1 package Math::BigFloat;
4 # Mike grinned. 'Two down, infinity to go' - Mike Nostrus in 'Before and After'
7 # The following hash values are internally used:
8 # _e : exponent (ref to $CALC object)
9 # _m : mantissa (ref to $CALC object)
11 # sign : +,-,+inf,-inf, or "NaN" if not a number
19 @ISA = qw(Exporter Math::BigInt);
22 # $_trap_inf/$_trap_nan are internal and should never be accessed from outside
23 use vars qw/$AUTOLOAD $accuracy $precision $div_scale $round_mode $rnd_mode
24 $upgrade $downgrade $_trap_nan $_trap_inf/;
25 my $class = "Math::BigFloat";
28 '<=>' => sub { $_[2] ?
29 ref($_[0])->bcmp($_[1],$_[0]) :
30 ref($_[0])->bcmp($_[0],$_[1])},
31 'int' => sub { $_[0]->as_number() }, # 'trunc' to bigint
34 ##############################################################################
35 # global constants, flags and assorted stuff
37 # the following are public, but their usage is not recommended. Use the
38 # accessor methods instead.
40 # class constants, use Class->constant_name() to access
41 $round_mode = 'even'; # one of 'even', 'odd', '+inf', '-inf', 'zero' or 'trunc'
48 # the package we are using for our private parts, defaults to:
49 # Math::BigInt->config()->{lib}
50 my $MBI = 'Math::BigInt::FastCalc';
52 # are NaNs ok? (otherwise it dies when encountering an NaN) set w/ config()
54 # the same for infinity
57 # constant for easier life
60 my $IMPORT = 0; # was import() called yet? used to make require work
62 # some digits of accuracy for blog(undef,10); which we use in blog() for speed
64 '2.3025850929940456840179914546843642076011014886287729760333279009675726097';
65 my $LOG_10_A = length($LOG_10)-1;
68 '0.6931471805599453094172321214581765680755001343602552541206800094933936220';
69 my $LOG_2_A = length($LOG_2)-1;
70 my $HALF = '0.5'; # made into an object if necc.
72 ##############################################################################
73 # the old code had $rnd_mode, so we need to support it, too
75 sub TIESCALAR { my ($class) = @_; bless \$round_mode, $class; }
76 sub FETCH { return $round_mode; }
77 sub STORE { $rnd_mode = $_[0]->round_mode($_[1]); }
81 # when someone set's $rnd_mode, we catch this and check the value to see
82 # whether it is valid or not.
83 $rnd_mode = 'even'; tie $rnd_mode, 'Math::BigFloat';
86 ##############################################################################
89 # valid method aliases for AUTOLOAD
90 my %methods = map { $_ => 1 }
91 qw / fadd fsub fmul fdiv fround ffround fsqrt fmod fstr fsstr fpow fnorm
92 fint facmp fcmp fzero fnan finf finc fdec flog ffac fneg
93 fceil ffloor frsft flsft fone flog froot
95 # valid method's that can be hand-ed up (for AUTOLOAD)
96 my %hand_ups = map { $_ => 1 }
97 qw / is_nan is_inf is_negative is_positive is_pos is_neg
98 accuracy precision div_scale round_mode fabs fnot
99 objectify upgrade downgrade
103 sub method_alias { exists $methods{$_[0]||''}; }
104 sub method_hand_up { exists $hand_ups{$_[0]||''}; }
107 ##############################################################################
112 # create a new BigFloat object from a string or another bigfloat object.
115 # sign => sign (+/-), or "NaN"
117 my ($class,$wanted,@r) = @_;
119 # avoid numify-calls by not using || on $wanted!
120 return $class->bzero() if !defined $wanted; # default to 0
121 return $wanted->copy() if UNIVERSAL::isa($wanted,'Math::BigFloat');
123 $class->import() if $IMPORT == 0; # make require work
125 my $self = {}; bless $self, $class;
126 # shortcut for bigints and its subclasses
127 if ((ref($wanted)) && (ref($wanted) ne $class))
129 $self->{_m} = $wanted->as_number()->{value}; # get us a bigint copy
130 $self->{_e} = $MBI->_zero();
132 $self->{sign} = $wanted->sign();
133 return $self->bnorm();
137 # handle '+inf', '-inf' first
138 if ($wanted =~ /^[+-]?inf\z/)
140 return $downgrade->new($wanted) if $downgrade;
142 $self->{sign} = $wanted; # set a default sign for bstr()
143 return $self->binf($wanted);
146 # shortcut for simple forms like '12' that neither have trailing nor leading
148 if ($wanted =~ /^([+-]?)([1-9][0-9]*[1-9])$/)
150 $self->{_e} = $MBI->_zero();
152 $self->{sign} = $1 || '+';
153 $self->{_m} = $MBI->_new($2);
154 return $self->round(@r) if !$downgrade;
157 my ($mis,$miv,$mfv,$es,$ev) = Math::BigInt::_split($wanted);
163 Carp::croak ("$wanted is not a number initialized to $class");
166 return $downgrade->bnan() if $downgrade;
168 $self->{_e} = $MBI->_zero();
170 $self->{_m} = $MBI->_zero();
171 $self->{sign} = $nan;
175 # make integer from mantissa by adjusting exp, then convert to int
176 $self->{_e} = $MBI->_new($$ev); # exponent
177 $self->{_es} = $$es || '+';
178 my $mantissa = "$$miv$$mfv"; # create mant.
179 $mantissa =~ s/^0+(\d)/$1/; # strip leading zeros
180 $self->{_m} = $MBI->_new($mantissa); # create mant.
182 # 3.123E0 = 3123E-3, and 3.123E-2 => 3123E-5
183 if (CORE::length($$mfv) != 0)
185 my $len = $MBI->_new( CORE::length($$mfv));
186 ($self->{_e}, $self->{_es}) =
187 _e_sub ($self->{_e}, $len, $self->{_es}, '+');
189 # we can only have trailing zeros on the mantissa if $$mfv eq ''
192 # Use a regexp to count the trailing zeros in $$miv instead of _zeros()
193 # because that is faster, especially when _m is not stored in base 10.
194 my $zeros = 0; $zeros = CORE::length($1) if $$miv =~ /[1-9](0*)$/;
197 my $z = $MBI->_new($zeros);
198 # turn '120e2' into '12e3'
199 $MBI->_rsft ( $self->{_m}, $z, 10);
200 _e_add ( $self->{_e}, $z, $self->{_es}, '+');
203 $self->{sign} = $$mis;
205 # for something like 0Ey, set y to 1, and -0 => +0
206 # Check $$miv for beeing '0' and $$mfv eq '', because otherwise _m could not
207 # have become 0. That's faster than to call $MBI->_is_zero().
208 $self->{sign} = '+', $self->{_e} = $MBI->_one()
209 if $$miv eq '0' and $$mfv eq '';
211 return $self->round(@r) if !$downgrade;
213 # if downgrade, inf, NaN or integers go down
215 if ($downgrade && $self->{_es} eq '+')
217 if ($MBI->_is_zero( $self->{_e} ))
219 return $downgrade->new($$mis . $MBI->_str( $self->{_m} ));
221 return $downgrade->new($self->bsstr());
223 $self->bnorm()->round(@r); # first normalize, then round
231 # if two arguments, the first one is the class to "swallow" subclasses
239 return unless ref($x); # only for objects
241 my $self = {}; bless $self,$c;
243 $self->{sign} = $x->{sign};
244 $self->{_es} = $x->{_es};
245 $self->{_m} = $MBI->_copy($x->{_m});
246 $self->{_e} = $MBI->_copy($x->{_e});
247 $self->{_a} = $x->{_a} if defined $x->{_a};
248 $self->{_p} = $x->{_p} if defined $x->{_p};
254 # used by parent class bone() to initialize number to NaN
260 my $class = ref($self);
261 Carp::croak ("Tried to set $self to NaN in $class\::_bnan()");
264 $IMPORT=1; # call our import only once
265 $self->{_m} = $MBI->_zero();
266 $self->{_e} = $MBI->_zero();
272 # used by parent class bone() to initialize number to +-inf
278 my $class = ref($self);
279 Carp::croak ("Tried to set $self to +-inf in $class\::_binf()");
282 $IMPORT=1; # call our import only once
283 $self->{_m} = $MBI->_zero();
284 $self->{_e} = $MBI->_zero();
290 # used by parent class bone() to initialize number to 1
292 $IMPORT=1; # call our import only once
293 $self->{_m} = $MBI->_one();
294 $self->{_e} = $MBI->_zero();
300 # used by parent class bone() to initialize number to 0
302 $IMPORT=1; # call our import only once
303 $self->{_m} = $MBI->_zero();
304 $self->{_e} = $MBI->_one();
310 my ($self,$class) = @_;
311 return if $class =~ /^Math::BigInt/; # we aren't one of these
312 UNIVERSAL::isa($self,$class);
317 # return (later set?) configuration data as hash ref
318 my $class = shift || 'Math::BigFloat';
320 my $cfg = $class->SUPER::config(@_);
322 # now we need only to override the ones that are different from our parent
323 $cfg->{class} = $class;
328 ##############################################################################
329 # string conversation
333 # (ref to BFLOAT or num_str ) return num_str
334 # Convert number from internal format to (non-scientific) string format.
335 # internal format is always normalized (no leading zeros, "-0" => "+0")
336 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
338 if ($x->{sign} !~ /^[+-]$/)
340 return $x->{sign} unless $x->{sign} eq '+inf'; # -inf, NaN
344 my $es = '0'; my $len = 1; my $cad = 0; my $dot = '.';
347 my $not_zero = !($x->{sign} eq '+' && $MBI->_is_zero($x->{_m}));
350 $es = $MBI->_str($x->{_m});
351 $len = CORE::length($es);
352 my $e = $MBI->_num($x->{_e});
353 $e = -$e if $x->{_es} eq '-';
357 # if _e is bigger than a scalar, the following will blow your memory
360 my $r = abs($e) - $len;
361 $es = '0.'. ('0' x $r) . $es; $cad = -($len+$r);
365 substr($es,$e,0) = '.'; $cad = $MBI->_num($x->{_e});
366 $cad = -$cad if $x->{_es} eq '-';
372 $es .= '0' x $e; $len += $e; $cad = 0;
376 $es = '-'.$es if $x->{sign} eq '-';
377 # if set accuracy or precision, pad with zeros on the right side
378 if ((defined $x->{_a}) && ($not_zero))
380 # 123400 => 6, 0.1234 => 4, 0.001234 => 4
381 my $zeros = $x->{_a} - $cad; # cad == 0 => 12340
382 $zeros = $x->{_a} - $len if $cad != $len;
383 $es .= $dot.'0' x $zeros if $zeros > 0;
385 elsif ((($x->{_p} || 0) < 0))
387 # 123400 => 6, 0.1234 => 4, 0.001234 => 6
388 my $zeros = -$x->{_p} + $cad;
389 $es .= $dot.'0' x $zeros if $zeros > 0;
396 # (ref to BFLOAT or num_str ) return num_str
397 # Convert number from internal format to scientific string format.
