Simulated annealing: Difference between revisions
Line 1,177: | Line 1,177: | ||
39 29 28 18 19 9 8 7 6 5 4 3 2 1 0) |
39 29 28 18 19 9 8 7 6 5 4 3 2 1 0) |
||
</pre> |
</pre> |
||
=={{header|Fortran}}== |
|||
{{trans|Ada}} |
|||
<lang fortran>module simanneal_support |
|||
implicit none |
|||
! |
|||
! The following two integer kinds are meant to be treated as |
|||
! synonyms. |
|||
! |
|||
! selected_int_kind (2) = integers in the range of at least -100 to |
|||
! +100. |
|||
! |
|||
integer, parameter :: city_location_kind = selected_int_kind (2) |
|||
integer, parameter :: path_index_kind = city_location_kind |
|||
! |
|||
! selected_int_kind (1) = integers in the range of at least -10 to |
|||
! +10. |
|||
! |
|||
integer, parameter :: coordinate_kind = selected_int_kind(1) |
|||
! |
|||
! selected_real_kind (6) = floating point with at least 6 decimal |
|||
! digits of precision. |
|||
! |
|||
integer, parameter :: float_kind = selected_real_kind (6) |
|||
! |
|||
! Shorthand notations. |
|||
! |
|||
integer, parameter :: clk = city_location_kind |
|||
integer, parameter :: pik = path_index_kind |
|||
integer, parameter :: cok = coordinate_kind |
|||
integer, parameter :: flk = float_kind |
|||
type path_vector |
|||
integer(kind = clk) :: elem(0:99) |
|||
end type path_vector |
|||
contains |
|||
function random_integer (imin, imax) result (n) |
|||
integer, intent(in) :: imin, imax |
|||
integer :: n |
|||
real(kind = flk) :: randnum |
|||
call random_number (randnum) |
|||
n = imin + floor ((imax - imin + 1) * randnum) |
|||
end function random_integer |
|||
function i_coord (loc) result (i) |
|||
integer(kind = clk), intent(in) :: loc |
|||
integer(kind = cok) :: i |
|||
i = loc / 10_clk |
|||
end function i_coord |
|||
function j_coord (loc) result (j) |
|||
integer(kind = clk), intent(in) :: loc |
|||
integer(kind = cok) :: j |
|||
j = mod (loc, 10_clk) |
|||
end function j_coord |
|||
function location (i, j) result (loc) |
|||
integer(kind = cok), intent(in) :: i, j |
|||
integer(kind = clk) :: loc |
|||
loc = (10_clk * i) + j |
|||
end function location |
|||
subroutine randomize_path_vector (path) |
|||
type(path_vector), intent(out) :: path |
|||
integer(kind = pik) :: i, j |
|||
integer(kind = clk) :: xi, xj |
|||
do i = 0_pik, 99_pik |
|||
path%elem(i) = i |
|||
end do |
|||
! Do a Fisher-Yates shuffle of elements 1 .. 99. |
|||
do i = 1_pik, 98_pik |
|||
j = int (random_integer (i + 1, 99), kind = pik) |
|||
xi = path%elem(i) |
|||
xj = path%elem(j) |
|||
path%elem(i) = xj |
|||
path%elem(j) = xi |
|||
end do |
|||
end subroutine randomize_path_vector |
|||
function distance (loc1, loc2) result (dist) |
|||
integer(kind = clk), intent(in) :: loc1, loc2 |
|||
real(kind = flk) :: dist |
|||
integer(kind = cok) :: i1, j1 |
|||
integer(kind = cok) :: i2, j2 |
|||
integer :: di, dj |
|||
i1 = i_coord (loc1) |
|||
j1 = j_coord (loc1) |
|||
i2 = i_coord (loc2) |
|||
j2 = j_coord (loc2) |
|||
di = i1 - i2 |
|||
dj = j1 - j2 |
|||
dist = sqrt (real ((di * di) + (dj * dj), kind = flk)) |
|||
end function distance |
|||
function path_length (path) result (len) |
|||
type(path_vector), intent(in) :: path |
|||
real(kind = flk) :: len |
|||
