, x.sqrt(100))
assert_equal(1125899906842624, x.sqrt(200))
assert_equal(1125899906842624, x.sqrt(300)) # I don’t understand the
meaning of argument…
x = BigDecimal.new(“-” + (2100).to_s)
assert_raise(FloatDomainError) { x.sqrt(1) }
x = BigDecimal.new((2200).to_s)

Yukihiro M. e$B$5$s$O=q$-$^$7$?e(B:

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

(snip)

TAKANO_Mitsuhiro · September 30, 2008, 10:27am

e$B9bLn$G$9!#e(B

e$B<h$j9~$_$"$j$,$H$&$4$6$$$^!A$9!#e(B

e$B%3%_%C%H8"$O$^$@$$$?$@$$$F$$$^$;$s$,!"$$$D$+$$$?$@$1$k$/$i$$$N@*$$$Ge(B

e$B9W8%$G$-$l$P!"$H;W$$$^$9!#e(B

Yukihiro M. e$B$5$s$O=q$-$^$7$?e(B:

TAKANO_Mitsuhiro · September 30, 2008, 10:01am

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

In message “Re: [ruby-dev:36634] Re: [IA-64]BigDecimal#sqrt
e$B$N;EMMe(B”
on Tue, 30 Sep 2008 16:49:47 +0900, TAKANO Mitsuhiro
[email protected] writes:

||http://www.ruby-lang.org/ja/man/html/BigDecimal.html
||sqrt
|e$B!!e(Bae$B$NM-8z7ee(B n e$B7e$NJ?J}:,e(B(n e$B$NJ?J}:,$G$O$“$j$^$;$se(B)e$B$r%K%e!<%H%sK!$G7W;;$7e(B
|e$B$^$9!#e(B
|e$B!!e(Bc = a.sqrt(n)
|
|e$B$3$N%I%-%e%a%s%H$N0UL#$,$h$/$o$+$i$J$+$C$?$N$Ge(B
|e$B>/$7e(BBigDecimale$B$N%=!<%9%3!<%I$rFI$s$G$$^$7$?!#e(B
|http://www.atdot.net/sp/readonly/7jyz7k_takano32
|
|e$B$I$&$d$i!"0z?t$G$Oe(Bsqrte$B$NM-8z?t;z$N2<8B$rM?$($k$h$&$K8+$($?$N$Ge(B
|e$B%F%9%H$GBg$-$J0z?t$rM?$($k=$@5$r$7$?$ie(BIA-64e$B$G$b%Q%9$7$^$7$?!#e(B
|sqrte$B$N0z?t$N;H$$J}$H$7$F$b@5$7$$$H;W$$$^$9!#e(B
|# e$B$=$b$=$b!"e(B1, 2, 3 e$B$N$h$&$J0z?t$G$O%F%9%H$K$J$C$F$$$J$$$h$&$G$9!#e(B
|# e$B$J$<$J$i!"e(BBigDecimal_sqrte$B$K$h$k$HM?$($i$l$?0z?t$,%G%U%)%k%H$Ne(B
|# e$B:GBgM-8z?t;z$h$j>.$5$$$H$-$O%G%U%)%k%H$NCM$,M%@h$5$l$k$?$a$G$9!#e(B
|
|e$B$D$^$j!"e(Bsqrte$B$N0z?t$r$G$C$+$/$9$k$HLa$j$Ne(BBigDecimale$B$NM-8z?t;z$,Bg$-$/$J$j!"e(B
|precse$B$NFsHVL$NMWAG$NCM$bBg$-$/$J$k$h$&$G$9!#e(B
|
|e$B$J$s$@$+i%C$C$]$/$J$C$F$-$?$N$G!”$H$j$"$($:%Q%C%A$rAw$j$^$9!#e(B