398 # internal format is always normalized (no leading zeros, "-0E0" => "+0E0")
399 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
401 if ($x->{sign} !~ /^[+-]$/)
403 return $x->{sign} unless $x->{sign} eq '+inf'; # -inf, NaN
406 my $sep = 'e'.$x->{_es};
407 my $sign = $x->{sign}; $sign = '' if $sign eq '+';
408 $sign . $MBI->_str($x->{_m}) . $sep . $MBI->_str($x->{_e});
413 # Make a number from a BigFloat object
414 # simple return a string and let Perl's atoi()/atof() handle the rest
415 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
419 ##############################################################################
420 # public stuff (usually prefixed with "b")
424 # (BINT or num_str) return BINT
425 # negate number or make a negated number from string
426 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
428 return $x if $x->modify('bneg');
430 # for +0 dont negate (to have always normalized +0). Does nothing for 'NaN'
431 $x->{sign} =~ tr/+-/-+/ unless ($x->{sign} eq '+' && $MBI->_is_zero($x->{_m}));
436 # XXX TODO this must be overwritten and return NaN for non-integer values
437 # band(), bior(), bxor(), too
440 # $class->SUPER::bnot($class,@_);
445 # Compares 2 values. Returns one of undef, <0, =0, >0. (suitable for sort)
448 my ($self,$x,$y) = (ref($_[0]),@_);
449 # objectify is costly, so avoid it
450 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
452 ($self,$x,$y) = objectify(2,@_);
455 return $upgrade->bcmp($x,$y) if defined $upgrade &&
456 ((!$x->isa($self)) || (!$y->isa($self)));
458 if (($x->{sign} !~ /^[+-]$/) || ($y->{sign} !~ /^[+-]$/))
460 # handle +-inf and NaN
461 return undef if (($x->{sign} eq $nan) || ($y->{sign} eq $nan));
462 return 0 if ($x->{sign} eq $y->{sign}) && ($x->{sign} =~ /^[+-]inf$/);
463 return +1 if $x->{sign} eq '+inf';
464 return -1 if $x->{sign} eq '-inf';
465 return -1 if $y->{sign} eq '+inf';
469 # check sign for speed first
470 return 1 if $x->{sign} eq '+' && $y->{sign} eq '-'; # does also 0 <=> -y
471 return -1 if $x->{sign} eq '-' && $y->{sign} eq '+'; # does also -x <=> 0
474 my $xz = $x->is_zero();
475 my $yz = $y->is_zero();
476 return 0 if $xz && $yz; # 0 <=> 0
477 return -1 if $xz && $y->{sign} eq '+'; # 0 <=> +y
478 return 1 if $yz && $x->{sign} eq '+'; # +x <=> 0
480 # adjust so that exponents are equal
481 my $lxm = $MBI->_len($x->{_m});
482 my $lym = $MBI->_len($y->{_m});
483 # the numify somewhat limits our length, but makes it much faster
484 my ($xes,$yes) = (1,1);
485 $xes = -1 if $x->{_es} ne '+';
486 $yes = -1 if $y->{_es} ne '+';
487 my $lx = $lxm + $xes * $MBI->_num($x->{_e});
488 my $ly = $lym + $yes * $MBI->_num($y->{_e});
489 my $l = $lx - $ly; $l = -$l if $x->{sign} eq '-';
490 return $l <=> 0 if $l != 0;
492 # lengths (corrected by exponent) are equal
493 # so make mantissa equal length by padding with zero (shift left)
494 my $diff = $lxm - $lym;
495 my $xm = $x->{_m}; # not yet copy it
499 $ym = $MBI->_copy($y->{_m});
500 $ym = $MBI->_lsft($ym, $MBI->_new($diff), 10);
504 $xm = $MBI->_copy($x->{_m});
505 $xm = $MBI->_lsft($xm, $MBI->_new(-$diff), 10);
507 my $rc = $MBI->_acmp($xm,$ym);
508 $rc = -$rc if $x->{sign} eq '-'; # -124 < -123
514 # Compares 2 values, ignoring their signs.
515 # Returns one of undef, <0, =0, >0. (suitable for sort)
518 my ($self,$x,$y) = (ref($_[0]),@_);
519 # objectify is costly, so avoid it
520 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
522 ($self,$x,$y) = objectify(2,@_);
525 return $upgrade->bacmp($x,$y) if defined $upgrade &&
526 ((!$x->isa($self)) || (!$y->isa($self)));
528 # handle +-inf and NaN's
529 if ($x->{sign} !~ /^[+-]$/ || $y->{sign} !~ /^[+-]$/)
531 return undef if (($x->{sign} eq $nan) || ($y->{sign} eq $nan));
532 return 0 if ($x->is_inf() && $y->is_inf());
533 return 1 if ($x->is_inf() && !$y->is_inf());
538 my $xz = $x->is_zero();
539 my $yz = $y->is_zero();
540 return 0 if $xz && $yz; # 0 <=> 0
541 return -1 if $xz && !$yz; # 0 <=> +y
542 return 1 if $yz && !$xz; # +x <=> 0
544 # adjust so that exponents are equal
545 my $lxm = $MBI->_len($x->{_m});
546 my $lym = $MBI->_len($y->{_m});
547 my ($xes,$yes) = (1,1);
548 $xes = -1 if $x->{_es} ne '+';
549 $yes = -1 if $y->{_es} ne '+';
550 # the numify somewhat limits our length, but makes it much faster
551 my $lx = $lxm + $xes * $MBI->_num($x->{_e});
552 my $ly = $lym + $yes * $MBI->_num($y->{_e});
554 return $l <=> 0 if $l != 0;
556 # lengths (corrected by exponent) are equal
557 # so make mantissa equal-length by padding with zero (shift left)
558 my $diff = $lxm - $lym;
559 my $xm = $x->{_m}; # not yet copy it
563 $ym = $MBI->_copy($y->{_m});
564 $ym = $MBI->_lsft($ym, $MBI->_new($diff), 10);
568 $xm = $MBI->_copy($x->{_m});
569 $xm = $MBI->_lsft($xm, $MBI->_new(-$diff), 10);
571 $MBI->_acmp($xm,$ym);
576 # add second arg (BFLOAT or string) to first (BFLOAT) (modifies first)
577 # return result as BFLOAT
580 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
581 # objectify is costly, so avoid it
582 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
584 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
587 # inf and NaN handling
588 if (($x->{sign} !~ /^[+-]$/) || ($y->{sign} !~ /^[+-]$/))
591 return $x->bnan() if (($x->{sign} eq $nan) || ($y->{sign} eq $nan));
593 if (($x->{sign} =~ /^[+-]inf$/) && ($y->{sign} =~ /^[+-]inf$/))
595 # +inf++inf or -inf+-inf => same, rest is NaN
596 return $x if $x->{sign} eq $y->{sign};
599 # +-inf + something => +inf; something +-inf => +-inf
600 $x->{sign} = $y->{sign}, return $x if $y->{sign} =~ /^[+-]inf$/;
604 return $upgrade->badd($x,$y,$a,$p,$r) if defined $upgrade &&
605 ((!$x->isa($self)) || (!$y->isa($self)));
607 # speed: no add for 0+y or x+0
608 return $x->bround($a,$p,$r) if $y->is_zero(); # x+0
609 if ($x->is_zero()) # 0+y
611 # make copy, clobbering up x (modify in place!)
612 $x->{_e} = $MBI->_copy($y->{_e});
613 $x->{_es} = $y->{_es};
614 $x->{_m} = $MBI->_copy($y->{_m});
615 $x->{sign} = $y->{sign} || $nan;
616 return $x->round($a,$p,$r,$y);
619 # take lower of the two e's and adapt m1 to it to match m2
621 $e = $MBI->_zero() if !defined $e; # if no BFLOAT?
622 $e = $MBI->_copy($e); # make copy (didn't do it yet)
626 ($e,$es) = _e_sub($e, $x->{_e}, $y->{_es} || '+', $x->{_es});
628 my $add = $MBI->_copy($y->{_m});
630 if ($es eq '-') # < 0
632 $MBI->_lsft( $x->{_m}, $e, 10);
633 ($x->{_e},$x->{_es}) = _e_add($x->{_e}, $e, $x->{_es}, $es);
635 elsif (!$MBI->_is_zero($e)) # > 0
637 $MBI->_lsft($add, $e, 10);
639 # else: both e are the same, so just leave them
641 if ($x->{sign} eq $y->{sign})
644 $x->{_m} = $MBI->_add($x->{_m}, $add);
648 ($x->{_m}, $x->{sign}) =
649 _e_add($x->{_m}, $add, $x->{sign}, $y->{sign});
652 # delete trailing zeros, then round
653 $x->bnorm()->round($a,$p,$r,$y);
656 # sub bsub is inherited from Math::BigInt!
660 # increment arg by one
661 my ($self,$x,@r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
663 if ($x->{_es} eq '-')
665 return $x->badd($self->bone(),@r); # digits after dot
668 if (!$MBI->_is_zero($x->{_e})) # _e == 0 for NaN, inf, -inf
670 # 1e2 => 100, so after the shift below _m has a '0' as last digit
671 $x->{_m} = $MBI->_lsft($x->{_m}, $x->{_e},10); # 1e2 => 100
672 $x->{_e} = $MBI->_zero(); # normalize
674 # we know that the last digit of $x will be '1' or '9', depending on the
678 if ($x->{sign} eq '+')
680 $MBI->_inc($x->{_m});
681 return $x->bnorm()->bround(@r);
683 elsif ($x->{sign} eq '-')
685 $MBI->_dec($x->{_m});
686 $x->{sign} = '+' if $MBI->_is_zero($x->{_m}); # -1 +1 => -0 => +0
687 return $x->bnorm()->bround(@r);
689 # inf, nan handling etc
690 $x->badd($self->bone(),@r); # badd() does round
695 # decrement arg by one
696 my ($self,$x,@r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
698 if ($x->{_es} eq '-')
700 return $x->badd($self->bone('-'),@r); # digits after dot
703 if (!$MBI->_is_zero($x->{_e}))
705 $x->{_m} = $MBI->_lsft($x->{_m}, $x->{_e},10); # 1e2 => 100
706 $x->{_e} = $MBI->_zero(); # normalize
710 my $zero = $x->is_zero();
712 if (($x->{sign} eq '-') || $zero)
714 $MBI->_inc($x->{_m});
715 $x->{sign} = '-' if $zero; # 0 => 1 => -1
716 $x->{sign} = '+' if $MBI->_is_zero($x->{_m}); # -1 +1 => -0 => +0
717 return $x->bnorm()->round(@r);
720 elsif ($x->{sign} eq '+')
722 $MBI->_dec($x->{_m});
723 return $x->bnorm()->round(@r);
725 # inf, nan handling etc
726 $x->badd($self->bone('-'),@r); # does round
733 my ($self,$x,$base,$a,$p,$r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
735 # $base > 0, $base != 1; if $base == undef default to $base == e
738 # we need to limit the accuracy to protect against overflow
741 ($x,@params) = $x->_find_round_parameters($a,$p,$r);
743 # also takes care of the "error in _find_round_parameters?" case
744 return $x->bnan() if $x->{sign} ne '+' || $x->is_zero();
747 # no rounding at all, so must use fallback
748 if (scalar @params == 0)
750 # simulate old behaviour
751 $params[0] = $self->div_scale(); # and round to it as accuracy
752 $params[1] = undef; # P = undef
753 $scale = $params[0]+4; # at least four more for proper round
754 $params[2] = $r; # round mode by caller or undef
755 $fallback = 1; # to clear a/p afterwards
759 # the 4 below is empirical, and there might be cases where it is not
761 $scale = abs($params[0] || $params[1]) + 4; # take whatever is defined
764 return $x->bzero(@params) if $x->is_one();
765 # base not defined => base == Euler's constant e
768 # make object, since we don't feed it through objectify() to still get the
769 # case of $base == undef
770 $base = $self->new($base) unless ref($base);
771 # $base > 0; $base != 1
772 return $x->bnan() if $base->is_zero() || $base->is_one() ||
773 $base->{sign} ne '+';
774 # if $x == $base, we know the result must be 1.0
775 if ($x->bcmp($base) == 0)
777 $x->bone('+',@params);
780 # clear a/p after round, since user did not request it
781 delete $x->{_a}; delete $x->{_p};
787 # when user set globals, they would interfere with our calculation, so
788 # disable them and later re-enable them
790 my $abr = "$self\::accuracy"; my $ab = $$abr; $$abr = undef;
791 my $pbr = "$self\::precision"; my $pb = $$pbr; $$pbr = undef;
792 # we also need to disable any set A or P on $x (_find_round_parameters took
793 # them already into account), since these would interfere, too
794 delete $x->{_a}; delete $x->{_p};
795 # need to disable $upgrade in BigInt, to avoid deep recursion
796 local $Math::BigInt::upgrade = undef;
797 local $Math::BigFloat::downgrade = undef;
799 # upgrade $x if $x is not a BigFloat (handle BigInt input)
800 if (!$x->isa('Math::BigFloat'))
802 $x = Math::BigFloat->new($x);
808 # If the base is defined and an integer, try to calculate integer result
809 # first. This is very fast, and in case the real result was found, we can
811 if (defined $base && $base->is_int() && $x->is_int())
813 my $i = $MBI->_copy( $x->{_m} );
814 $MBI->_lsft( $i, $x->{_e}, 10 ) unless $MBI->_is_zero($x->{_e});
815 my $int = Math::BigInt->bzero();
817 $int->blog($base->as_number());
819 if ($base->as_number()->bpow($int) == $x)
821 # found result, return it
822 $x->{_m} = $int->{value};
823 $x->{_e} = $MBI->_zero();
832 # first calculate the log to base e (using reduction by 10 (and probably 2))
833 $self->_log_10($x,$scale);
835 # and if a different base was requested, convert it
838 $base = Math::BigFloat->new($base) unless $base->isa('Math::BigFloat');
839 # not ln, but some other base (don't modify $base)
840 $x->bdiv( $base->copy()->blog(undef,$scale), $scale );
844 # shortcut to not run through _find_round_parameters again
845 if (defined $params[0])
847 $x->bround($params[0],$params[2]); # then round accordingly
851 $x->bfround($params[1],$params[2]); # then round accordingly
855 # clear a/p after round, since user did not request it
856 delete $x->{_a}; delete $x->{_p};
859 $$abr = $ab; $$pbr = $pb;
866 # internal log function to calculate ln() based on Taylor series.