integer(kind = pik) :: i |
|||
len = distance (path%elem(0_pik), path%elem(99_pik)) |
|||
do i = 0_pik, 98_pik |
|||
len = len + distance (path%elem(i), path%elem(i + 1_pik)) |
|||
end do |
|||
end function path_length |
|||
subroutine find_neighbors (loc, neighbors, num_neighbors) |
|||
integer(kind = clk), intent(in) :: loc |
|||
integer(kind = clk), intent(out) :: neighbors(1:8) |
|||
integer, intent(out) :: num_neighbors |
|||
integer(kind = cok) :: i, j |
|||
integer(kind = clk) :: c1, c2, c3, c4, c5, c6, c7, c8 |
|||
c1 = 0_clk |
|||
c2 = 0_clk |
|||
c3 = 0_clk |
|||
c4 = 0_clk |
|||
c5 = 0_clk |
|||
c6 = 0_clk |
|||
c7 = 0_clk |
|||
c8 = 0_clk |
|||
i = i_coord (loc) |
|||
j = j_coord (loc) |
|||
if (i < 9_cok) then |
|||
c1 = location (i + 1_cok, j) |
|||
if (j < 9_cok) then |
|||
c2 = location (i + 1_cok, j + 1_cok) |
|||
end if |
|||
if (0_cok < j) then |
|||
c3 = location (i + 1_cok, j - 1_cok) |
|||
end if |
|||
end if |
|||
if (0_cok < i) then |
|||
c4 = location (i - 1_cok, j) |
|||
if (j < 9_cok) then |
|||
c5 = location (i - 1_cok, j + 1_cok) |
|||
end if |
|||
if (0_cok < j) then |
|||
c6 = location (i - 1_cok, j - 1_cok) |
|||
end if |
|||
end if |
|||
if (j < 9_cok) then |
|||
c7 = location (i, j + 1_cok) |
|||
end if |
|||
if (0_cok < j) then |
|||
c8 = location (i, j - 1_cok) |
|||
end if |
|||
num_neighbors = 0 |
|||
call add_neighbor (c1) |
|||
call add_neighbor (c2) |
|||
call add_neighbor (c3) |
|||
call add_neighbor (c4) |
|||
call add_neighbor (c5) |
|||
call add_neighbor (c6) |
|||
call add_neighbor (c7) |
|||
call add_neighbor (c8) |
|||
contains |
|||
subroutine add_neighbor (neighbor) |
|||
integer(kind = clk), intent(in) :: neighbor |
|||
if (neighbor /= 0_clk) then |
|||
num_neighbors = num_neighbors + 1 |
|||
neighbors(num_neighbors) = neighbor |
|||
end if |
|||
end subroutine add_neighbor |
|||
end subroutine find_neighbors |
|||
function make_neighbor_path (path) result (neighbor_path) |
|||
type(path_vector), intent(in) :: path |
|||
type(path_vector) :: neighbor_path |
|||
integer(kind = clk) :: u, v |
|||
integer(kind = clk) :: neighbors(1:8) |
|||
integer :: num_neighbors |
|||
integer(kind = pik) :: j, iu, iv |
|||
neighbor_path = path |
|||
u = int (random_integer (1, 99), kind = clk) |
|||
call find_neighbors (u, neighbors, num_neighbors) |
|||
v = neighbors (random_integer (1, num_neighbors)) |
|||
j = 0_pik |
|||
iu = 0_pik |
|||
iv = 0_pik |
|||
do while (iu == 0_pik .or. iv == 0_pik) |
|||
if (neighbor_path%elem(j + 1) == u) then |
|||
iu = j + 1 |
|||
else if (neighbor_path%elem(j + 1) == v) then |
|||
iv = j + 1 |
|||
end if |
|||
j = j + 1 |
|||
end do |
|||
neighbor_path%elem(iu) = v |
|||
neighbor_path%elem(iv) = u |
|||
end function make_neighbor_path |
|||
function temperature (kT, kmax, k) result (temp) |
|||
real(kind = flk), intent(in) :: kT |
|||
integer, intent(in) :: kmax, k |
|||
real(kind = flk) :: temp |
|||
real(kind = flk) :: kf, kmaxf |
|||
kf = real (k, kind = flk) |
|||
kmaxf = real (kmax, kind = flk) |
|||
temp = kT * (1.0_flk - (kf / kmaxf)) |
|||
end function temperature |
|||
function probability (delta_E, T) result (prob) |
|||
real(kind = flk), intent(in) :: delta_E, T |
|||
real(kind = flk) :: prob |