e$B%3%_%C%H8"$O$^$@$5$7$"$2$F$$$^$;$s$G$7$?$C$1!#e(B
e$B<h$j9~$s$G$*$-$^$9!#e(B

TAKANO_Mitsuhiro · October 4, 2008, 7:43am

e$B$?$+$N$G$9!#e(B

Yukihiro M. e$B$5$s$O=q$-$^$7$?e(B:

e$B%3%_%C%H8"$O$^$@$5$7$"$2$F$$$^$;$s$G$7$?$C$1!#e(B
e$B<h$j9~$s$G$*$-$^$9!#e(B

e$B$9$$^$;$s!#e(B
MacBook(x86_64)e$B4D6-$GF0$+$7$F$$?$ie(BFaile$B$7$F$7$^$$$^$7$?!&!&!&e(Borz
e$B!J$?$^$?$^$&$^$/F0$$$F$7$^$&4D6-$b$"$k$N$+$7$i!#!Ke(B

e$B@:EY$N>.$5$$$H$3$m$N:9$r<h$C$?$b$N$H@:EY$NBg$-$$$H$3$m$N:9$r<h$C$FHf3S$7e(B
e$B$?$[$&$,e(B
e$B3N<B$H9M$(!"0J2<$N$h$&$K=q$-49$($^$7$?!#$?$S$?$S$9$_$^$;$s!#e(B

e$B$J$s$@$+IT0B$J$N$G!"<1<T$NJ}$G$I$J$?$+e(BBigDecimale$B$K$*>$7$$$+$?$,$$$i$Ce(B
e$B$7$c$C$?$ie(B

e$B$<$R$4;XE&$*4j$$$7$^$9!#e(B

Index: test/bigdecimal/test_bigdecimal.rb

— test/bigdecimal/test_bigdecimal.rb (revision 19669)
+++ test/bigdecimal/test_bigdecimal.rb (working copy)
@@ -466,9 +466,7 @@
x = BigDecimal.new(“0.09”)
assert_in_delta(0.3, x.sqrt(1), 0.001)
x = BigDecimal.new((2**100).to_s)

assert_equal(1125899906842624, x.sqrt(100))
assert_equal(1125899906842624, x.sqrt(200))
assert_equal(1125899906842624, x.sqrt(300)) # I don’t understand the
meaning of argument…

assert_equal(true, x.sqrt(100) - x.sqrt(200) > x.sqrt(300) -
x.sqrt(400))
x = BigDecimal.new("-" + (2100).to_s)
assert_raise(FloatDomainError) { x.sqrt(1) }
x = BigDecimal.new((2200).to_s)

TAKANO_Mitsuhiro · October 5, 2008, 4:05pm

e$B$?$S$?$S$9$_$^$;$s!#e(B
e$B$?$+$N$G$9!#e(B

Yukihiro M. e$B$5$s$O=q$-$^$7$?e(B:

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

In message “Re: [ruby-dev:36662] Re: [IA-64]BigDecimal#sqrt e$B$N;EMMe(B”
on Sat, 4 Oct 2008 14:41:21 +0900, TAKANO Mitsuhiro [email protected] writes:

|MacBook(x86_64)e$B4D6-$GF0$+$7$F$_$?$ie(BFaile$B$7$F$7$^$$$^$7$?!&!&!&e(Borz
|e$B!J$?$^$?$^$&$^$/F0$$$F$7$^$&4D6-$b$"$k$N$+$7$i!#!Ke(B

e$B$&!<$`!#e(B

e$B$J$<$+$^$?e(BFaile$B$7$F$^$7$?!&!&!&e(B @ i386
orz

|e$B@:EY$N>.$5$$$H$3$m$N:9$r<h$C$?$b$N$H@:EY$NBg$-$$$H$3$m$N:9$r<h$C$FHf3S$7e(B
|e$B$?$[$&$,e(B
|e$B3N<B$H9M$(!"0J2<$N$h$&$K=q$-49$($^$7$?!#$?$S$?$S$9$_$^$;$s!#e(B