867 # Modifies $x in place.
868 my ($self,$x,$scale) = @_;
870 # in case of $x == 1, result is 0
871 return $x->bzero() if $x->is_one();
873 # http://www.efunda.com/math/taylor_series/logarithmic.cfm?search_string=log
877 # Taylor: | u 1 u^3 1 u^5 |
878 # ln (x) = 2 | --- + - * --- + - * --- + ... | x > 0
879 # |_ v 3 v^3 5 v^5 _|
881 # This takes much more steps to calculate the result and is thus not used
884 # Taylor: | u 1 u^2 1 u^3 |
885 # ln (x) = 2 | --- + - * --- + - * --- + ... | x > 1/2
886 # |_ x 2 x^2 3 x^3 _|
888 my ($limit,$v,$u,$below,$factor,$two,$next,$over,$f);
890 $v = $x->copy(); $v->binc(); # v = x+1
891 $x->bdec(); $u = $x->copy(); # u = x-1; x = x-1
892 $x->bdiv($v,$scale); # first term: u/v
895 $u *= $u; $v *= $v; # u^2, v^2
896 $below->bmul($v); # u^3, v^3
898 $factor = $self->new(3); $f = $self->new(2);
900 my $steps = 0 if DEBUG;
901 $limit = $self->new("1E-". ($scale-1));
904 # we calculate the next term, and add it to the last
905 # when the next term is below our limit, it won't affect the outcome
906 # anymore, so we stop
908 # calculating the next term simple from over/below will result in quite
909 # a time hog if the input has many digits, since over and below will
910 # accumulate more and more digits, and the result will also have many
911 # digits, but in the end it is rounded to $scale digits anyway. So if we
912 # round $over and $below first, we save a lot of time for the division
913 # (not with log(1.2345), but try log (123**123) to see what I mean. This
914 # can introduce a rounding error if the division result would be f.i.
915 # 0.1234500000001 and we round it to 5 digits it would become 0.12346, but
916 # if we truncated $over and $below we might get 0.12345. Does this matter
917 # for the end result? So we give $over and $below 4 more digits to be
918 # on the safe side (unscientific error handling as usual... :+D
920 $next = $over->copy->bround($scale+4)->bdiv(
921 $below->copy->bmul($factor)->bround($scale+4),
925 ## $next = $over->copy()->bdiv($below->copy()->bmul($factor),$scale);
927 last if $next->bacmp($limit) <= 0;
929 delete $next->{_a}; delete $next->{_p};
931 # calculate things for the next term
932 $over *= $u; $below *= $v; $factor->badd($f);
935 $steps++; print "step $steps = $x\n" if $steps % 10 == 0;
938 $x->bmul($f); # $x *= 2
939 print "took $steps steps\n" if DEBUG;
944 # Internal log function based on reducing input to the range of 0.1 .. 9.99
945 # and then "correcting" the result to the proper one. Modifies $x in place.
946 my ($self,$x,$scale) = @_;
948 # taking blog() from numbers greater than 10 takes a *very long* time, so we
949 # break the computation down into parts based on the observation that:
950 # blog(x*y) = blog(x) + blog(y)
951 # We set $y here to multiples of 10 so that $x is below 1 (the smaller $x is
952 # the faster it get's, especially because 2*$x takes about 10 times as long,
953 # so by dividing $x by 10 we make it at least factor 100 faster...)
955 # The same observation is valid for numbers smaller than 0.1 (e.g. computing
956 # log(1) is fastest, and the farther away we get from 1, the longer it takes)
957 # so we also 'break' this down by multiplying $x with 10 and subtract the
958 # log(10) afterwards to get the correct result.
960 # calculate nr of digits before dot
961 my $dbd = $MBI->_num($x->{_e});
962 $dbd = -$dbd if $x->{_es} eq '-';
963 $dbd += $MBI->_len($x->{_m});
965 # more than one digit (e.g. at least 10), but *not* exactly 10 to avoid
968 my $calc = 1; # do some calculation?
970 # disable the shortcut for 10, since we need log(10) and this would recurse
972 if ($x->{_es} eq '+' && $MBI->_is_one($x->{_e}) && $MBI->_is_one($x->{_m}))
974 $dbd = 0; # disable shortcut
975 # we can use the cached value in these cases
976 if ($scale <= $LOG_10_A)
978 $x->bzero(); $x->badd($LOG_10);
979 $calc = 0; # no need to calc, but round
984 # disable the shortcut for 2, since we maybe have it cached
985 if (($MBI->_is_zero($x->{_e}) && $MBI->_is_two($x->{_m})))
987 $dbd = 0; # disable shortcut
988 # we can use the cached value in these cases
989 if ($scale <= $LOG_2_A)
991 $x->bzero(); $x->badd($LOG_2);
992 $calc = 0; # no need to calc, but round
997 # if $x = 0.1, we know the result must be 0-log(10)
998 if ($calc != 0 && $x->{_es} eq '-' && $MBI->_is_one($x->{_e}) &&
999 $MBI->_is_one($x->{_m}))
1001 $dbd = 0; # disable shortcut
1002 # we can use the cached value in these cases
1003 if ($scale <= $LOG_10_A)
1005 $x->bzero(); $x->bsub($LOG_10);
1006 $calc = 0; # no need to calc, but round
1010 return if $calc == 0; # already have the result
1012 # default: these correction factors are undef and thus not used
1013 my $l_10; # value of ln(10) to A of $scale
1014 my $l_2; # value of ln(2) to A of $scale
1016 # $x == 2 => 1, $x == 13 => 2, $x == 0.1 => 0, $x == 0.01 => -1
1017 # so don't do this shortcut for 1 or 0
1018 if (($dbd > 1) || ($dbd < 0))
1020 # convert our cached value to an object if not already (avoid doing this
1021 # at import() time, since not everybody needs this)
1022 $LOG_10 = $self->new($LOG_10,undef,undef) unless ref $LOG_10;
1024 #print "x = $x, dbd = $dbd, calc = $calc\n";
1025 # got more than one digit before the dot, or more than one zero after the
1027 # log(123) == log(1.23) + log(10) * 2
1028 # log(0.0123) == log(1.23) - log(10) * 2
1030 if ($scale <= $LOG_10_A)
1033 $l_10 = $LOG_10->copy(); # copy for mul
1037 # else: slower, compute it (but don't cache it, because it could be big)
1038 # also disable downgrade for this code path
1039 local $Math::BigFloat::downgrade = undef;
1040 $l_10 = $self->new(10)->blog(undef,$scale); # scale+4, actually
1042 $dbd-- if ($dbd > 1); # 20 => dbd=2, so make it dbd=1
1043 $l_10->bmul( $self->new($dbd)); # log(10) * (digits_before_dot-1)
1050 ($x->{_e}, $x->{_es}) =
1051 _e_sub( $x->{_e}, $MBI->_new($dbd), $x->{_es}, $dbd_sign); # 123 => 1.23
1055 # Now: 0.1 <= $x < 10 (and possible correction in l_10)
1057 ### Since $x in the range 0.5 .. 1.5 is MUCH faster, we do a repeated div
1058 ### or mul by 2 (maximum times 3, since x < 10 and x > 0.1)
1060 $HALF = $self->new($HALF) unless ref($HALF);
1062 my $twos = 0; # default: none (0 times)
1063 my $two = $self->new(2);
1064 while ($x->bacmp($HALF) <= 0)
1066 $twos--; $x->bmul($two);
1068 while ($x->bacmp($two) >= 0)
1070 $twos++; $x->bdiv($two,$scale+4); # keep all digits
1072 # $twos > 0 => did mul 2, < 0 => did div 2 (never both)
1073 # calculate correction factor based on ln(2)
1076 $LOG_2 = $self->new($LOG_2,undef,undef) unless ref $LOG_2;
1077 if ($scale <= $LOG_2_A)
1080 $l_2 = $LOG_2->copy(); # copy for mul
1084 # else: slower, compute it (but don't cache it, because it could be big)
1085 # also disable downgrade for this code path
1086 local $Math::BigFloat::downgrade = undef;
1087 $l_2 = $two->blog(undef,$scale); # scale+4, actually
1089 $l_2->bmul($twos); # * -2 => subtract, * 2 => add
1092 $self->_log($x,$scale); # need to do the "normal" way
1093 $x->badd($l_10) if defined $l_10; # correct it by ln(10)
1094 $x->badd($l_2) if defined $l_2; # and maybe by ln(2)
1095 # all done, $x contains now the result
1100 # (BFLOAT or num_str, BFLOAT or num_str) return BFLOAT
1101 # does not modify arguments, but returns new object
1102 # Lowest Common Multiplicator
1104 my ($self,@arg) = objectify(0,@_);
1105 my $x = $self->new(shift @arg);
1106 while (@arg) { $x = Math::BigInt::__lcm($x,shift @arg); }
1112 # (BINT or num_str, BINT or num_str) return BINT
1113 # does not modify arguments, but returns new object
1116 $y = __PACKAGE__->new($y) if !ref($y);
1118 my $x = $y->copy()->babs(); # keep arguments
1120 return $x->bnan() if $x->{sign} !~ /^[+-]$/ # x NaN?
1121 || !$x->is_int(); # only for integers now
1125 my $t = shift; $t = $self->new($t) if !ref($t);
1126 $y = $t->copy()->babs();
1128 return $x->bnan() if $y->{sign} !~ /^[+-]$/ # y NaN?
1129 || !$y->is_int(); # only for integers now
1131 # greatest common divisor
1132 while (! $y->is_zero())
1134 ($x,$y) = ($y->copy(), $x->copy()->bmod($y));
1137 last if $x->is_one();
1142 ##############################################################################
1146 # Internal helper sub to take two positive integers and their signs and
1147 # then add them. Input ($CALC,$CALC,('+'|'-'),('+'|'-')),
1148 # output ($CALC,('+'|'-'))
1149 my ($x,$y,$xs,$ys) = @_;
1151 # if the signs are equal we can add them (-5 + -3 => -(5 + 3) => -8)
1154 $x = $MBI->_add ($x, $y ); # a+b
1155 # the sign follows $xs
1159 my $a = $MBI->_acmp($x,$y);
1162 $x = $MBI->_sub ($x , $y); # abs sub
1166 $x = $MBI->_zero(); # result is 0
1171 $x = $MBI->_sub ( $y, $x, 1 ); # abs sub
1179 # Internal helper sub to take two positive integers and their signs and
1180 # then subtract them. Input ($CALC,$CALC,('+'|'-'),('+'|'-')),
1181 # output ($CALC,('+'|'-'))
1182 my ($x,$y,$xs,$ys) = @_;
1186 _e_add($x,$y,$xs,$ys); # call add (does subtract now)
1189 ###############################################################################
1190 # is_foo methods (is_negative, is_positive are inherited from BigInt)
1194 # return true if arg (BFLOAT or num_str) is an integer
1195 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
1197 return 1 if ($x->{sign} =~ /^[+-]$/) && # NaN and +-inf aren't
1198 $x->{_es} eq '+'; # 1e-1 => no integer
1204 # return true if arg (BFLOAT or num_str) is zero
1205 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
1207 return 1 if $x->{sign} eq '+' && $MBI->_is_zero($x->{_m});
1213 # return true if arg (BFLOAT or num_str) is +1 or -1 if signis given
1214 my ($self,$x,$sign) = ref($_[0]) ? (undef,@_) : objectify(1,@_);
1216 $sign = '+' if !defined $sign || $sign ne '-';
1218 if ($x->{sign} eq $sign &&
1219 $MBI->_is_zero($x->{_e}) && $MBI->_is_one($x->{_m}));
1225 # return true if arg (BFLOAT or num_str) is odd or false if even
1226 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
1228 return 1 if ($x->{sign} =~ /^[+-]$/) && # NaN & +-inf aren't
1229 ($MBI->_is_zero($x->{_e}) && $MBI->_is_odd($x->{_m}));
1235 # return true if arg (BINT or num_str) is even or false if odd
1236 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
1238 return 0 if $x->{sign} !~ /^[+-]$/; # NaN & +-inf aren't
1239 return 1 if ($x->{_es} eq '+' # 123.45 is never
1240 && $MBI->_is_even($x->{_m})); # but 1200 is
1246 # multiply two numbers -- stolen from Knuth Vol 2 pg 233
1247 # (BINT or num_str, BINT or num_str) return BINT
1250 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
1251 # objectify is costly, so avoid it
1252 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
1254 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
1257 return $x->bnan() if (($x->{sign} eq $nan) || ($y->{sign} eq $nan));
1260 if (($x->{sign} =~ /^[+-]inf$/) || ($y->{sign} =~ /^[+-]inf$/))
1262 return $x->bnan() if $x->is_zero() || $y->is_zero();
1263 # result will always be +-inf:
1264 # +inf * +/+inf => +inf, -inf * -/-inf => +inf
1265 # +inf * -/-inf => -inf, -inf * +/+inf => -inf
1266 return $x->binf() if ($x->{sign} =~ /^\+/ && $y->{sign} =~ /^\+/);
1267 return $x->binf() if ($x->{sign} =~ /^-/ && $y->{sign} =~ /^-/);
1268 return $x->binf('-');
1271 return $x->bzero() if $x->is_zero() || $y->is_zero();
1273 return $upgrade->bmul($x,$y,$a,$p,$r) if defined $upgrade &&
1274 ((!$x->isa($self)) || (!$y->isa($self)));
1276 # aEb * cEd = (a*c)E(b+d)
1277 $MBI->_mul($x->{_m},$y->{_m});
1278 ($x->{_e}, $x->{_es}) = _e_add($x->{_e}, $y->{_e}, $x->{_es}, $y->{_es});
1281 $x->{sign} = $x->{sign} ne $y->{sign} ? '-' : '+';
1282 return $x->bnorm()->round($a,$p,$r,$y);
1287 # (dividend: BFLOAT or num_str, divisor: BFLOAT or num_str) return
1288 # (BFLOAT,BFLOAT) (quo,rem) or BFLOAT (only rem)
1291 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
1292 # objectify is costly, so avoid it
1293 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
1295 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
1298 return $self->_div_inf($x,$y)
1299 if (($x->{sign} !~ /^[+-]$/) || ($y->{sign} !~ /^[+-]$/) || $y->is_zero());
1301 # x== 0 # also: or y == 1 or y == -1
1302 return wantarray ? ($x,$self->bzero()) : $x if $x->is_zero();
1305 return $upgrade->bdiv($upgrade->new($x),$y,$a,$p,$r) if defined $upgrade;
1307 # we need to limit the accuracy to protect against overflow
1309 my (@params,$scale);
1310 ($x,@params) = $x->_find_round_parameters($a,$p,$r,$y);
1312 return $x if $x->is_nan(); # error in _find_round_parameters?