|||
if (T == 0.0_flk) then |
|||
prob = 0.0_flk |
|||
else |
|||
prob = exp (-(delta_E / T)) |
|||
end if |
|||
end function probability |
|||
subroutine show (k, T, E) |
|||
integer, intent(in) :: k |
|||
real(kind = flk), intent(in) :: T, E |
|||
write (*, 10) k, T, E |
|||
10 format (1X, I7, 1X, F7.1, 1X, F10.2) |
|||
end subroutine show |
|||
subroutine display_path (path) |
|||
type(path_vector), intent(in) :: path |
|||
integer(kind = pik) :: i |
|||
999 format () |
|||
100 format (' ->') |
|||
110 format (' ') |
|||
120 format (I2) |
|||
do i = 0_pik, 99_pik |
|||
write (*, 120, advance = 'no') path%elem(i) |
|||
write (*, 100, advance = 'no') |
|||
if (mod (i, 8_pik) == 7_pik) then |
|||
write (*, 999, advance = 'yes') |
|||
else |
|||
write (*, 110, advance = 'no') |
|||
end if |
|||
end do |
|||
write (*, 120, advance = 'no') path%elem(0_pik) |
|||
end subroutine display_path |
|||
subroutine simulate_annealing (kT, kmax, initial_path, final_path) |
|||
real(kind = flk), intent(in) :: kT |
|||
integer, intent(in) :: kmax |
|||
type(path_vector), intent(in) :: initial_path |
|||
type(path_vector), intent(inout) :: final_path |
|||
integer :: kshow |
|||
integer :: k |
|||
real(kind = flk) :: E, E_trial, T |
|||
type(path_vector) :: path, trial |
|||
real(kind = flk) :: randnum |
|||
kshow = kmax / 10 |
|||
path = initial_path |
|||
E = path_length (path) |
|||
do k = 0, kmax |
|||
T = temperature (kT, kmax, k) |
|||
if (mod (k, kshow) == 0) call show (k, T, E) |
|||
trial = make_neighbor_path (path) |
|||
E_trial = path_length (trial) |
|||
if (E_trial <= E) then |
|||
path = trial |
|||
E = E_trial |
|||
else |
|||
call random_number (randnum) |
|||
if (randnum <= probability (E_trial - E, T)) then |
|||
path = trial |
|||
E = E_trial |
|||
end if |
|||
end if |
|||
end do |
|||
final_path = path |
|||
end subroutine simulate_annealing |
|||
end module simanneal_support |
|||
program simanneal |
|||
use, non_intrinsic :: simanneal_support |
|||
implicit none |
|||
real(kind = flk), parameter :: kT = 1.0_flk |
|||
integer, parameter :: kmax = 1000000 |
|||
type(path_vector) :: initial_path |
|||
type(path_vector) :: final_path |
|||
call random_seed |
|||
call randomize_path_vector (initial_path) |
|||
10 format () |
|||
20 format (' kT: ', F0.2) |
|||
30 format (' kmax: ', I0) |
|||
40 format (' k T E(s)') |
|||
50 format (' --------------------------') |
|||
60 format ('Final E(s): ', F0.2) |
|||
write (*, 10) |
|||
write (*, 20) kT |
|||
write (*, 30) kmax |
|||
write (*, 10) |
|||
write (*, 40) |
|||
write (*, 50) |
|||
call simulate_annealing (kT, kmax, initial_path, final_path) |
|||
write (*, 10) |
|||
call display_path (final_path) |
|||
write (*, 10) |
|||
write (*, 10) |
|||
write (*, 60) path_length (final_path) |
|||
write (*, 10) |
|||
end program simanneal</lang> |
|||
=={{header|Go}}== |
=={{header|Go}}== |
Revision as of 21:36, 1 May 2022
Quoted from the Wikipedia page : Simulated annealing (SA) is a probabilistic technique for approximating the global optimum of a given function. Simulated annealing interprets slow cooling as a slow decrease in the probability of temporarily accepting worse solutions as it explores the solution space.