e$B@dBPCM$r$H$l$P$$$$$N$+$7$i!&!&!&$H9M$($^$7$?$,!"LLE]$J$N$G!"e(B
BigDecimal_sqrt
e$B$N%=!<%9DL$j$KM-8z?t;z$rHf3S$9$k$h$&$K=$@5$7$^$7$?!#e(B

e$BN;2r$G$9!#$?$@!“$$$?$@$$$?%Q%C%Ae(B(e$BA02s$b$G$9$,e(B)e$B$N9TF,%9%Z!<%9e(B
e$B$,A4It>C$($F$$$k>e$K@^$jJV$7$F$7$^$C$F$$$k$h$&$G$9!#%Q%C%A$re(B
e$BEj9F$5$l$k>l9g$K$O!”$=$NJU$KCm0U$7$F$$$?$@$1$k$H$5$$$o$$$G$9!#e(B

e$B$&$R$g!#$9$$^$;$s!#e(BTerminal.appe$B$+$ie(Bsvn
diffe$B$r%3%T$C$?$i$=$&$J$C$A$c$$$^e(B
e$B$7$?!&!&!&e(B
e$B:#EY$Oe(BMacVime$B$G3+$-D>$7$F$+$i%3%T!<$7$F$$^$7$?!#e(B

irbe$B$Ge(Bx64e$B$He(Bi386e$B$N3NG’$r$7!“3N<B$KBg$-$JI}$,$”$k$3$H$r3NG’$7$^$7$?$N$Ge(B

e$BBg>fIW$J$O$:$G$9!#$[$s$H$K$?$S$?$S$9$_$^$;$s!#e(B

x = BigDecimal.new((2**100).to_s)
=> #<BigDecimal:363c60,‘0.1267650600 2282294014 9670320537 6E31’,32(36)>
x.sqrt(300).precs
=> [16, 324]
x.sqrt(1200).precs
=> [16, 1224]

Index: test/bigdecimal/test_bigdecimal.rb

— test/bigdecimal/test_bigdecimal.rb (revision 19688)
+++ test/bigdecimal/test_bigdecimal.rb (working copy)
@@ -466,9 +466,7 @@
x = BigDecimal.new(“0.09”)
assert_in_delta(0.3, x.sqrt(1), 0.001)
x = BigDecimal.new((2**100).to_s)

assert_equal(1125899906842624, x.sqrt(100))
assert_equal(1125899906842624, x.sqrt(200))
assert_equal(1125899906842624, x.sqrt(300)) # I don’t understand the
meaning of argument…

assert_equal(true, x.sqrt(300).precs.last < x.sqrt(1200).precs.last)
x = BigDecimal.new(“-” + (2100).to_s)
assert_raise(FloatDomainError) { x.sqrt(1) }
x = BigDecimal.new((2200).to_s)

TAKANO_Mitsuhiro · October 9, 2008, 3:35pm

e$BK-J!$G$9!#e(B

In [ruby-dev:36662]

assert_equal(1125899906842624, x.sqrt(100))

assert_equal(1125899906842624, x.sqrt(200))

assert_equal(1125899906842624, x.sqrt(300)) # I don’t understand the meaning of argument…

assert_equal(true, x.sqrt(100) - x.sqrt(200) > x.sqrt(300) - x.sqrt(400))

y = BigDecimal(“1125899906842624”)
prec = y.precs[0]
assert_equal(true, (x.sqrt(100) - y).abs <
BigDecimal(“1E#{prec-100}”))
assert_equal(true, (x.sqrt(200) - y).abs <
BigDecimal(“1E#{prec-200}”))
assert_equal(true, (x.sqrt(300) - y).abs <
BigDecimal(“1E#{prec-300}”))

e$B$O$I$&$G$7$g$&!#e(B

TAKANO_Mitsuhiro · October 5, 2008, 4:36pm

ãŸã‹ã®ã§ã™ã€‚

TAKANO Mitsuhiro ã•ã‚“ã¯æ›¸ãã¾ã—ãŸ:

