The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
# Copyright 2010, 2011, 2012, 2013, 2014 Kevin Ryde

# This file is part of Math-NumSeq.
#
# Math-NumSeq is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 3, or (at your option) any later
# version.
#
# Math-NumSeq is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
# or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
# for more details.
#
# You should have received a copy of the GNU General Public License along
# with Math-NumSeq.  If not, see <http://www.gnu.org/licenses/>.

package Math::NumSeq::LucasNumbers;
use 5.004;
use strict;

use vars '$VERSION','@ISA';
$VERSION = 72;
use Math::NumSeq::Base::Sparse;
@ISA = ('Math::NumSeq::Base::Sparse');

use Math::NumSeq;
*_is_infinite = \&Math::NumSeq::_is_infinite;
*_to_bigint = \&Math::NumSeq::_to_bigint;

use Math::NumSeq::Fibonacci;
*_blog2_estimate = \&Math::NumSeq::Fibonacci::_blog2_estimate;
*_bit_split_hightolow = \&Math::NumSeq::Fibonacci::_bit_split_hightolow;


# uncomment this to run the ### lines
# use Smart::Comments;


# use constant name => Math::NumSeq::__('Lucas Numbers');
sub description {
  my ($self) = @_;
  if (ref $self && $self->i_start == 0) {
    return Math::NumSeq::__('Lucas numbers 2, 1, 3, 4, 7, 11, 18, 29, etc, being L(i+1) = L(i) + L(i-1) starting from 2,1.  This is the same recurrence as the Fibonacci numbers, but a different starting point.');
  } else {
    return Math::NumSeq::__('Lucas numbers 1, 3, 4, 7, 11, 18, 29, etc, being L(i+1) = L(i) + L(i-1) starting from 1,3.  This is the same recurrence as the Fibonacci numbers, but a different starting point.');
  }
}

# negatives at i odd negative, otherwise minimum 1 at i=1
sub values_min {
  my ($self) = @_;
  my $i = $self->i_start;
  if ($i <= 0 && $i % 2 == 0) {
    # i even, increase to make i odd and i<=1
    $i += 1;
  }
  return $self->ith($i);
}

sub characteristic_increasing {
  my ($self) = @_;
  # not increasing at i=0 value=2 then i=1 value=1
  return ($self->i_start >= 1);
}
sub characteristic_increasing_from_i {
  my ($self) = @_;
  return _max($self->i_start,1);
}
use constant characteristic_integer => 1;
use constant default_i_start => 1;

sub _max {
  my $ret = shift;
  while (@_) {
    my $next = shift;
    if ($next > $ret) {
      $ret = $next;
    }
  }
  return $ret;
}

#------------------------------------------------------------------------------
# cf A000285 starting 1,4
#    A022086 starting 0,3
#    A022087 starting 0,4
#    A022095 starting 1,5
#    A022130 starting 4,9
#    A080023 closest to log(phi), 2,3,4,7,11
#    A169985 nearestint(phi^n) 1,2,3,4,7,11,18
{
  my %oeis_anum = (
                   # OEIS-Catalogue array begin
                   0 => 'A000032', # i_start=0  # Lucas starting at 2,1,3,...
                   1 => 'A000204', #            # Lucas starting at   1,3,...
                   # OEIS-Catalogue array end
                  );
  sub oeis_anum {
    my ($self) = @_;
    return $oeis_anum{$self->i_start};
  }
}

#------------------------------------------------------------------------------

# the biggest f0 for which both f0 and f1 fit into a UV, and which therefore
# for the next step will require BigInt
#
my $uv_limit;
my $uv_i_limit;
{
  # Float integers too in 32 bits ?
  # my $max = 1;
  # for (1 .. 256) {
  #   my $try = $max*2 + 1;
  #   ### $try
  #   if ($try == 2*$max || $try == 2*$max+2) {
  #     last;
  #   }
  #   $max = $try;
  # }
  my $max = ~0;

  # f1+f0 > i
  # f0 > i-f1
  # check i-f1 as the stopping point, so that if i=UV_MAX then won't
  # overflow a UV trying to get to f1>=i
  #
  my $i = 0;
  my $f0 = 1;
  my $f1 = 3;
  my $prev_f0;
  while ($f0 <= $max - $f1) {
    $prev_f0 = $f0;
    ($f1,$f0) = ($f1+$f0,$f1);
    $i++;
  }
  ### $prev_f0
  ### $f0
  ### $f1
  ### ~0 : ~0