Pseudo code from Wikipedia
Notations : T : temperature. Decreases to 0. s : a system state E(s) : Energy at s. The function we want to minimize ∆E : variation of E, from state s to state s_next P(∆E , T) : Probability to move from s to s_next. if ( ∆E < 0 ) P = 1 else P = exp ( - ∆E / T) . Decreases as T → 0 Pseudo-code: Let s = s0 -- initial state For k = 0 through kmax (exclusive): T ← temperature(k , kmax) Pick a random neighbour state , s_next ← neighbour(s) ∆E ← E(s) - E(s_next) If P(∆E , T) ≥ random(0, 1), move to the new state: s ← s_next Output: the final state s
Problem statement
We want to apply SA to the travelling salesman problem. There are 100 cities, numbered 0 to 99, located on a plane, at integer coordinates i,j : 0 <= i,j < 10 . The city at (i,j) has number 10*i + j. The cities are all connected : the graph is complete : you can go from one city to any other city in one step.
The salesman wants to start from city 0, visit all cities, each one time, and go back to city 0. The travel cost between two cities is the euclidian distance between there cities. The total travel cost is the total path length.
A path s is a sequence (0 a b ...z 0) where (a b ..z) is a permutation of the numbers (1 2 .. 99). The path length = E(s) is the sum d(0,a) + d(a,b) + ... + d(z,0) , where d(u,v) is the distance between two cities. Naturally, we want to minimize E(s).
Definition : The neighbours of a city are the closest cities at distance 1 horizontally/vertically, or √2 diagonally. A corner city (0,9,90,99) has 3 neighbours. A center city has 8 neighbours.
Distances between cities d ( 0, 7) → 7 d ( 0, 99) → 12.7279 d ( 23, 78) → 7.0711 d ( 33, 44) → 1.4142 // sqrt(2)
Task
Apply SA to the travelling salesman problem, using the following set of parameters/functions :
- kT = 1 (Multiplication by kT is a placeholder, representing computing temperature as a function of 1-k/kmax):
- temperature (k, kmax) = kT * (1 - k/kmax)
- neighbour (s) : Pick a random city u > 0 . Pick a random neighbour city v > 0 of u , among u's 8 (max) neighbours on the grid. Swap u and v in s . This gives the new state s_next.
- kmax = 1000_000
- s0 = a random permutation
For k = 0 to kmax by step kmax/10 , display k, T, E(s). Display the final state s_final, and E(s_final).
You will see that the Energy may grow to a local optimum, before decreasing to a global optimum.