ã†ã²ã‚‡ã€‚ã™ã¿ã¾ã›ã‚“ã€‚Terminal.appã‹ã‚‰svn diffã‚’ã‚³ãƒ”ã£ãŸã‚‰ãã†ãªã£ã¡ã‚ƒã„ã¾
ã—ãŸãƒ»ãƒ»ãƒ»
ä»Šåº¦ã¯MacVimã§é–‹ãç›´ã—ã¦ã‹ã‚‰ã‚³ãƒ”ãƒ¼ã—ã¦ã¿ã¾ã—ãŸã€‚

ãœã‚“ãœã‚“ã ã‚ã§ã—ãŸãƒ»ãƒ»ãƒ»

ãŠã¨ãªã—ãæ·»ä»˜ã—ã¾ã™ã€‚

TAKANO_Mitsuhiro · October 4, 2008, 3:06pm

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

In message “Re: [ruby-dev:36662] Re: [IA-64]BigDecimal#sqrt
e$B$N;EMMe(B”
on Sat, 4 Oct 2008 14:41:21 +0900, TAKANO Mitsuhiro
[email protected] writes:

|MacBook(x86_64)e$B4D6-$GF0$+$7$F$_$?$ie(BFaile$B$7$F$7$^$$$^$7$?!&!&!&e(Borz
|e$B!J$?$^$?$^$&$^$/F0$$$F$7$^$&4D6-$b$"$k$N$+$7$i!#!Ke(B

e$B$&!<$`!#e(B

|e$B@:EY$N>.$5$$$H$3$m$N:9$r<h$C$?$b$N$H@:EY$NBg$-$$$H$3$m$N:9$r<h$C$FHf3S$7e(B
|e$B$?$[$&$,e(B
|e$B3N<B$H9M$(!"0J2<$N$h$&$K=q$-49$($^$7$?!#$?$S$?$S$9$_$^$;$s!#e(B

e$BN;2r$G$9!#$?$@!“$$$?$@$$$?%Q%C%Ae(B(e$BA02s$b$G$9$,e(B)e$B$N9TF,%9%Z!<%9e(B
e$B$,A4It>C$($F$$$k>e$K@^$jJV$7$F$7$^$C$F$$$k$h$&$G$9!#%Q%C%A$re(B
e$BEj9F$5$l$k>l9g$K$O!”$=$NJU$KCm0U$7$F$$$?$@$1$k$H$5$$$o$$$G$9!#e(B

e$B$C$F!“%3%_%C%H8”$r$5$7$"$2$l$P$$$$$N$+!#e(B

TAKANO_Mitsuhiro · October 14, 2008, 10:58am

e$B$^$D$b$He(B e$B$f$-$R$m$G$9e(B

In message “Re: [ruby-dev:36696] Re: [IA-64]BigDecimal#sqrt
e$B$N;EMMe(B”
on Thu, 9 Oct 2008 22:54:48 +0900, TOYOFUKU Chikanobu
[email protected] writes:

| y = BigDecimal(“1125899906842624”)
| e = y.exponent
| assert_equal(true, (x.sqrt(100) - y).abs < BigDecimal(“1E#{e-100}”))
| assert_equal(true, (x.sqrt(200) - y).abs < BigDecimal(“1E#{e-200}”))
| assert_equal(true, (x.sqrt(300) - y).abs < BigDecimal(“1E#{e-300}”))
|
|e$B$G$"$k$Y$-$G$7$?!#e(B
|e$B$3$N>l9g$?$^$?$^@0?t$@$C$?$N$Ge(B prec e$B$,e(B e e$B$H0lCW$7$F$^$7$?$,!#e(B

e$B$3$N@~$G=$@5$7$h$&$H;W$$$^$9$,!“;d$,M}2r$7$F$J$$$?$a$K4V0c$$e(B
e$B$,$”$k$H$$$1$^$;$s$+$i!"e(Btest_bigdecimal.rbe$B$KBP$9$k%Q%C%A$N7Ae(B
e$B$G$$$?$@$1$^$;$s$+!)e(B