  $uv_limit = $prev_f0;
  $uv_i_limit = $i;
  ### $uv_limit
  ### $uv_i_limit
  ### assert: __PACKAGE__->ith($uv_i_limit) == $uv_limit
};

#------------------------------------------------------------------------------

sub rewind {
  my ($self) = @_;
  $self->seek_to_i($self->i_start);
}
sub seek_to_i {
  my ($self, $i) = @_;
  $self->{'i'} = $i;
  if (abs($i) >= $uv_i_limit) {
    $i = Math::NumSeq::_to_bigint($i);
  }
  $self->{'f0'} = $self->ith($i);
  $self->{'f1'} = $self->ith($i+1);

  # or perhaps
  # ($self->{'f0'}, $self->{'f1'}) = $self->ith_pair($i);
}

sub next {
  my ($self) = @_;
  ### LucasNumbers next(): "f0=$self->{'f0'}, f1=$self->{'f1'}"

  (my $ret,
   $self->{'f0'},
   $self->{'f1'})
    = ($self->{'f0'},
       $self->{'f1'},
       $self->{'f0'}+$self->{'f1'});
  ### $ret

  if ($ret == $uv_limit) {
    $self->{'f0'} = Math::NumSeq::_to_bigint($self->{'f0'});
    $self->{'f1'} = Math::NumSeq::_to_bigint($self->{'f1'});
  }

  return ($self->{'i'}++, $ret);
}

# F[4] = (F[2]+L[2])^2/2 - 3*F[2]^2 - 2*(-1)^2
#      = (1+3)^2/4 - 3*1^2 - 2
#      = 16/4 - 3 - 2

# F[3] = ((F[1]+L[1])^2 - 2*(-1)^1)/4 + F[1]^2
#      = ((1+3)^2 - -2)/4 + 1^2
#      = (16 + 2)/4 + 1
#      = (16 + 2)/4 + 1

# F[3] = (F[1]+L[1])^2/4 + F[1]^2
#      = (1+3)^2/4 + 1^2
#      = 16/4 + 1
#      = 5

# -8,5,-3,2,-1,1, 0, 1,1,2,3,5,8,13,21
# -11,7,-4,3,-1, 2, 1,3,4,7,11,18,29
#

sub ith {
  my ($self, $i) = @_;
  ### LucasNumbers ith(): $i

  if (_is_infinite($i)) {
    return $i;
  }
  $i = int($i);
  if ($i == 0) {
    return 2;
  }

  # automatic BigInt if not another bignum class
  my $to_bigint = ($i >= $uv_i_limit || $i <= -$uv_i_limit && ! ref $i);
  ### $to_bigint

  my $lowzeros = 0;
  until ($i % 2) {
    $lowzeros++;
    $i /= 2;
  }

  my ($L0) = $self->ith_pair($i);
  ### $L0
  if ($to_bigint) {
    ### to bigint ...
    $L0 = _to_bigint($L0);
    ### $L0
  }

  ### apply lowzeros: $lowzeros
  my $add = -2;
  while ($lowzeros--) {
    $L0 *= $L0;
    $L0 -= $add;
    $add = 2;
    ### lowzeros L0: "$L0 " . ref $L0
  }

  ### final ith() ...
  ### L0: "$L0"

  return $L0;


  # {
  # # cf plain interative
  # $i--;
  # my $f0 = ($i * 0) + 1;  # inherit bignum 1
  # my $f1 = $f0 + 2;       # inherit bignum 3
  # while (--$i > 0) {
  #   $f0 += $f1;
  #
  #   unless (--$i > 0) {
  #     return $f0;
  #   }
  #   $f1 += $f0;
  # }
  # return $f1;
  # }
}

sub ith_pair {
  my ($self, $i) = @_;
  ### LucasNumbers ith_pair(): $i

  if (_is_infinite($i)) {
    return ($i,$i);
  }
  $i = int($i);

  my $neg;
  if ($i < 0) {
    $i = -1 - $i;    # L[-k] = L[-1-k] * (-1)^k
    $neg = 1;
  }

  my ($Lk, $Lkplus1);
  if ($i == 0) {
    $Lk = 2;       # L[0] = 2
    $Lkplus1 = 1;  # L[1] = 1