Illustrated example Temperature charts
Numerical example
kT = 1 E(s0) = 529.9158 k: 0 T: 1 Es: 529.9158 k: 100000 T: 0.9 Es: 201.1726 k: 200000 T: 0.8 Es: 178.1723 k: 300000 T: 0.7 Es: 154.7069 k: 400000 T: 0.6 Es: 158.1412 <== local optimum k: 500000 T: 0.5 Es: 133.856 k: 600000 T: 0.4 Es: 129.5684 k: 700000 T: 0.3 Es: 112.6919 k: 800000 T: 0.2 Es: 105.799 k: 900000 T: 0.1 Es: 102.8284 k: 1000000 T: 0 Es: 102.2426 E(s_final) = 102.2426 Path s_final = ( 0 10 11 21 31 20 30 40 50 60 70 80 90 91 81 71 73 83 84 74 64 54 55 65 75 76 66 67 77 78 68 58 48 47 57 56 46 36 37 27 26 16 15 5 6 7 17 18 8 9 19 29 28 38 39 49 59 69 79 89 99 98 88 87 97 96 86 85 95 94 93 92 82 72 62 61 51 41 42 52 63 53 43 32 22 12 13 23 33 34 44 45 35 25 24 14 4 3 2 1 0)
Extra credit
Tune the parameters kT, kmax, or use different temperature() and/or neighbour() functions to demonstrate a quicker convergence, or a better optimum.
Ada
This implementation is adapted from the C, which was adapted from the Scheme. It uses fixed-point numbers for no better reason than to demonstrate that Ada has fixed-point numbers support built in.
<lang ada>----------------------------------------------------------------------
--
-- The Rosetta Code simulated annealing task in Ada.
--
-- This implementation demonstrates that Ada has fixed-point numbers
-- support built in. Otherwise there is no particular reason I used
-- fixed-point instead of floating-point numbers.
--
-- (Actually, for the square root and exponential, I cheat and use the
-- floating-point functions.)
--
with Ada.Numerics.Discrete_Random; with Ada.Numerics.Elementary_Functions; with Ada.Text_IO; use Ada.Text_IO;
procedure simanneal is
Bigint : constant := 1_000_000_000; Bigfpt : constant := 1_000_000_000.0;
-- Fixed point numbers. type Fixed_Point is delta 0.000_01 range 0.0 .. Bigfpt;
-- Integers. subtype Zero_or_One is Integer range 0 .. 1; subtype Coordinate is Integer range 0 .. 9; subtype City_Location is Integer range 0 .. 99; subtype Nonzero_City_Location is City_Location range 1 .. 99; subtype Path_Index is City_Location; subtype Nonzero_Path_Index is Nonzero_City_Location;
-- Arrays. type Path_Vector is array (Path_Index) of City_Location; type Neighborhood_Array is array (1 .. 8) of City_Location;
-- Random numbers. subtype Random_Number is Integer range 0 .. Bigint - 1; package Random_Numbers is new Ada.Numerics.Discrete_Random (Random_Number); use Random_Numbers;
gen : Generator;
function Randnum return Fixed_Point is begin return (Fixed_Point (Random (gen)) / Fixed_Point (Bigfpt)); end Randnum;
function Random_Natural (imin : Natural; imax : Natural) return Natural is begin -- There may be a tiny bias in the result, due to imax-imin+1 not -- being a divisor of Bigint. The algorithm should work, anyway. return imin + (Random (gen) rem (imax - imin + 1)); end Random_Natural;
function Random_City_Location (minloc : City_Location; maxloc : City_Location) return City_Location is begin return City_Location (Random_Natural (minloc, maxloc)); end Random_City_Location;