TAKANO_Mitsuhiro · October 9, 2008, 4:28pm

e$BK-J!$G$9!#e(B

In [ruby-dev:36695]

y = BigDecimal(“1125899906842624”)
prec = y.precs[0]
assert_equal(true, (x.sqrt(100) - y).abs < BigDecimal(“1E#{prec-100}”))
assert_equal(true, (x.sqrt(200) - y).abs < BigDecimal(“1E#{prec-200}”))
assert_equal(true, (x.sqrt(300) - y).abs < BigDecimal(“1E#{prec-300}”))

e$B$9$_$^$;$s!#e(B

y = BigDecimal(“1125899906842624”)
e = y.exponent
assert_equal(true, (x.sqrt(100) - y).abs < BigDecimal(“1E#{e-100}”))
assert_equal(true, (x.sqrt(200) - y).abs < BigDecimal(“1E#{e-200}”))
assert_equal(true, (x.sqrt(300) - y).abs < BigDecimal(“1E#{e-300}”))

e$B$G$"$k$Y$-$G$7$?!#e(B
e$B$3$N>l9g$?$^$?$^@0?t$@$C$?$N$Ge(B prec e$B$,e(B e
e$B$H0lCW$7$F$^$7$?$,!#e(B

TAKANO_Mitsuhiro · October 15, 2008, 6:38am

é«˜é‡Žã§ã™ã€‚

ã“ã®ç·šã§ä¿®æ£ã—ã‚ˆã†ã¨æ€ã„ã¾ã™ãŒã€ç§ãŒç†è§£ã—ã¦ãªã„ãŸã‚ã«é–“é•ã„
ãŒã‚ã‚‹ã¨ã„ã‘ã¾ã›ã‚“ã‹ã‚‰ã€test_bigdecimal.rbã«å¯¾ã™ã‚‹ãƒ‘ãƒƒãƒã®å½¢
ã§ã„ãŸã ã‘ã¾ã›ã‚“ã‹
ãƒ‘ãƒƒãƒã§ã™ã€‚
ã‚ˆã‚ã—ããŠé¡˜ã„ã—ã¾ã™ã€‚

[IA-64]BigDecimal#sqrt $B$N;EMM(B

e$B%F%9%H$N%3!<%I$Oe(B

x = BigDecimal.new((2**100).to_s) assert_equal(1125899906842624, x.sqrt(1)) assert_equal(1125899906842624, x.sqrt(2)) assert_equal(1125899906842624, x.sqrt(3)) # I don’t understand the meaning of argument…

http://www.atdot.net/sp/readonly/1coz7k_takano32

e$B0z?t$NCM$,0[$J$k$N$KF1$8CM$HHf3S$7$F$$$k$N$,$$$1$J$$$N$+$J!)$H9M$(!"e(B e$BB>$N%W%i%C%H%U%)!<%`$G$b0J2<$N$h$&$J%3!<%I$G<B9T$7$F$_$^$7$?!#e(B

require ‘bigdecimal’ puts BigDecimal.new( (2100).to_s ).sqrt(1) puts BigDecimal.new( (2100).to_s ).sqrt(2) puts BigDecimal.new( (2**100).to_s ).sqrt(3)

e$B$^$:$O$o$l$i$,!J!)!Ke(BIA-64e$B$G$9!#e(B ruby 1.9.0 (2008-09-30 revision 19570) [ia64-linux]

0.112589990684262399999999999999999975E16 0.112589990684262399999999999999999975E16 0.112589990684262399999999999999999975E16

e$B<!$Ke(B x64 e$B$G$9!#e(B ruby 1.8.6 (2007-09-24 patchlevel 111) [x86_64-linux]