1314 # no rounding at all, so must use fallback
1315 if (scalar @params == 0)
1317 # simulate old behaviour
1318 $params[0] = $self->div_scale(); # and round to it as accuracy
1319 $scale = $params[0]+4; # at least four more for proper round
1320 $params[2] = $r; # round mode by caller or undef
1321 $fallback = 1; # to clear a/p afterwards
1325 # the 4 below is empirical, and there might be cases where it is not
1327 $scale = abs($params[0] || $params[1]) + 4; # take whatever is defined
1330 my $rem; $rem = $self->bzero() if wantarray;
1332 $y = $self->new($y) unless $y->isa('Math::BigFloat');
1334 my $lx = $MBI->_len($x->{_m}); my $ly = $MBI->_len($y->{_m});
1335 $scale = $lx if $lx > $scale;
1336 $scale = $ly if $ly > $scale;
1337 my $diff = $ly - $lx;
1338 $scale += $diff if $diff > 0; # if lx << ly, but not if ly << lx!
1340 # already handled inf/NaN/-inf above:
1342 my $xsign = $x->{sign};
1343 $y->{sign} =~ tr/+-/-+/;
1345 # check that $y is not 1 nor -1 and cache the result:
1346 my $y_not_one = !($MBI->_is_zero($y->{_e}) && $MBI->_is_one($y->{_m}));
1348 if ($xsign ne $x->{sign})
1350 # special case of $x /= $x results in 1
1351 $x->bone(); # "fixes" also sign of $y, since $x is $y
1355 # correct $y's sign again
1356 $y->{sign} =~ tr/+-/-+/;
1357 # continue with normal div code:
1359 # make copy of $x in case of list context for later reminder calculation
1360 if (wantarray && $y_not_one)
1365 $x->{sign} = $x->{sign} ne $y->sign() ? '-' : '+';
1367 # check for / +-1 ( +/- 1E0)
1370 # promote BigInts and it's subclasses (except when already a BigFloat)
1371 $y = $self->new($y) unless $y->isa('Math::BigFloat');
1373 # calculate the result to $scale digits and then round it
1374 # a * 10 ** b / c * 10 ** d => a/c * 10 ** (b-d)
1375 $MBI->_lsft($x->{_m},$MBI->_new($scale),10);
1376 $MBI->_div ($x->{_m},$y->{_m}); # a/c
1378 # correct exponent of $x
1379 ($x->{_e},$x->{_es}) = _e_sub($x->{_e}, $y->{_e}, $x->{_es}, $y->{_es});
1380 # correct for 10**scale
1381 ($x->{_e},$x->{_es}) = _e_sub($x->{_e}, $MBI->_new($scale), $x->{_es}, '+');
1382 $x->bnorm(); # remove trailing 0's
1384 } # ende else $x != $y
1386 # shortcut to not run through _find_round_parameters again
1387 if (defined $params[0])
1389 delete $x->{_a}; # clear before round
1390 $x->bround($params[0],$params[2]); # then round accordingly
1394 delete $x->{_p}; # clear before round
1395 $x->bfround($params[1],$params[2]); # then round accordingly
1399 # clear a/p after round, since user did not request it
1400 delete $x->{_a}; delete $x->{_p};
1407 $rem->bmod($y,@params); # copy already done
1411 # clear a/p after round, since user did not request it
1412 delete $rem->{_a}; delete $rem->{_p};
1421 # (dividend: BFLOAT or num_str, divisor: BFLOAT or num_str) return reminder
1424 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
1425 # objectify is costly, so avoid it
1426 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
1428 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
1431 # handle NaN, inf, -inf
1432 if (($x->{sign} !~ /^[+-]$/) || ($y->{sign} !~ /^[+-]$/))
1434 my ($d,$re) = $self->SUPER::_div_inf($x,$y);
1435 $x->{sign} = $re->{sign};
1436 $x->{_e} = $re->{_e};
1437 $x->{_m} = $re->{_m};
1438 return $x->round($a,$p,$r,$y);
1442 return $x->bnan() if $x->is_zero();
1446 return $x->bzero() if $x->is_zero() ||
1447 # check that $y == -1 or +1:
1448 ($MBI->_is_zero($y->{_e}) && $MBI->_is_one($y->{_m}));
1450 my $cmp = $x->bacmp($y); # equal or $x < $y?
1451 return $x->bzero($a,$p) if $cmp == 0; # $x == $y => result 0
1453 # only $y of the operands negative?
1454 my $neg = 0; $neg = 1 if $x->{sign} ne $y->{sign};
1456 $x->{sign} = $y->{sign}; # calc sign first
1457 return $x->round($a,$p,$r) if $cmp < 0 && $neg == 0; # $x < $y => result $x
1459 my $ym = $MBI->_copy($y->{_m});
1462 $MBI->_lsft( $ym, $y->{_e}, 10)
1463 if $y->{_es} eq '+' && !$MBI->_is_zero($y->{_e});
1465 # if $y has digits after dot
1466 my $shifty = 0; # correct _e of $x by this
1467 if ($y->{_es} eq '-') # has digits after dot
1469 # 123 % 2.5 => 1230 % 25 => 5 => 0.5
1470 $shifty = $MBI->_num($y->{_e}); # no more digits after dot
1471 $MBI->_lsft($x->{_m}, $y->{_e}, 10);# 123 => 1230, $y->{_m} is already 25
1473 # $ym is now mantissa of $y based on exponent 0
1475 my $shiftx = 0; # correct _e of $x by this
1476 if ($x->{_es} eq '-') # has digits after dot
1478 # 123.4 % 20 => 1234 % 200
1479 $shiftx = $MBI->_num($x->{_e}); # no more digits after dot
1480 $MBI->_lsft($ym, $x->{_e}, 10); # 123 => 1230
1482 # 123e1 % 20 => 1230 % 20
1483 if ($x->{_es} eq '+' && !$MBI->_is_zero($x->{_e}))
1485 $MBI->_lsft( $x->{_m}, $x->{_e},10); # es => '+' here
1488 $x->{_e} = $MBI->_new($shiftx);
1490 $x->{_es} = '-' if $shiftx != 0 || $shifty != 0;
1491 $MBI->_add( $x->{_e}, $MBI->_new($shifty)) if $shifty != 0;
1493 # now mantissas are equalized, exponent of $x is adjusted, so calc result
1495 $x->{_m} = $MBI->_mod( $x->{_m}, $ym);
1497 $x->{sign} = '+' if $MBI->_is_zero($x->{_m}); # fix sign for -0
1500 if ($neg != 0) # one of them negative => correct in place
1503 $x->{_m} = $r->{_m};
1504 $x->{_e} = $r->{_e};
1505 $x->{_es} = $r->{_es};
1506 $x->{sign} = '+' if $MBI->_is_zero($x->{_m}); # fix sign for -0
1510 $x->round($a,$p,$r,$y); # round and return
1515 # calculate $y'th root of $x
1518 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
1519 # objectify is costly, so avoid it
1520 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
1522 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
1525 # NaN handling: $x ** 1/0, x or y NaN, or y inf/-inf or y == 0
1526 return $x->bnan() if $x->{sign} !~ /^\+/ || $y->is_zero() ||
1527 $y->{sign} !~ /^\+$/;
1529 return $x if $x->is_zero() || $x->is_one() || $x->is_inf() || $y->is_one();
1531 # we need to limit the accuracy to protect against overflow
1533 my (@params,$scale);
1534 ($x,@params) = $x->_find_round_parameters($a,$p,$r);
1536 return $x if $x->is_nan(); # error in _find_round_parameters?
1538 # no rounding at all, so must use fallback
1539 if (scalar @params == 0)
1541 # simulate old behaviour
1542 $params[0] = $self->div_scale(); # and round to it as accuracy
1543 $scale = $params[0]+4; # at least four more for proper round
1544 $params[2] = $r; # iound mode by caller or undef
1545 $fallback = 1; # to clear a/p afterwards
1549 # the 4 below is empirical, and there might be cases where it is not
1551 $scale = abs($params[0] || $params[1]) + 4; # take whatever is defined
1554 # when user set globals, they would interfere with our calculation, so
1555 # disable them and later re-enable them
1557 my $abr = "$self\::accuracy"; my $ab = $$abr; $$abr = undef;
1558 my $pbr = "$self\::precision"; my $pb = $$pbr; $$pbr = undef;
1559 # we also need to disable any set A or P on $x (_find_round_parameters took
1560 # them already into account), since these would interfere, too
1561 delete $x->{_a}; delete $x->{_p};
1562 # need to disable $upgrade in BigInt, to avoid deep recursion
1563 local $Math::BigInt::upgrade = undef; # should be really parent class vs MBI
1565 # remember sign and make $x positive, since -4 ** (1/2) => -2
1566 my $sign = 0; $sign = 1 if $x->{sign} eq '-'; $x->{sign} = '+';
1569 if ($y->isa('Math::BigFloat'))
1571 $is_two = ($y->{sign} eq '+' && $MBI->_is_two($y->{_m}) && $MBI->_is_zero($y->{_e}));
1575 $is_two = ($y == 2);
1578 # normal square root if $y == 2:
1581 $x->bsqrt($scale+4);
1583 elsif ($y->is_one('-'))
1586 my $u = $self->bone()->bdiv($x,$scale);
1587 # copy private parts over
1588 $x->{_m} = $u->{_m};
1589 $x->{_e} = $u->{_e};
1590 $x->{_es} = $u->{_es};
1594 # calculate the broot() as integer result first, and if it fits, return
1595 # it rightaway (but only if $x and $y are integer):
1597 my $done = 0; # not yet
1598 if ($y->is_int() && $x->is_int())
1600 my $i = $MBI->_copy( $x->{_m} );
1601 $MBI->_lsft( $i, $x->{_e}, 10 ) unless $MBI->_is_zero($x->{_e});
1602 my $int = Math::BigInt->bzero();
1604 $int->broot($y->as_number());
1606 if ($int->copy()->bpow($y) == $x)
1608 # found result, return it
1609 $x->{_m} = $int->{value};
1610 $x->{_e} = $MBI->_zero();
1618 my $u = $self->bone()->bdiv($y,$scale+4);
1619 delete $u->{_a}; delete $u->{_p}; # otherwise it conflicts
1620 $x->bpow($u,$scale+4); # el cheapo
1623 $x->bneg() if $sign == 1;
1625 # shortcut to not run through _find_round_parameters again
1626 if (defined $params[0])
1628 $x->bround($params[0],$params[2]); # then round accordingly
1632 $x->bfround($params[1],$params[2]); # then round accordingly
1636 # clear a/p after round, since user did not request it
1637 delete $x->{_a}; delete $x->{_p};
1640 $$abr = $ab; $$pbr = $pb;
1646 # calculate square root
1647 my ($self,$x,$a,$p,$r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
1649 return $x->bnan() if $x->{sign} !~ /^[+]/; # NaN, -inf or < 0
1650 return $x if $x->{sign} eq '+inf'; # sqrt(inf) == inf
1651 return $x->round($a,$p,$r) if $x->is_zero() || $x->is_one();
1653 # we need to limit the accuracy to protect against overflow
1655 my (@params,$scale);
1656 ($x,@params) = $x->_find_round_parameters($a,$p,$r);
1658 return $x if $x->is_nan(); # error in _find_round_parameters?