function Random_Path_Index (imin : Path_Index; imax : Path_Index) return Path_Index is begin return Random_City_Location (imin, imax); end Random_Path_Index;
package Natural_IO is new Ada.Text_IO.Integer_IO (Natural); package City_Location_IO is new Ada.Text_IO.Integer_IO (City_Location); package Fixed_Point_IO is new Ada.Text_IO.Fixed_IO (Fixed_Point);
function sqrt (x : Fixed_Point) return Fixed_Point is begin -- Cheat by using the floating-point routine. It is an exercise -- for the reader to write a true fixed-point function. return Fixed_Point (Ada.Numerics.Elementary_Functions.Sqrt (Float (x))); end sqrt;
function expneg (x : Fixed_Point) return Fixed_Point is begin -- Cheat by using the floating-point routine. It is an exercise -- for the reader to write a true fixed-point function. return Fixed_Point (Ada.Numerics.Elementary_Functions.Exp (-Float (x))); end expneg;
function i_Coord (loc : City_Location) return Coordinate is begin return loc / 10; end i_Coord;
function j_Coord (loc : City_Location) return Coordinate is begin return loc rem 10; end j_Coord;
function Location (i : Coordinate; j : Coordinate) return City_Location is begin return (10 * i) + j; end Location;
function distance (loc1 : City_Location; loc2 : City_Location) return Fixed_Point is i1, j1 : Coordinate; i2, j2 : Coordinate; di, dj : Coordinate; begin i1 := i_Coord (loc1); j1 := j_Coord (loc1); i2 := i_Coord (loc2); j2 := j_Coord (loc2); di := (if i1 < i2 then i2 - i1 else i1 - i2); dj := (if j1 < j2 then j2 - j1 else j1 - j2); return sqrt (Fixed_Point ((di * di) + (dj * dj))); end distance;
procedure Randomize_Path_Vector (path : out Path_Vector) is j : Nonzero_Path_Index; xi, xj : Nonzero_City_Location; begin for i in 0 .. 99 loop path (i) := i; end loop;
-- Do a Fisher-Yates shuffle of elements 1 .. 99. for i in 1 .. 98 loop j := Random_Path_Index (i + 1, 99); xi := path (i); xj := path (j); path (i) := xj; path (j) := xi; end loop; end Randomize_Path_Vector;
function Path_Length (path : Path_Vector) return Fixed_Point is len : Fixed_Point; begin len := distance (path (0), path (99)); for i in 0 .. 98 loop len := len + distance (path (i), path (i + 1)); end loop; return len; end Path_Length;
-- Switch the index of s to switch which s is current and which is -- the trial vector. s : array (0 .. 1) of Path_Vector;
Current : Zero_or_One;
function Trial return Zero_or_One is begin return 1 - Current; end Trial;
procedure Accept_Trial is begin Current := 1 - Current; end Accept_Trial;
procedure Find_Neighbors (loc : City_Location; neighbors : out Neighborhood_Array; num_neighbors : out Integer) is i, j : Coordinate; c1, c2, c3, c4, c5, c6, c7, c8 : City_Location := 0;
procedure Add_Neighbor (neighbor : City_Location) is begin if neighbor /= 0 then num_neighbors := num_neighbors + 1; neighbors (num_neighbors) := neighbor; end if; end Add_Neighbor;
begin i := i_Coord (loc); j := j_Coord (loc);
if i < 9 then c1 := Location (i + 1, j); if j < 9 then c2 := Location (i + 1, j + 1); end if; if 0 < j then c3 := Location (i + 1, j - 1); end if; end if; if 0 < i then c4 := Location (i - 1, j); if j < 9 then c5 := Location (i - 1, j + 1); end if; if 0 < j then c6 := Location (i - 1, j - 1); end if; end if; if j < 9 then c7 := Location (i, j + 1); end if; if 0 < j then c8 := Location (i, j - 1); end if;