  } else {
    # initial k=1
    my $zero = ($i >= $uv_i_limit && ! ref $i
                ? _to_bigint(0)   # automatic BigInt if not another bignum class
                : $i * 0);        # inherit bignum $i

    $Lk = 1 + $zero;       # L[k]   = L[1] = 1
    $Lkplus1 = 3 + $zero;  # L[k+1] = L[2] = 3
    my $add = -2;  # 2*(-1)^k


    my @bits = _bit_split_hightolow($i);
    ### @bits
    shift @bits; # drop high 1-bit

    while (@bits) {
      ### at: "Lk=$Lk Lkplus1=$Lkplus1 bit=$bits[0]"
      $Lk *= $Lk;
      $Lk -= $add;                 # L[2k] = L[k]^2 - 2*(-1)^k

      $Lkplus1 *= $Lkplus1;
      $Lkplus1 += $add;            # L[2k+2] = L[k+1]^2 + 2*(-1)^k

      # L[2k+1] = L[2k+2] - L[2k]
      my $Lmid = $Lkplus1 - $Lk;   # L[2k+1] = L[2k+2] - L[2k]

      if (shift @bits) {  # high to low
        $Lk = $Lmid;               # L[2k+1], L[2k+2]
        $add = -2;
      } else {
        $Lkplus1 = $Lmid;          # L[2k], L[2k+1]
        $add = 2;
      }
    }
  }
  ### loop final: "Lk=$Lk Lkplus1=$Lkplus1"

  if ($neg) {
    ($Lk,$Lkplus1) = ($Lkplus1, $Lk);
    if ($i % 2) {
      $Lkplus1 = -$Lkplus1;
    } else {
      $Lk = -$Lk;
    }
  }
  return ($Lk, $Lkplus1);
}

use constant 1.02 _PHI => (1 + sqrt(5)) / 2;

sub value_to_i_estimate {
  my ($self, $value) = @_;
  if (_is_infinite($value)) {
    return $value;
  }
  if ($value <= 0) {
    return 0;
  }
  if (defined (my $blog2 = _blog2_estimate($value))) {
    # i ~= log2(L(i)) / log2(phi)
    # with log2(x) = log(x)/log(2)
    return int($blog2 * (1 / (log(_PHI)/log(2))));
  }

  # i ~= log(L(i)) / log(phi)
  return int(log($value) * (1/log(_PHI)));
}

1;
__END__

=for stopwords Ryde Math-NumSeq Ith ie doublings bignum Lestimate Festimate-1 vice-versa

=head1 NAME

Math::NumSeq::LucasNumbers -- Lucas numbers

=head1 SYNOPSIS

 use Math::NumSeq::LucasNumbers;
 my $seq = Math::NumSeq::LucasNumbers->new;
 my ($i, $value) = $seq->next;

=head1 DESCRIPTION

The Lucas numbers, L(i) = L(i-1) + L(i-2) starting from values 1,3.

    1, 3, 4, 7, 11, 18, 29, 47, 76, 123, 199, 322, 521, 843, 1364,...
    starting i=1

This is the same recurrence as the Fibonacci numbers
(L<Math::NumSeq::Fibonacci>), but a different starting point.

    L[i+1] = L[i] + L[i-1]

Each Lucas number falls in between successive Fibonaccis, and in fact the
distance is a further Fibonacci,

    F[i+1] < L[i] < F[i+2]

    L[i] = F[i+1] + F[i-1]      # above F[i+1]
    L[i] = F[i+2] - F[i-2]      # below F[i+2]

=head2 Start

Optional C<i_start =E<gt> $i> can start the sequence from somewhere other
than the default i=1.  For example starting at i=0 gives value 2 at i=0,

    i_start => 0
    2, 1, 3, 4, 7, 11, 18, ...

=head1 FUNCTIONS

See L<Math::NumSeq/FUNCTIONS> for behaviour common to all sequence classes.

=over 4

=item C<$seq = Math::NumSeq::LucasNumbers-E<gt>new ()>

=item C<$seq = Math::NumSeq::LucasNumbers-E<gt>new (i_start =E<gt> $i)>

Create and return a new sequence object.

=back

=head2 Iterating

=over

=item C<($i, $value) = $seq-E<gt>next()>

Return the next index and value in the sequence.