1660 # no rounding at all, so must use fallback
1661 if (scalar @params == 0)
1663 # simulate old behaviour
1664 $params[0] = $self->div_scale(); # and round to it as accuracy
1665 $scale = $params[0]+4; # at least four more for proper round
1666 $params[2] = $r; # round mode by caller or undef
1667 $fallback = 1; # to clear a/p afterwards
1671 # the 4 below is empirical, and there might be cases where it is not
1673 $scale = abs($params[0] || $params[1]) + 4; # take whatever is defined
1676 # when user set globals, they would interfere with our calculation, so
1677 # disable them and later re-enable them
1679 my $abr = "$self\::accuracy"; my $ab = $$abr; $$abr = undef;
1680 my $pbr = "$self\::precision"; my $pb = $$pbr; $$pbr = undef;
1681 # we also need to disable any set A or P on $x (_find_round_parameters took
1682 # them already into account), since these would interfere, too
1683 delete $x->{_a}; delete $x->{_p};
1684 # need to disable $upgrade in BigInt, to avoid deep recursion
1685 local $Math::BigInt::upgrade = undef; # should be really parent class vs MBI
1687 my $i = $MBI->_copy( $x->{_m} );
1688 $MBI->_lsft( $i, $x->{_e}, 10 ) unless $MBI->_is_zero($x->{_e});
1689 my $xas = Math::BigInt->bzero();
1692 my $gs = $xas->copy()->bsqrt(); # some guess
1694 if (($x->{_es} ne '-') # guess can't be accurate if there are
1695 # digits after the dot
1696 && ($xas->bacmp($gs * $gs) == 0)) # guess hit the nail on the head?
1698 # exact result, copy result over to keep $x
1699 $x->{_m} = $gs->{value}; $x->{_e} = $MBI->_zero(); $x->{_es} = '+';
1701 # shortcut to not run through _find_round_parameters again
1702 if (defined $params[0])
1704 $x->bround($params[0],$params[2]); # then round accordingly
1708 $x->bfround($params[1],$params[2]); # then round accordingly
1712 # clear a/p after round, since user did not request it
1713 delete $x->{_a}; delete $x->{_p};
1715 # re-enable A and P, upgrade is taken care of by "local"
1716 ${"$self\::accuracy"} = $ab; ${"$self\::precision"} = $pb;
1720 # sqrt(2) = 1.4 because sqrt(2*100) = 1.4*10; so we can increase the accuracy
1721 # of the result by multipyling the input by 100 and then divide the integer
1722 # result of sqrt(input) by 10. Rounding afterwards returns the real result.
1724 # The following steps will transform 123.456 (in $x) into 123456 (in $y1)
1725 my $y1 = $MBI->_copy($x->{_m});
1727 my $length = $MBI->_len($y1);
1729 # Now calculate how many digits the result of sqrt(y1) would have
1730 my $digits = int($length / 2);
1732 # But we need at least $scale digits, so calculate how many are missing
1733 my $shift = $scale - $digits;
1735 # That should never happen (we take care of integer guesses above)
1736 # $shift = 0 if $shift < 0;
1738 # Multiply in steps of 100, by shifting left two times the "missing" digits
1739 my $s2 = $shift * 2;
1741 # We now make sure that $y1 has the same odd or even number of digits than
1742 # $x had. So when _e of $x is odd, we must shift $y1 by one digit left,
1743 # because we always must multiply by steps of 100 (sqrt(100) is 10) and not
1744 # steps of 10. The length of $x does not count, since an even or odd number
1745 # of digits before the dot is not changed by adding an even number of digits
1746 # after the dot (the result is still odd or even digits long).
1747 $s2++ if $MBI->_is_odd($x->{_e});
1749 $MBI->_lsft( $y1, $MBI->_new($s2), 10);
1751 # now take the square root and truncate to integer
1752 $y1 = $MBI->_sqrt($y1);
1754 # By "shifting" $y1 right (by creating a negative _e) we calculate the final
1755 # result, which is than later rounded to the desired scale.
1757 # calculate how many zeros $x had after the '.' (or before it, depending
1758 # on sign of $dat, the result should have half as many:
1759 my $dat = $MBI->_num($x->{_e});
1760 $dat = -$dat if $x->{_es} eq '-';
1765 # no zeros after the dot (e.g. 1.23, 0.49 etc)
1766 # preserve half as many digits before the dot than the input had
1767 # (but round this "up")
1768 $dat = int(($dat+1)/2);
1772 $dat = int(($dat)/2);
1774 $dat -= $MBI->_len($y1);
1778 $x->{_e} = $MBI->_new( $dat );
1783 $x->{_e} = $MBI->_new( $dat );
1789 # shortcut to not run through _find_round_parameters again
1790 if (defined $params[0])
1792 $x->bround($params[0],$params[2]); # then round accordingly
1796 $x->bfround($params[1],$params[2]); # then round accordingly
1800 # clear a/p after round, since user did not request it
1801 delete $x->{_a}; delete $x->{_p};
1804 $$abr = $ab; $$pbr = $pb;
1810 # (BFLOAT or num_str, BFLOAT or num_str) return BFLOAT
1811 # compute factorial number, modifies first argument
1814 my ($self,$x,@r) = (ref($_[0]),@_);
1815 # objectify is costly, so avoid it
1816 ($self,$x,@r) = objectify(1,@_) if !ref($x);
1818 return $x if $x->{sign} eq '+inf'; # inf => inf
1820 if (($x->{sign} ne '+') || # inf, NaN, <0 etc => NaN
1821 ($x->{_es} ne '+')); # digits after dot?
1823 # use BigInt's bfac() for faster calc
1824 if (! $MBI->_is_zero($x->{_e}))
1826 $MBI->_lsft($x->{_m}, $x->{_e},10); # change 12e1 to 120e0
1827 $x->{_e} = $MBI->_zero(); # normalize
1830 $MBI->_fac($x->{_m}); # calculate factorial
1831 $x->bnorm()->round(@r); # norm again and round result
1836 # Calculate a power where $y is a non-integer, like 2 ** 0.5
1837 my ($x,$y,$a,$p,$r) = @_;
1840 # if $y == 0.5, it is sqrt($x)
1841 $HALF = $self->new($HALF) unless ref($HALF);
1842 return $x->bsqrt($a,$p,$r,$y) if $y->bcmp($HALF) == 0;
1845 # a ** x == e ** (x * ln a)
1849 # Taylor: | u u^2 u^3 |
1850 # x ** y = 1 + | --- + --- + ----- + ... |
1853 # we need to limit the accuracy to protect against overflow
1855 my ($scale,@params);
1856 ($x,@params) = $x->_find_round_parameters($a,$p,$r);
1858 return $x if $x->is_nan(); # error in _find_round_parameters?
1860 # no rounding at all, so must use fallback
1861 if (scalar @params == 0)
1863 # simulate old behaviour
1864 $params[0] = $self->div_scale(); # and round to it as accuracy
1865 $params[1] = undef; # disable P
1866 $scale = $params[0]+4; # at least four more for proper round
1867 $params[2] = $r; # round mode by caller or undef
1868 $fallback = 1; # to clear a/p afterwards
1872 # the 4 below is empirical, and there might be cases where it is not
1874 $scale = abs($params[0] || $params[1]) + 4; # take whatever is defined
1877 # when user set globals, they would interfere with our calculation, so
1878 # disable them and later re-enable them
1880 my $abr = "$self\::accuracy"; my $ab = $$abr; $$abr = undef;
1881 my $pbr = "$self\::precision"; my $pb = $$pbr; $$pbr = undef;
1882 # we also need to disable any set A or P on $x (_find_round_parameters took
1883 # them already into account), since these would interfere, too
1884 delete $x->{_a}; delete $x->{_p};
1885 # need to disable $upgrade in BigInt, to avoid deep recursion
1886 local $Math::BigInt::upgrade = undef;
1888 my ($limit,$v,$u,$below,$factor,$next,$over);
1890 $u = $x->copy()->blog(undef,$scale)->bmul($y);
1891 $v = $self->bone(); # 1
1892 $factor = $self->new(2); # 2
1893 $x->bone(); # first term: 1
1895 $below = $v->copy();
1898 $limit = $self->new("1E-". ($scale-1));
1902 # we calculate the next term, and add it to the last
1903 # when the next term is below our limit, it won't affect the outcome
1904 # anymore, so we stop
1905 $next = $over->copy()->bdiv($below,$scale);
1906 last if $next->bacmp($limit) <= 0;
1908 # calculate things for the next term
1909 $over *= $u; $below *= $factor; $factor->binc();
1911 last if $x->{sign} !~ /^[-+]$/;
1916 # shortcut to not run through _find_round_parameters again
1917 if (defined $params[0])
1919 $x->bround($params[0],$params[2]); # then round accordingly
1923 $x->bfround($params[1],$params[2]); # then round accordingly
1927 # clear a/p after round, since user did not request it
1928 delete $x->{_a}; delete $x->{_p};
1931 $$abr = $ab; $$pbr = $pb;
1937 # (BFLOAT or num_str, BFLOAT or num_str) return BFLOAT
1938 # compute power of two numbers, second arg is used as integer
1939 # modifies first argument
1942 my ($self,$x,$y,$a,$p,$r) = (ref($_[0]),@_);
1943 # objectify is costly, so avoid it
1944 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
1946 ($self,$x,$y,$a,$p,$r) = objectify(2,@_);
1949 return $x->bnan() if $x->{sign} eq $nan || $y->{sign} eq $nan;
1950 return $x if $x->{sign} =~ /^[+-]inf$/;
1953 return $x->bnan() if $x->{sign} eq '-' && $y->{sign} eq '-';
1955 # cache the result of is_zero
1956 my $y_is_zero = $y->is_zero();
1957 return $x->bone() if $y_is_zero;
1958 return $x if $x->is_one() || $y->is_one();
1960 my $x_is_zero = $x->is_zero();
1961 return $x->_pow($y,$a,$p,$r) if !$x_is_zero && !$y->is_int(); # non-integer power
1963 my $y1 = $y->as_number()->{value}; # make MBI part
1966 if ($x->{sign} eq '-' && $MBI->_is_one($x->{_m}) && $MBI->_is_zero($x->{_e}))
1968 # if $x == -1 and odd/even y => +1/-1 because +-1 ^ (+-1) => +-1
1969 return $MBI->_is_odd($y1) ? $x : $x->babs(1);
1973 return $x->bone() if $y_is_zero;
1974 return $x if $y->{sign} eq '+'; # 0**y => 0 (if not y <= 0)
1975 # 0 ** -y => 1 / (0 ** y) => 1 / 0! (1 / 0 => +inf)
1980 $new_sign = $MBI->_is_odd($y1) ? '-' : '+' if $x->{sign} ne '+';
1982 # calculate $x->{_m} ** $y and $x->{_e} * $y separately (faster)
1983 $x->{_m} = $MBI->_pow( $x->{_m}, $y1);
1984 $x->{_e} = $MBI->_mul ($x->{_e}, $y1);
1986 $x->{sign} = $new_sign;
1988 if ($y->{sign} eq '-')
1990 # modify $x in place!
1991 my $z = $x->copy(); $x->bone();
1992 return $x->bdiv($z,$a,$p,$r); # round in one go (might ignore y's A!)
1994 $x->round($a,$p,$r,$y);
1997 ###############################################################################
1998 # rounding functions
2002 # precision: round to the $Nth digit left (+$n) or right (-$n) from the '.'
2003 # $n == 0 means round to integer
2004 # expects and returns normalized numbers!