num_neighbors := 0; Add_Neighbor (c1); Add_Neighbor (c2); Add_Neighbor (c3); Add_Neighbor (c4); Add_Neighbor (c5); Add_Neighbor (c6); Add_Neighbor (c7); Add_Neighbor (c8); end Find_Neighbors;
procedure Make_Neighbor_Path is u, v : City_Location; neighbors : Neighborhood_Array; num_neighbors : Integer; j, iu, iv : Path_Index; begin for i in 0 .. 99 loop s (Trial) := s (Current); end loop;
u := Random_City_Location (1, 99); Find_Neighbors (u, neighbors, num_neighbors); v := neighbors (Random_Natural (1, num_neighbors));
j := 0; iu := 0; iv := 0; while iu = 0 or iv = 0 loop if s (Trial) (j + 1) = u then iu := j + 1; elsif s (Trial) (j + 1) = v then iv := j + 1; end if; j := j + 1; end loop; s (Trial) (iu) := v; s (Trial) (iv) := u; end Make_Neighbor_Path;
function Temperature (kT : Fixed_Point; kmax : Natural; k : Natural) return Fixed_Point is begin return kT * (Fixed_Point (1) - (Fixed_Point (k) / Fixed_Point (kmax))); end Temperature;
function Probability (delta_E : Fixed_Point; T : Fixed_Point) return Fixed_Point is prob : Fixed_Point; begin if T = Fixed_Point (0.0) then prob := Fixed_Point (0.0); else prob := expneg (delta_E / T); end if; return prob; end Probability;
procedure Show (k : Natural; T : Fixed_Point; E : Fixed_Point) is begin Put (" "); Natural_IO.Put (k, Width => 7); Put (" "); Fixed_Point_IO.Put (T, Fore => 5, Aft => 1); Put (" "); Fixed_Point_IO.Put (E, Fore => 7, Aft => 2); Put_Line (""); end Show;
procedure Display_Path (path : Path_Vector) is begin for i in 0 .. 99 loop City_Location_IO.Put (path (i), Width => 2); Put (" ->"); if i rem 8 = 7 then Put_Line (""); else Put (" "); end if; end loop; City_Location_IO.Put (path (0), Width => 2); end Display_Path;
procedure Simulate_Annealing (kT : Fixed_Point; kmax : Natural) is kshow : Natural := kmax / 10; E : Fixed_Point; E_trial : Fixed_Point; T : Fixed_Point; P : Fixed_Point; begin E := Path_Length (s (Current)); for k in 0 .. kmax loop T := Temperature (kT, kmax, k); if k rem kshow = 0 then Show (k, T, E); end if; Make_Neighbor_Path; E_trial := Path_Length (s (Trial)); if E_trial <= E then Accept_Trial; E := E_trial; else P := Probability (E_trial - E, T); if P = Fixed_Point (1) or else Randnum <= P then Accept_Trial; E := E_trial; end if; end if; end loop; end Simulate_Annealing;
kT : constant := Fixed_Point (1.0); kmax : constant := 1_000_000;
begin
Reset (gen);
Current := 0; Randomize_Path_Vector (s (Current));
Put_Line (""); Put (" kT:"); Put_Line (Fixed_Point'Image (kT)); Put (" kmax:"); Put_Line (Natural'Image (kmax)); Put_Line (""); Put_Line (" k T E(s)"); Simulate_Annealing (kT, kmax); Put_Line (""); Put_Line ("Final path:"); Display_Path (s (Current)); Put_Line (""); Put_Line (""); Put ("Final E(s): "); Fixed_Point_IO.Put (Path_Length (s (Current)), Fore => 3, Aft => 2); Put_Line (""); Put_Line ("");
end simanneal;
</lang>
- Output:
An example run. In the following, you could use gnatmake instead of gprbuild.
