3
2

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 5 years have passed since last update.

Fortran で priority queue を作ってみた

Last updated at Posted at 2019-07-15

動機

priority queue は Dijkstra 法などに用いられ、コレクション内の最小・最大の要素を高速に取り出すことができる便利なデータ構造です。しかしながら Fortran には実装されていないので、自分で実装します。今回は integer を要素として持ち、コレクション内の最小値を取り出せるものを作成します。

priority queue とは

priority queue は追加する要素に優先度をつけ、取り出す際には優先度の高いものから取り出すことのできるデータ構造です。実装にはヒープ構造を用いるものが多いようです。 priority queue に要素を追加するのは $O(\mathrm{log}N)$ 、priority queue から要素を取り出すのは $O(\mathrm{log}N)$ の時間で行うことができます。

環境

macOS Sierra (10.12.6)
GNU Fortran (GCC) 6.3.0

プログラム

実際に実装したのが以下のものです。

priority_queue.f08
module mod_priority_queue
  implicit none

  type t_priority_queue
    integer :: num = 0
    integer, pointer :: heap(:) => null()
  end type t_priority_queue

contains

  subroutine offer(pq,item)
    implicit none
    type(t_priority_queue), intent(inout) :: pq
    integer, intent(in) :: item
    integer :: n, i, t
    integer, allocatable :: tmp(:)

    if (.not.associated(pq%heap)) allocate(pq%heap(1))
    if (pq%num == size(pq%heap)) then
      allocate(tmp(pq%num))
      tmp = pq%heap
      deallocate(pq%heap)
      allocate(pq%heap(2*pq%num))
      pq%heap(1:pq%num) = tmp
      deallocate(tmp)
    end if

    pq%num = pq%num+1
    pq%heap(pq%num) = item

    n = pq%num
    do while (n > 1)
      i = n/2
      if (pq%heap(n) < pq%heap(i)) then
        t = pq%heap(n)
        pq%heap(n) = pq%heap(i)
        pq%heap(i) = t
      end if
      n = i
    end do
    return
  end subroutine offer

  subroutine clear(pq)
    implicit none
    type(t_priority_queue), intent(inout) :: pq

    if (associated(pq%heap)) deallocate(pq%heap)
    pq%num = 0
    return
  end subroutine clear

  function poll(pq) result(item)
    implicit none
    type(t_priority_queue), intent(inout) :: pq
    integer :: item, n, i, j, tmp

    n = pq%num
    item = pq%heap(1)
    pq%heap(1) = pq%heap(n)
    pq%num = pq%num-1

    i = 1
    do while (2*i < n)
      j = 2*i
      if (j+1 < n .and. pq%heap(j+1) < pq%heap(j)) j = j+1
      if (pq%heap(j) < pq%heap(i)) then
        tmp = pq%heap(j)
        pq%heap(j) = pq%heap(i)
        pq%heap(i) = tmp
      end if
      i = j
    end do
    return
  end function poll

  function peek(pq) result(item)
    implicit none
    type(t_priority_queue), intent(inout) :: pq
    integer :: item

    item = pq%heap(1)
    return
  end function peek

end module mod_priority_queue

また、実装できているか確認するために以下のプログラムも書きました。

test_priority_queue.f08
program test_priority_queue
  use mod_priority_queue
  implicit none
  integer, parameter :: n = 6
  type(t_priority_queue) :: pq
  integer :: x(n), i

  call random_seed_clock()
  call randint(x,10*n)

  do i = 1, n
    call offer(pq,x(i))
    write(*,'("offer: ",i0," size: ",i0)') x(i), pq%num
  end do

  do while (pq%num > 0)
    write(*,'("poll: ",i0," size: ",i0)') poll(pq), pq%num
  end do

  call randint(x,10*n)

  do i = 1, n
    call offer(pq,x(i))
    write(*,'("offer: ",i0," size: ",i0)') x(i), pq%num
  end do

  write(*,'("peek: ",i0," size: ",i0)') peek(pq), pq%num

  call clear(pq)
  write(*,'("clear, size: ",i0)') pq%num

  call randint(x,10*n)

  do i = 1, n
    call offer(pq,x(i))
    write(*,'("offer: ",i0," size: ",i0)') x(i), pq%num
  end do

  write(*,'("peek: ",i0," size: ",i0)') peek(pq), pq%num

  stop
contains

  subroutine random_seed_clock()
    implicit none
    integer :: nseed, clock
    integer, allocatable :: seed(:)

    call system_clock(clock)
    call random_seed(size=nseed)
    allocate(seed(nseed))
    seed = clock
    call random_seed(put=seed)
    deallocate(seed)
    return
  end subroutine random_seed_clock

  subroutine randint(x,maxint)
    implicit none
    integer, intent(out) :: x(:)
    integer, intent(in) :: maxint
    real(8) :: r(size(x))

    call random_number(r)
    x = int(maxint*r)
    return
  end subroutine randint

end program test_priority_queue

実行結果は以下のとおりです。

offer: 22 size: 1
offer: 19 size: 2
offer: 37 size: 3
offer: 40 size: 4
offer: 56 size: 5
offer: 50 size: 6
poll: 19 size: 5
poll: 22 size: 4
poll: 37 size: 3
poll: 40 size: 2
poll: 50 size: 1
poll: 56 size: 0
offer: 43 size: 1
offer: 57 size: 2
offer: 41 size: 3
offer: 23 size: 4
offer: 2 size: 5
offer: 50 size: 6
peek: 2 size: 6
clear, size: 0
offer: 8 size: 1
offer: 43 size: 2
offer: 6 size: 3
offer: 8 size: 4
offer: 29 size: 5
offer: 38 size: 6
peek: 6 size: 6

これで逐次的に最小値を取り出せるデータ構造が手に入りました。

追記 2019/08/11

不必要な部分を削り、よりオブジェクト指向な書き方に変更しました (ソースコード) 。

3
2
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
3
2

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?