2005 my $x = shift; my $self = ref($x) || $x; $x = $self->new(shift) if !ref($x);
2007 my ($scale,$mode) = $x->_scale_p(@_);
2008 return $x if !defined $scale || $x->modify('bfround'); # no-op
2010 # never round a 0, +-inf, NaN
2013 $x->{_p} = $scale if !defined $x->{_p} || $x->{_p} < $scale; # -3 < -2
2016 return $x if $x->{sign} !~ /^[+-]$/;
2018 # don't round if x already has lower precision
2019 return $x if (defined $x->{_p} && $x->{_p} < 0 && $scale < $x->{_p});
2021 $x->{_p} = $scale; # remember round in any case
2022 delete $x->{_a}; # and clear A
2025 # round right from the '.'
2027 return $x if $x->{_es} eq '+'; # e >= 0 => nothing to round
2029 $scale = -$scale; # positive for simplicity
2030 my $len = $MBI->_len($x->{_m}); # length of mantissa
2032 # the following poses a restriction on _e, but if _e is bigger than a
2033 # scalar, you got other problems (memory etc) anyway
2034 my $dad = -(0+ ($x->{_es}.$MBI->_num($x->{_e}))); # digits after dot
2035 my $zad = 0; # zeros after dot
2036 $zad = $dad - $len if (-$dad < -$len); # for 0.00..00xxx style
2038 # p rint "scale $scale dad $dad zad $zad len $len\n";
2039 # number bsstr len zad dad
2040 # 0.123 123e-3 3 0 3
2041 # 0.0123 123e-4 3 1 4
2044 # 1.2345 12345e-4 5 0 4
2046 # do not round after/right of the $dad
2047 return $x if $scale > $dad; # 0.123, scale >= 3 => exit
2049 # round to zero if rounding inside the $zad, but not for last zero like:
2050 # 0.0065, scale -2, round last '0' with following '65' (scale == zad case)
2051 return $x->bzero() if $scale < $zad;
2052 if ($scale == $zad) # for 0.006, scale -3 and trunc
2058 # adjust round-point to be inside mantissa
2061 $scale = $scale-$zad;
2065 my $dbd = $len - $dad; $dbd = 0 if $dbd < 0; # digits before dot
2066 $scale = $dbd+$scale;
2072 # round left from the '.'
2074 # 123 => 100 means length(123) = 3 - $scale (2) => 1
2076 my $dbt = $MBI->_len($x->{_m});
2078 my $dbd = $dbt + ($x->{_es} . $MBI->_num($x->{_e}));
2079 # should be the same, so treat it as this
2080 $scale = 1 if $scale == 0;
2081 # shortcut if already integer
2082 return $x if $scale == 1 && $dbt <= $dbd;
2083 # maximum digits before dot
2088 # not enough digits before dot, so round to zero
2091 elsif ( $scale == $dbd )
2098 $scale = $dbd - $scale;
2101 # pass sign to bround for rounding modes '+inf' and '-inf'
2102 my $m = bless { sign => $x->{sign}, value => $x->{_m} }, 'Math::BigInt';
2103 $m->bround($scale,$mode);
2104 $x->{_m} = $m->{value}; # get our mantissa back
2110 # accuracy: preserve $N digits, and overwrite the rest with 0's
2111 my $x = shift; my $self = ref($x) || $x; $x = $self->new(shift) if !ref($x);
2113 if (($_[0] || 0) < 0)
2115 require Carp; Carp::croak ('bround() needs positive accuracy');
2118 my ($scale,$mode) = $x->_scale_a(@_);
2119 return $x if !defined $scale || $x->modify('bround'); # no-op
2121 # scale is now either $x->{_a}, $accuracy, or the user parameter
2122 # test whether $x already has lower accuracy, do nothing in this case
2123 # but do round if the accuracy is the same, since a math operation might
2124 # want to round a number with A=5 to 5 digits afterwards again
2125 return $x if defined $x->{_a} && $x->{_a} < $scale;
2127 # scale < 0 makes no sense
2128 # scale == 0 => keep all digits
2129 # never round a +-inf, NaN
2130 return $x if ($scale <= 0) || $x->{sign} !~ /^[+-]$/;
2132 # 1: never round a 0
2133 # 2: if we should keep more digits than the mantissa has, do nothing
2134 if ($x->is_zero() || $MBI->_len($x->{_m}) <= $scale)
2136 $x->{_a} = $scale if !defined $x->{_a} || $x->{_a} > $scale;
2140 # pass sign to bround for '+inf' and '-inf' rounding modes
2141 my $m = bless { sign => $x->{sign}, value => $x->{_m} }, 'Math::BigInt';
2143 $m->bround($scale,$mode); # round mantissa
2144 $x->{_m} = $m->{value}; # get our mantissa back
2145 $x->{_a} = $scale; # remember rounding
2146 delete $x->{_p}; # and clear P
2147 $x->bnorm(); # del trailing zeros gen. by bround()
2152 # return integer less or equal then $x
2153 my ($self,$x,$a,$p,$r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
2155 return $x if $x->modify('bfloor');
2157 return $x if $x->{sign} !~ /^[+-]$/; # nan, +inf, -inf
2159 # if $x has digits after dot
2160 if ($x->{_es} eq '-')
2162 $x->{_m} = $MBI->_rsft($x->{_m},$x->{_e},10); # cut off digits after dot
2163 $x->{_e} = $MBI->_zero(); # trunc/norm
2164 $x->{_es} = '+'; # abs e
2165 $MBI->_inc($x->{_m}) if $x->{sign} eq '-'; # increment if negative
2167 $x->round($a,$p,$r);
2172 # return integer greater or equal then $x
2173 my ($self,$x,$a,$p,$r) = ref($_[0]) ? (ref($_[0]),@_) : objectify(1,@_);
2175 return $x if $x->modify('bceil');
2176 return $x if $x->{sign} !~ /^[+-]$/; # nan, +inf, -inf
2178 # if $x has digits after dot
2179 if ($x->{_es} eq '-')
2181 $x->{_m} = $MBI->_rsft($x->{_m},$x->{_e},10); # cut off digits after dot
2182 $x->{_e} = $MBI->_zero(); # trunc/norm
2183 $x->{_es} = '+'; # abs e
2184 $MBI->_inc($x->{_m}) if $x->{sign} eq '+'; # increment if positive
2186 $x->round($a,$p,$r);
2191 # shift right by $y (divide by power of $n)
2194 my ($self,$x,$y,$n,$a,$p,$r) = (ref($_[0]),@_);
2195 # objectify is costly, so avoid it
2196 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
2198 ($self,$x,$y,$n,$a,$p,$r) = objectify(2,@_);
2201 return $x if $x->modify('brsft');
2202 return $x if $x->{sign} !~ /^[+-]$/; # nan, +inf, -inf
2204 $n = 2 if !defined $n; $n = $self->new($n);
2205 $x->bdiv($n->bpow($y),$a,$p,$r,$y);
2210 # shift left by $y (multiply by power of $n)
2213 my ($self,$x,$y,$n,$a,$p,$r) = (ref($_[0]),@_);
2214 # objectify is costly, so avoid it
2215 if ((!ref($_[0])) || (ref($_[0]) ne ref($_[1])))
2217 ($self,$x,$y,$n,$a,$p,$r) = objectify(2,@_);
2220 return $x if $x->modify('blsft');
2221 return $x if $x->{sign} !~ /^[+-]$/; # nan, +inf, -inf
2223 $n = 2 if !defined $n; $n = $self->new($n);
2224 $x->bmul($n->bpow($y),$a,$p,$r,$y);
2227 ###############################################################################
2231 # going through AUTOLOAD for every DESTROY is costly, avoid it by empty sub
2236 # make fxxx and bxxx both work by selectively mapping fxxx() to MBF::bxxx()
2237 # or falling back to MBI::bxxx()
2238 my $name = $AUTOLOAD;
2240 $name =~ s/(.*):://; # split package
2241 my $c = $1 || $class;
2243 $c->import() if $IMPORT == 0;
2244 if (!method_alias($name))
2248 # delayed load of Carp and avoid recursion
2250 Carp::croak ("$c: Can't call a method without name");
2252 if (!method_hand_up($name))
2254 # delayed load of Carp and avoid recursion
2256 Carp::croak ("Can't call $c\-\>$name, not a valid method");
2258 # try one level up, but subst. bxxx() for fxxx() since MBI only got bxxx()
2260 return &{"Math::BigInt"."::$name"}(@_);
2262 my $bname = $name; $bname =~ s/^f/b/;
2270 # return a copy of the exponent
2271 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2273 if ($x->{sign} !~ /^[+-]$/)
2275 my $s = $x->{sign}; $s =~ s/^[+-]//;
2276 return Math::BigInt->new($s); # -inf, +inf => +inf
2278 Math::BigInt->new( $x->{_es} . $MBI->_str($x->{_e}));
2283 # return a copy of the mantissa
2284 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2286 if ($x->{sign} !~ /^[+-]$/)
2288 my $s = $x->{sign}; $s =~ s/^[+]//;
2289 return Math::BigInt->new($s); # -inf, +inf => +inf
2291 my $m = Math::BigInt->new( $MBI->_str($x->{_m}));
2292 $m->bneg() if $x->{sign} eq '-';
2299 # return a copy of both the exponent and the mantissa
2300 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2302 if ($x->{sign} !~ /^[+-]$/)
2304 my $s = $x->{sign}; $s =~ s/^[+]//; my $se = $s; $se =~ s/^[-]//;
2305 return ($self->new($s),$self->new($se)); # +inf => inf and -inf,+inf => inf
2307 my $m = Math::BigInt->bzero();
2308 $m->{value} = $MBI->_copy($x->{_m});
2309 $m->bneg() if $x->{sign} eq '-';
2310 ($m, Math::BigInt->new( $x->{_es} . $MBI->_num($x->{_e}) ));
2313 ##############################################################################
2314 # private stuff (internal use only)
2320 my $lib = ''; my @a;
2322 for ( my $i = 0; $i < $l ; $i++)
2324 if ( $_[$i] eq ':constant' )
2326 # This causes overlord er load to step in. 'binary' and 'integer'
2327 # are handled by BigInt.
2328 overload::constant float => sub { $self->new(shift); };
2330 elsif ($_[$i] eq 'upgrade')
2332 # this causes upgrading
2333 $upgrade = $_[$i+1]; # or undef to disable
2336 elsif ($_[$i] eq 'downgrade')
2338 # this causes downgrading
2339 $downgrade = $_[$i+1]; # or undef to disable
2342 elsif ($_[$i] eq 'lib')
2344 # alternative library
2345 $lib = $_[$i+1] || ''; # default Calc
2348 elsif ($_[$i] eq 'with')
2350 # alternative class for our private parts()
2351 # XXX: no longer supported
2352 # $MBI = $_[$i+1] || 'Math::BigInt';
2361 $lib =~ tr/a-zA-Z0-9,://cd; # restrict to sane characters
2362 # let use Math::BigInt lib => 'GMP'; use Math::BigFloat; still work
2363 my $mbilib = eval { Math::BigInt->config()->{lib} };
2364 if ((defined $mbilib) && ($MBI eq 'Math::BigInt::Calc'))
2366 # MBI already loaded
2367 Math::BigInt->import('lib',"$lib,$mbilib", 'objectify');
2371 # MBI not loaded, or with ne "Math::BigInt::Calc"
2372 $lib .= ",$mbilib" if defined $mbilib;
2373 $lib =~ s/^,//; # don't leave empty
2375 # replacement library can handle lib statement, but also could ignore it
2377 # Perl < 5.6.0 dies with "out of memory!" when eval() and ':constant' is
2378 # used in the same script, or eval inside import(). So we require MBI:
2379 require Math::BigInt;
2380 Math::BigInt->import( lib => $lib, 'objectify' );
2384 require Carp; Carp::croak ("Couldn't load $lib: $! $@");
2386 # find out which one was actually loaded
2387 $MBI = Math::BigInt->config()->{lib};
2389 # register us with MBI to get notified of future lib changes
2390 Math::BigInt::_register_callback( $self, sub { $MBI = $_[0]; } );
2392 # any non :constant stuff is handled by our parent, Exporter
2393 # even if @_ is empty, to give it a chance
2394 $self->SUPER::import(@a); # for subclasses
2395 $self->export_to_level(1,$self,@a); # need this, too
2400 # adjust m and e so that m is smallest possible
2401 # round number according to accuracy and precision settings
2402 my ($self,$x) = ref($_[0]) ? (undef,$_[0]) : objectify(1,@_);
2404 return $x if $x->{sign} !~ /^[+-]$/; # inf, nan etc
2406 my $zeros = $MBI->_zeros($x->{_m}); # correct for trailing zeros
2409 my $z = $MBI->_new($zeros);
2410 $x->{_m} = $MBI->_rsft ($x->{_m}, $z, 10);
2411 if ($x->{_es} eq '-')
2413 if ($MBI->_acmp($x->{_e},$z) >= 0)
2415 $x->{_e} = $MBI->_sub ($x->{_e}, $z);
2416 $x->{_es} = '+' if $MBI->_is_zero($x->{_e});
2420 $x->{_e} = $MBI->_sub ( $MBI->_copy($z), $x->{_e});
2426 $x->{_e} = $MBI->_add ($x->{_e}, $z);
2431 # $x can only be 0Ey if there are no trailing zeros ('0' has 0 trailing
2432 # zeros). So, for something like 0Ey, set y to 1, and -0 => +0
2433 $x->{sign} = '+', $x->{_es} = '+', $x->{_e} = $MBI->_one()
2434 if $MBI->_is_zero($x->{_m});
2437 $x; # MBI bnorm is no-op, so dont call it
2440 ##############################################################################
2444 # return number as hexadecimal string (only for integers defined)
2445 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2447 return $x->bstr() if $x->{sign} !~ /^[+-]$/; # inf, nan etc
2448 return '0x0' if $x->is_zero();
2450 return $nan if $x->{_es} ne '+'; # how to do 1e-1 in hex!?