0.112589990684262399999999999999999975E16 0.112589990684262399999999999999999975E16 0.112589990684262399999999999999999975E16

e$B:G8e$Ke(B IA-32 e$B$G$9!#e(B ruby 1.8.7 (2008-08-11 patchlevel 72) [i686-linux]

0.1125899906842624E16 0.1125899906842624E16 0.1125899906842624E16

IA-64e$B$G$O@:EY$,$"$,$C$?$N$G!"Hf3S$9$k7e?t$,B?$/$J$C$F$$$k!)e(B

LL Futuree$B$N$H$-$O$"$j$,$H$&$4$6$$$^!A$9!*!*e(B

e$B$=$b$=$b!"e(B1, 2, 3

e$B:GBgM-8z?t;z$h$j>.$5$$$H$-$O%G%U%)%k%H$NCM$,M%@h$5$l$k$?$a$G$9!#e(B

Index: test/bigdecimal/test_bigdecimal.rb

e$B9W8%$G$-$l$P!"$H;W$$$^$9!#e(B

e$B$<$R$4;XE&$*4j$$$7$^$9!#e(B

Index: test/bigdecimal/test_bigdecimal.rb

e$BBg>fIW$J$O$:$G$9!#$[$s$H$K$?$S$?$S$9$_$^$;$s!#e(B

Index: test/bigdecimal/test_bigdecimal.rb

e$B$O$I$&$G$7$g$&!#e(B

ãƒ¡ãƒ¼ãƒ©ã ã£ãŸã®ã‹ãƒ¼ã£ã€‚

e$B$C$F!“%3%_%C%H8”$r$5$7$"$2$l$P$$$$$N$+!#e(B

e$B$G$"$k$Y$-$G$7$?!#e(B e$B$3$N>l9g$?$^$?$^@0?t$@$C$?$N$Ge(B prec e$B$,e(B e e$B$H0lCW$7$F$^$7$?$,!#e(B

[IA-64]BigDecimal#sqrt $B$N;EMM(B

x = BigDecimal.new((2**100).to_s)
assert_equal(1125899906842624, x.sqrt(1))
assert_equal(1125899906842624, x.sqrt(2))
assert_equal(1125899906842624, x.sqrt(3)) # I don’t understand the
meaning of argument…

e$B0z?t$NCM$,0[$J$k$N$KF1$8CM$HHf3S$7$F$$$k$N$,$$$1$J$$$N$+$J!)$H9M$(!"e(B
e$BB>$N%W%i%C%H%U%)!<%`$G$b0J2<$N$h$&$J%3!<%I$G<B9T$7$F$_$^$7$?!#e(B

require ‘bigdecimal’
puts BigDecimal.new( (2100).to_s ).sqrt(1)
puts BigDecimal.new( (2100).to_s ).sqrt(2)
puts BigDecimal.new( (2**100).to_s ).sqrt(3)

e$B$^$:$O$o$l$i$,!J!)!Ke(BIA-64e$B$G$9!#e(B
ruby 1.9.0 (2008-09-30 revision 19570) [ia64-linux]

0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16

e$B<!$Ke(B x64 e$B$G$9!#e(B
ruby 1.8.6 (2007-09-24 patchlevel 111) [x86_64-linux]

0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16
0.112589990684262399999999999999999975E16

e$B:G8e$Ke(B IA-32 e$B$G$9!#e(B
ruby 1.8.7 (2008-08-11 patchlevel 72) [i686-linux]

0.1125899906842624E16
0.1125899906842624E16
0.1125899906842624E16

LL Futuree$B$N$H$-$O$"$j$,$H$&$4$6$$$^!A$9!!e(B

ãƒ¡ãƒ¼ãƒ©ã ã£ãŸã®ã‹ãƒ¼ã£ã€‚

e$B$G$"$k$Y$-$G$7$?!#e(B
e$B$3$N>l9g$?$^$?$^@0?t$@$C$?$N$Ge(B prec e$B$,e(B e
e$B$H0lCW$7$F$^$7$?$,!#e(B