When C<$value> exceeds the range of a Perl unsigned integer the return is a
C<Math::BigInt> to preserve precision.

=item C<$seq-E<gt>seek_to_i($i)>

Move the current sequence position to C<$i>.  The next call to C<next()>
will return C<$i> and corresponding value.

=back

=head2 Random Access

=over

=item C<$value = $seq-E<gt>ith($i)>

Return the C<$i>'th Lucas number.

=item C<$bool = $seq-E<gt>pred($value)>

Return true if C<$value> is a Lucas number.

=item C<$i = $seq-E<gt>value_to_i_estimate($value)>

Return an estimate of the i corresponding to C<$value>.  See L</Value to i
Estimate> below.

=back

=head1 FORMULAS

=head2 Ith Pair

A pair of Lucas numbers L[k], L[k+1] can be calculated together by a
powering algorithm with two squares per doubling,

    L[2k]   = L[k]^2   - 2*(-1)^k
    L[2k+2] = L[k+1]^2 + 2*(-1)^k

    L[2k+1] = L[2k+2] - L[2k]

    if bit==0 then take L[2k], L[2k+1]
    if bit==1 then take L[2k+1], L[2k+2]

The 2*(-1)^k terms mean adding or subtracting 2 according to k odd or even.
This means add or subtract according to the previous bit handled.

=head2 Ith

For any trailing zero bits of i the final doublings can be done by L[2k]
alone which is one square per 0-bit.

    ith_pair(odd part of i) to get L[2k+1]  (ignore L[2k+2])
    loop L[2k] = L[k]^2 - 2*(-1)^k for each trailing 0-bit of i

=head2 Value to i Estimate

L[i] increases as a power of phi, the golden ratio.  The exact value is

    L[i] = phi^i + beta^i    # exactly

    phi = (1+sqrt(5))/2 = 1.618
    beta = -1/phi = -0.618

Since abs(beta)E<lt>1 the beta^i term quickly becomes small.  So taking a
log (natural logarithm) to get i,

    log(L[i]) ~= i*log(phi)
    i ~= log(L[i]) * 1/log(phi)

Or the same using log base 2 which can be estimated from the highest bit
position of a bignum,

    log2(L[i]) ~= i*log2(phi)
    i ~= log2(L[i]) * 1/log2(phi)

This is very close to the Fibonacci formula (see
L<Math::NumSeq::Fibonacci/Value to i Estimate>), being bigger by

    Lestimate(value) - Festimate(value)
      = log(value) / log(phi) - (log(value) + log(phi-beta)) / log(phi)
      = -log(phi-beta) / log(phi)
      = -1.67

On that basis it could even be close enough to take Lestimate = Festimate-1,
or vice-versa.

=head2 Ith Fibonacci and Lucas

It's also possible to calculate a Fibonacci F[k] and Lucas L[k] together by
a similar powering algorithm with two squares per doubling, but a couple of
extra factors,

    F[2k] = (F[k]+L[k])^2/2 - 3*F[k]^2 - 2*(-1)^k
    L[2k] =                   5*F[k]^2 + 2*(-1)^k

    F[2k+1] =    ((F[k]+L[k])/2)^2 + F[k]^2
    L[2k+1] = 5*(((F[k]+L[k])/2)^2 - F[k]^2) - 4*(-1)^k

Or the conversions between a pair of Fibonacci and Lucas are

    F[k]   = ( - L[k] + 2*L[k+1])/5
    F[k+1] = ( 2*L[k] +   L[k+1])/5   = (F[k] + L[k])/2

    L[k]   =  - F[k] + 2*F[k+1]
    L[k+1] =  2*F[k] +   F[k+1]       = (5*F[k] + L[k])/2

=cut

# MY-PARI-CHECK: [-1,2;2,1] * [-1,2;2,1] == [5,0;0,5]

=pod

=head1 SEE ALSO

L<Math::NumSeq>,
L<Math::NumSeq::Fibonacci>,
L<Math::NumSeq::Pell>

=head1 HOME PAGE

L<http://user42.tuxfamily.org/math-numseq/index.html>

=head1 LICENSE

Copyright 2010, 2011, 2012, 2013, 2014 Kevin Ryde

Math-NumSeq is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
version.

Math-NumSeq is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
more details.

You should have received a copy of the GNU General Public License along with
Math-NumSeq.  If not, see <http://www.gnu.org/licenses/>.

=cut