2452 my $z = $MBI->_copy($x->{_m});
2453 if (! $MBI->_is_zero($x->{_e})) # > 0
2455 $MBI->_lsft( $z, $x->{_e},10);
2457 $z = Math::BigInt->new( $x->{sign} . $MBI->_num($z));
2463 # return number as binary digit string (only for integers defined)
2464 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2466 return $x->bstr() if $x->{sign} !~ /^[+-]$/; # inf, nan etc
2467 return '0b0' if $x->is_zero();
2469 return $nan if $x->{_es} ne '+'; # how to do 1e-1 in hex!?
2471 my $z = $MBI->_copy($x->{_m});
2472 if (! $MBI->_is_zero($x->{_e})) # > 0
2474 $MBI->_lsft( $z, $x->{_e},10);
2476 $z = Math::BigInt->new( $x->{sign} . $MBI->_num($z));
2482 # return copy as a bigint representation of this BigFloat number
2483 my ($self,$x) = ref($_[0]) ? (ref($_[0]),$_[0]) : objectify(1,@_);
2485 my $z = $MBI->_copy($x->{_m});
2486 if ($x->{_es} eq '-') # < 0
2488 $MBI->_rsft( $z, $x->{_e},10);
2490 elsif (! $MBI->_is_zero($x->{_e})) # > 0
2492 $MBI->_lsft( $z, $x->{_e},10);
2494 $z = Math::BigInt->new( $x->{sign} . $MBI->_num($z));
2501 my $class = ref($x) || $x;
2502 $x = $class->new(shift) unless ref($x);
2504 return 1 if $MBI->_is_zero($x->{_m});
2506 my $len = $MBI->_len($x->{_m});
2507 $len += $MBI->_num($x->{_e}) if $x->{_es} eq '+';
2511 $t = $MBI->_num($x->{_e}) if $x->{_es} eq '-';
2522 Math::BigFloat - Arbitrary size floating point math package
2529 $x = Math::BigFloat->new($str); # defaults to 0
2530 $nan = Math::BigFloat->bnan(); # create a NotANumber
2531 $zero = Math::BigFloat->bzero(); # create a +0
2532 $inf = Math::BigFloat->binf(); # create a +inf
2533 $inf = Math::BigFloat->binf('-'); # create a -inf
2534 $one = Math::BigFloat->bone(); # create a +1
2535 $one = Math::BigFloat->bone('-'); # create a -1
2538 $x->is_zero(); # true if arg is +0
2539 $x->is_nan(); # true if arg is NaN
2540 $x->is_one(); # true if arg is +1
2541 $x->is_one('-'); # true if arg is -1
2542 $x->is_odd(); # true if odd, false for even
2543 $x->is_even(); # true if even, false for odd
2544 $x->is_pos(); # true if >= 0
2545 $x->is_neg(); # true if < 0
2546 $x->is_inf(sign); # true if +inf, or -inf (default is '+')
2548 $x->bcmp($y); # compare numbers (undef,<0,=0,>0)
2549 $x->bacmp($y); # compare absolutely (undef,<0,=0,>0)
2550 $x->sign(); # return the sign, either +,- or NaN
2551 $x->digit($n); # return the nth digit, counting from right
2552 $x->digit(-$n); # return the nth digit, counting from left
2554 # The following all modify their first argument. If you want to preserve
2555 # $x, use $z = $x->copy()->bXXX($y); See under L<CAVEATS> for why this is
2556 # neccessary when mixing $a = $b assigments with non-overloaded math.
2559 $x->bzero(); # set $i to 0
2560 $x->bnan(); # set $i to NaN
2561 $x->bone(); # set $x to +1
2562 $x->bone('-'); # set $x to -1
2563 $x->binf(); # set $x to inf
2564 $x->binf('-'); # set $x to -inf
2566 $x->bneg(); # negation
2567 $x->babs(); # absolute value
2568 $x->bnorm(); # normalize (no-op)
2569 $x->bnot(); # two's complement (bit wise not)
2570 $x->binc(); # increment x by 1
2571 $x->bdec(); # decrement x by 1
2573 $x->badd($y); # addition (add $y to $x)
2574 $x->bsub($y); # subtraction (subtract $y from $x)
2575 $x->bmul($y); # multiplication (multiply $x by $y)
2576 $x->bdiv($y); # divide, set $x to quotient
2577 # return (quo,rem) or quo if scalar
2579 $x->bmod($y); # modulus ($x % $y)
2580 $x->bpow($y); # power of arguments ($x ** $y)
2581 $x->blsft($y); # left shift
2582 $x->brsft($y); # right shift
2583 # return (quo,rem) or quo if scalar
2585 $x->blog(); # logarithm of $x to base e (Euler's number)
2586 $x->blog($base); # logarithm of $x to base $base (f.i. 2)
2588 $x->band($y); # bit-wise and
2589 $x->bior($y); # bit-wise inclusive or
2590 $x->bxor($y); # bit-wise exclusive or
2591 $x->bnot(); # bit-wise not (two's complement)
2593 $x->bsqrt(); # calculate square-root
2594 $x->broot($y); # $y'th root of $x (e.g. $y == 3 => cubic root)
2595 $x->bfac(); # factorial of $x (1*2*3*4*..$x)
2597 $x->bround($N); # accuracy: preserve $N digits
2598 $x->bfround($N); # precision: round to the $Nth digit
2600 $x->bfloor(); # return integer less or equal than $x
2601 $x->bceil(); # return integer greater or equal than $x
2603 # The following do not modify their arguments:
2605 bgcd(@values); # greatest common divisor
2606 blcm(@values); # lowest common multiplicator
2608 $x->bstr(); # return string
2609 $x->bsstr(); # return string in scientific notation
2611 $x->as_int(); # return $x as BigInt
2612 $x->exponent(); # return exponent as BigInt
2613 $x->mantissa(); # return mantissa as BigInt
2614 $x->parts(); # return (mantissa,exponent) as BigInt
2616 $x->length(); # number of digits (w/o sign and '.')
2617 ($l,$f) = $x->length(); # number of digits, and length of fraction
2619 $x->precision(); # return P of $x (or global, if P of $x undef)
2620 $x->precision($n); # set P of $x to $n
2621 $x->accuracy(); # return A of $x (or global, if A of $x undef)
2622 $x->accuracy($n); # set A $x to $n
2624 # these get/set the appropriate global value for all BigFloat objects
2625 Math::BigFloat->precision(); # Precision
2626 Math::BigFloat->accuracy(); # Accuracy
2627 Math::BigFloat->round_mode(); # rounding mode
2631 All operators (inlcuding basic math operations) are overloaded if you
2632 declare your big floating point numbers as
2634 $i = new Math::BigFloat '12_3.456_789_123_456_789E-2';
2636 Operations with overloaded operators preserve the arguments, which is
2637 exactly what you expect.
2639 =head2 Canonical notation
2641 Input to these routines are either BigFloat objects, or strings of the
2642 following four forms:
2656 C</^[+-]\d+E[+-]?\d+$/>
2660 C</^[+-]\d*\.\d+E[+-]?\d+$/>
2664 all with optional leading and trailing zeros and/or spaces. Additonally,
2665 numbers are allowed to have an underscore between any two digits.
2667 Empty strings as well as other illegal numbers results in 'NaN'.
2669 bnorm() on a BigFloat object is now effectively a no-op, since the numbers
2670 are always stored in normalized form. On a string, it creates a BigFloat
2675 Output values are BigFloat objects (normalized), except for bstr() and bsstr().
2677 The string output will always have leading and trailing zeros stripped and drop
2678 a plus sign. C<bstr()> will give you always the form with a decimal point,
2679 while C<bsstr()> (s for scientific) gives you the scientific notation.
2681 Input bstr() bsstr()
2683 ' -123 123 123' '-123123123' '-123123123E0'
2684 '00.0123' '0.0123' '123E-4'
2685 '123.45E-2' '1.2345' '12345E-4'
2686 '10E+3' '10000' '1E4'
2688 Some routines (C<is_odd()>, C<is_even()>, C<is_zero()>, C<is_one()>,
2689 C<is_nan()>) return true or false, while others (C<bcmp()>, C<bacmp()>)
2690 return either undef, <0, 0 or >0 and are suited for sort.
2692 Actual math is done by using the class defined with C<with => Class;> (which
2693 defaults to BigInts) to represent the mantissa and exponent.
2695 The sign C</^[+-]$/> is stored separately. The string 'NaN' is used to
2696 represent the result when input arguments are not numbers, as well as
2697 the result of dividing by zero.
2699 =head2 C<mantissa()>, C<exponent()> and C<parts()>
2701 C<mantissa()> and C<exponent()> return the said parts of the BigFloat
2702 as BigInts such that:
2704 $m = $x->mantissa();
2705 $e = $x->exponent();
2706 $y = $m * ( 10 ** $e );
2707 print "ok\n" if $x == $y;
2709 C<< ($m,$e) = $x->parts(); >> is just a shortcut giving you both of them.
2711 A zero is represented and returned as C<0E1>, B<not> C<0E0> (after Knuth).
2713 Currently the mantissa is reduced as much as possible, favouring higher
2714 exponents over lower ones (e.g. returning 1e7 instead of 10e6 or 10000000e0).
2715 This might change in the future, so do not depend on it.
2717 =head2 Accuracy vs. Precision
2719 See also: L<Rounding|Rounding>.
2721 Math::BigFloat supports both precision (rounding to a certain place before or
2722 after the dot) and accuracy (rounding to a certain number of digits). For a
2723 full documentation, examples and tips on these topics please see the large
2724 section about rounding in L<Math::BigInt>.
2726 Since things like C<sqrt(2)> or C<1 / 3> must presented with a limited
2727 accuracy lest a operation consumes all resources, each operation produces
2728 no more than the requested number of digits.
2730 If there is no gloabl precision or accuracy set, B<and> the operation in
2731 question was not called with a requested precision or accuracy, B<and> the
2732 input $x has no accuracy or precision set, then a fallback parameter will
2733 be used. For historical reasons, it is called C<div_scale> and can be accessed
2736 $d = Math::BigFloat->div_scale(); # query
2737 Math::BigFloat->div_scale($n); # set to $n digits
2739 The default value for C<div_scale> is 40.
2741 In case the result of one operation has more digits than specified,
2742 it is rounded. The rounding mode taken is either the default mode, or the one
2743 supplied to the operation after the I<scale>:
2745 $x = Math::BigFloat->new(2);
2746 Math::BigFloat->accuracy(5); # 5 digits max
2747 $y = $x->copy()->bdiv(3); # will give 0.66667
2748 $y = $x->copy()->bdiv(3,6); # will give 0.666667
2749 $y = $x->copy()->bdiv(3,6,undef,'odd'); # will give 0.666667
2750 Math::BigFloat->round_mode('zero');
2751 $y = $x->copy()->bdiv(3,6); # will also give 0.666667
2753 Note that C<< Math::BigFloat->accuracy() >> and C<< Math::BigFloat->precision() >>
2754 set the global variables, and thus B<any> newly created number will be subject
2755 to the global rounding B<immidiately>. This means that in the examples above, the
2756 C<3> as argument to C<bdiv()> will also get an accuracy of B<5>.
2758 It is less confusing to either calculate the result fully, and afterwards
2759 round it explicitely, or use the additional parameters to the math
2763 $x = Math::BigFloat->new(2);
2764 $y = $x->copy()->bdiv(3);
2765 print $y->bround(5),"\n"; # will give 0.66667
2770 $x = Math::BigFloat->new(2);
2771 $y = $x->copy()->bdiv(3,5); # will give 0.66667
2778 =item ffround ( +$scale )
2780 Rounds to the $scale'th place left from the '.', counting from the dot.
2781 The first digit is numbered 1.
2783 =item ffround ( -$scale )
2785 Rounds to the $scale'th place right from the '.', counting from the dot.
2789 Rounds to an integer.
2791 =item fround ( +$scale )
2793 Preserves accuracy to $scale digits from the left (aka significant digits)
2794 and pads the rest with zeros. If the number is between 1 and -1, the
2795 significant digits count from the first non-zero after the '.'
2797 =item fround ( -$scale ) and fround ( 0 )
2799 These are effectively no-ops.
2803 All rounding functions take as a second parameter a rounding mode from one of
2804 the following: 'even', 'odd', '+inf', '-inf', 'zero' or 'trunc'.
2806 The default rounding mode is 'even'. By using
2807 C<< Math::BigFloat->round_mode($round_mode); >> you can get and set the default
2808 mode for subsequent rounding. The usage of C<$Math::BigFloat::$round_mode> is
2809 no longer supported.
2810 The second parameter to the round functions then overrides the default
2813 The C<as_number()> function returns a BigInt from a Math::BigFloat. It uses
2814 'trunc' as rounding mode to make it equivalent to:
2819 You can override this by passing the desired rounding mode as parameter to
2822 $x = Math::BigFloat->new(2.5);
2823 $y = $x->as_number('odd'); # $y = 3
2829 $x->accuracy(5); # local for $x
2830 CLASS->accuracy(5); # global for all members of CLASS
2831 # Note: This also applies to new()!
2833 $A = $x->accuracy(); # read out accuracy that affects $x
2834 $A = CLASS->accuracy(); # read out global accuracy
2836 Set or get the global or local accuracy, aka how many significant digits the
2837 results have. If you set a global accuracy, then this also applies to new()!
2839 Warning! The accuracy I<sticks>, e.g. once you created a number under the
2840 influence of C<< CLASS->accuracy($A) >>, all results from math operations with
2841 that number will also be rounded.
2843 In most cases, you should probably round the results explicitely using one of
2844 L<round()>, L<bround()> or L<bfround()> or by passing the desired accuracy
2845 to the math operation as additional parameter:
2847 my $x = Math::BigInt->new(30000);
2848 my $y = Math::BigInt->new(7);
2849 print scalar $x->copy()->bdiv($y, 2); # print 4300
2850 print scalar $x->copy()->bdiv($y)->bround(2); # print 4300
2854 $x->precision(-2); # local for $x, round at the second digit right of the dot
2855 $x->precision(2); # ditto, round at the second digit left of the dot
2857 CLASS->precision(5); # Global for all members of CLASS
2858 # This also applies to new()!
2859 CLASS->precision(-5); # ditto
2861 $P = CLASS->precision(); # read out global precision
2862 $P = $x->precision(); # read out precision that affects $x
2864 Note: You probably want to use L<accuracy()> instead. With L<accuracy> you
2865 set the number of digits each result should have, with L<precision> you
2866 set the place where to round!
2868 =head1 Autocreating constants
2870 After C<use Math::BigFloat ':constant'> all the floating point constants
2871 in the given scope are converted to C<Math::BigFloat>. This conversion
2872 happens at compile time.
2876 perl -MMath::BigFloat=:constant -e 'print 2E-100,"\n"'
2878 prints the value of C<2E-100>. Note that without conversion of
2879 constants the expression 2E-100 will be calculated as normal floating point
2882 Please note that ':constant' does not affect integer constants, nor binary
2883 nor hexadecimal constants. Use L<bignum> or L<Math::BigInt> to get this to
2888 Math with the numbers is done (by default) by a module called
2889 Math::BigInt::Calc. This is equivalent to saying:
2891 use Math::BigFloat lib => 'Calc';
2893 You can change this by using:
2895 use Math::BigFloat lib => 'BitVect';
2897 The following would first try to find Math::BigInt::Foo, then
2898 Math::BigInt::Bar, and when this also fails, revert to Math::BigInt::Calc:
2900 use Math::BigFloat lib => 'Foo,Math::BigInt::Bar';
2902 Calc.pm uses as internal format an array of elements of some decimal base
2903 (usually 1e7, but this might be differen for some systems) with the least
2904 significant digit first, while BitVect.pm uses a bit vector of base 2, most
2905 significant bit first. Other modules might use even different means of
2906 representing the numbers. See the respective module documentation for further
2909 Please note that Math::BigFloat does B<not> use the denoted library itself,
2910 but it merely passes the lib argument to Math::BigInt. So, instead of the need
2913 use Math::BigInt lib => 'GMP';
2916 you can roll it all into one line:
2918 use Math::BigFloat lib => 'GMP';
2920 It is also possible to just require Math::BigFloat:
2922 require Math::BigFloat;
2924 This will load the neccessary things (like BigInt) when they are needed, and
2927 Use the lib, Luke! And see L<Using Math::BigInt::Lite> for more details than
2928 you ever wanted to know about loading a different library.
2930 =head2 Using Math::BigInt::Lite
2932 It is possible to use L<Math::BigInt::Lite> with Math::BigFloat:
2935 use Math::BigFloat with => 'Math::BigInt::Lite';
2937 There is no need to "use Math::BigInt" or "use Math::BigInt::Lite", but you
2938 can combine these if you want. For instance, you may want to use
2939 Math::BigInt objects in your main script, too.
2943 use Math::BigFloat with => 'Math::BigInt::Lite';
2945 Of course, you can combine this with the C<lib> parameter.
2948 use Math::BigFloat with => 'Math::BigInt::Lite', lib => 'GMP,Pari';
2950 There is no need for a "use Math::BigInt;" statement, even if you want to
2951 use Math::BigInt's, since Math::BigFloat will needs Math::BigInt and thus
2952 always loads it. But if you add it, add it B<before>:
2956 use Math::BigFloat with => 'Math::BigInt::Lite', lib => 'GMP,Pari';
2958 Notice that the module with the last C<lib> will "win" and thus
2959 it's lib will be used if the lib is available:
2962 use Math::BigInt lib => 'Bar,Baz';
2963 use Math::BigFloat with => 'Math::BigInt::Lite', lib => 'Foo';
2965 That would try to load Foo, Bar, Baz and Calc (in that order). Or in other
2966 words, Math::BigFloat will try to retain previously loaded libs when you
2967 don't specify it onem but if you specify one, it will try to load them.
2969 Actually, the lib loading order would be "Bar,Baz,Calc", and then
2970 "Foo,Bar,Baz,Calc", but independend of which lib exists, the result is the
2971 same as trying the latter load alone, except for the fact that one of Bar or
2972 Baz might be loaded needlessly in an intermidiate step (and thus hang around
2973 and waste memory). If neither Bar nor Baz exist (or don't work/compile), they
2974 will still be tried to be loaded, but this is not as time/memory consuming as
2975 actually loading one of them. Still, this type of usage is not recommended due
2978 The old way (loading the lib only in BigInt) still works though:
2981 use Math::BigInt lib => 'Bar,Baz';
2984 You can even load Math::BigInt afterwards:
2988 use Math::BigInt lib => 'Bar,Baz';
2990 But this has the same problems like #5, it will first load Calc
2991 (Math::BigFloat needs Math::BigInt and thus loads it) and then later Bar or
2992 Baz, depending on which of them works and is usable/loadable. Since this
2993 loads Calc unnecc., it is not recommended.
2995 Since it also possible to just require Math::BigFloat, this poses the question
2996 about what libary this will use:
2998 require Math::BigFloat;
2999 my $x = Math::BigFloat->new(123); $x += 123;
3001 It will use Calc. Please note that the call to import() is still done, but
3002 only when you use for the first time some Math::BigFloat math (it is triggered
3003 via any constructor, so the first time you create a Math::BigFloat, the load
3004 will happen in the background). This means:
3006 require Math::BigFloat;
3007 Math::BigFloat->import ( lib => 'Foo,Bar' );
3009 would be the same as:
3011 use Math::BigFloat lib => 'Foo, Bar';
3013 But don't try to be clever to insert some operations in between:
3015 require Math::BigFloat;
3016 my $x = Math::BigFloat->bone() + 4; # load BigInt and Calc
3017 Math::BigFloat->import( lib => 'Pari' ); # load Pari, too
3018 $x = Math::BigFloat->bone()+4; # now use Pari
3020 While this works, it loads Calc needlessly. But maybe you just wanted that?
3022 B<Examples #3 is highly recommended> for daily usage.
3026 Please see the file BUGS in the CPAN distribution Math::BigInt for known bugs.
3032 =item stringify, bstr()
3034 Both stringify and bstr() now drop the leading '+'. The old code would return
3035 '+1.23', the new returns '1.23'. See the documentation in L<Math::BigInt> for
3036 reasoning and details.
3040 The following will probably not do what you expect:
3042 print $c->bdiv(123.456),"\n";
3044 It prints both quotient and reminder since print works in list context. Also,
3045 bdiv() will modify $c, so be carefull. You probably want to use
3047 print $c / 123.456,"\n";
3048 print scalar $c->bdiv(123.456),"\n"; # or if you want to modify $c
3052 =item Modifying and =
3056 $x = Math::BigFloat->new(5);
3059 It will not do what you think, e.g. making a copy of $x. Instead it just makes
3060 a second reference to the B<same> object and stores it in $y. Thus anything
3061 that modifies $x will modify $y (except overloaded math operators), and vice
3062 versa. See L<Math::BigInt> for details and how to avoid that.
3066 C<bpow()> now modifies the first argument, unlike the old code which left
3067 it alone and only returned the result. This is to be consistent with
3068 C<badd()> etc. The first will modify $x, the second one won't:
3070 print bpow($x,$i),"\n"; # modify $x
3071 print $x->bpow($i),"\n"; # ditto
3072 print $x ** $i,"\n"; # leave $x alone
3074 =item precision() vs. accuracy()
3076 A common pitfall is to use L<precision()> when you want to round a result to
3077 a certain number of digits:
3081 Math::BigFloat->precision(4); # does not do what you think it does
3082 my $x = Math::BigFloat->new(12345); # rounds $x to "12000"!
3083 print "$x\n"; # print "12000"
3084 my $y = Math::BigFloat->new(3); # rounds $y to "0"!
3085 print "$y\n"; # print "0"
3086 $z = $x / $y; # 12000 / 0 => NaN!
3088 print $z->precision(),"\n"; # 4
3090 Replacing L<precision> with L<accuracy> is probably not what you want, either:
3094 Math::BigFloat->accuracy(4); # enables global rounding:
3095 my $x = Math::BigFloat->new(123456); # rounded immidiately to "12350"
3096 print "$x\n"; # print "123500"
3097 my $y = Math::BigFloat->new(3); # rounded to "3
3098 print "$y\n"; # print "3"
3099 print $z = $x->copy()->bdiv($y),"\n"; # 41170
3100 print $z->accuracy(),"\n"; # 4
3102 What you want to use instead is:
3106 my $x = Math::BigFloat->new(123456); # no rounding
3107 print "$x\n"; # print "123456"
3108 my $y = Math::BigFloat->new(3); # no rounding
3109 print "$y\n"; # print "3"
3110 print $z = $x->copy()->bdiv($y,4),"\n"; # 41150
3111 print $z->accuracy(),"\n"; # undef
3113 In addition to computing what you expected, the last example also does B<not>
3114 "taint" the result with an accuracy or precision setting, which would
3115 influence any further operation.
3121 L<Math::BigInt>, L<Math::BigRat> and L<Math::Big> as well as
3122 L<Math::BigInt::BitVect>, L<Math::BigInt::Pari> and L<Math::BigInt::GMP>.
3124 The pragmas L<bignum>, L<bigint> and L<bigrat> might also be of interest
3125 because they solve the autoupgrading/downgrading issue, at least partly.
3128 L<http://search.cpan.org/search?mode=module&query=Math%3A%3ABigInt> contains
3129 more documentation including a full version history, testcases, empty
3130 subclass files and benchmarks.
3134 This program is free software; you may redistribute it and/or modify it under
3135 the same terms as Perl itself.
3139 Mark Biggar, overloaded interface by Ilya Zakharevich.
3140 Completely rewritten by Tels L<http://bloodgate.com> in 2001 - 2004, and still