$ gprbuild -q simanneal.adb && ./simanneal kT: 1.00000 kmax: 1000000 k T E(s) 0 1.0 525.23 100000 0.9 189.97 200000 0.8 180.33 300000 0.7 153.31 400000 0.6 156.18 500000 0.5 136.17 600000 0.4 119.56 700000 0.3 110.51 800000 0.2 106.21 900000 0.1 102.89 1000000 0.0 102.89 Final path: 0 -> 10 -> 11 -> 21 -> 20 -> 30 -> 31 -> 32 -> 22 -> 23 -> 33 -> 43 -> 42 -> 52 -> 51 -> 41 -> 40 -> 50 -> 60 -> 70 -> 80 -> 90 -> 91 -> 92 -> 93 -> 84 -> 94 -> 95 -> 85 -> 86 -> 96 -> 97 -> 98 -> 99 -> 89 -> 88 -> 87 -> 77 -> 67 -> 57 -> 58 -> 68 -> 78 -> 79 -> 69 -> 59 -> 49 -> 39 -> 29 -> 19 -> 9 -> 8 -> 7 -> 6 -> 25 -> 24 -> 34 -> 35 -> 44 -> 54 -> 53 -> 63 -> 62 -> 61 -> 71 -> 81 -> 72 -> 82 -> 83 -> 73 -> 74 -> 64 -> 65 -> 75 -> 76 -> 66 -> 56 -> 55 -> 45 -> 46 -> 47 -> 48 -> 38 -> 37 -> 36 -> 26 -> 27 -> 28 -> 18 -> 17 -> 16 -> 15 -> 5 -> 4 -> 14 -> 3 -> 13 -> 12 -> 2 -> 1 -> 0 Final E(s): 102.89
C
For your platform you might have to modify parts, such as the call to getentropy(3).
You can easily change the kind of floating point used. I apologize for false precision in printouts using the default single precision floating point.
Some might notice the calculations of random integers are done in a way that may introduce a bias, which is miniscule as long as the integer is much smaller than 2 to the 31st power. I mention this now so no one will complain about it later.
<lang c>#include <math.h>
- include <stdio.h>
- include <stdlib.h>
- include <stdint.h>
- include <string.h>
- include <unistd.h>
- define VECSZ 100
- define STATESZ 64
typedef float floating_pt;
- define EXP expf
- define SQRT sqrtf
static floating_pt randnum (void) {
return (floating_pt) ((double) (random () & 2147483647) / 2147483648.0);
}
static void shuffle (uint8_t vec[], size_t i, size_t n) {
/* A Fisher-Yates shuffle of n elements of vec, starting at index i. */ for (size_t j = 0; j != n; j += 1) { size_t k = i + j + (random () % (n - j)); uint8_t xi = vec[i]; uint8_t xk = vec[k]; vec[i] = xk; vec[k] = xi; }
}
static void init_s (uint8_t vec[VECSZ]) {
for (uint8_t j = 0; j != VECSZ; j += 1) vec[j] = j; shuffle (vec, 1, VECSZ - 1);
}
static inline void add_neighbor (uint8_t neigh[8],
unsigned int *neigh_size, uint8_t neighbor)
{
if (neighbor != 0) { neigh[*neigh_size] = neighbor; *neigh_size += 1; }
}
static void neighborhood (uint8_t neigh[8],
unsigned int *neigh_size, uint8_t city)
{
/* Find all non-zero neighbor cities. */
const uint8_t i = city / 10; const uint8_t j = city % 10;
uint8_t c0 = 0; uint8_t c1 = 0; uint8_t c2 = 0; uint8_t c3 = 0; uint8_t c4 = 0; uint8_t c5 = 0; uint8_t c6 = 0; uint8_t c7 = 0;
if (i < 9) { c0 = (10 * (i + 1)) + j; if (j < 9) c1 = (10 * (i + 1)) + (j + 1); if (0 < j) c2 = (10 * (i + 1)) + (j - 1); } if (0 < i) { c3 = (10 * (i - 1)) + j; if (j < 9) c4 = (10 * (i - 1)) + (j + 1); if (0 < j) c5 = (10 * (i - 1)) + (j - 1); } if (j < 9) c6 = (10 * i) + (j + 1); if (0 < j) c7 = (10 * i) + (j - 1);
*neigh_size = 0; add_neighbor (neigh, neigh_size, c0); add_neighbor (neigh, neigh_size, c1); add_neighbor (neigh, neigh_size, c2); add_neighbor (neigh, neigh_size, c3); add_neighbor (neigh, neigh_size, c4); add_neighbor (neigh, neigh_size, c5); add_neighbor (neigh, neigh_size, c6); add_neighbor (neigh, neigh_size, c7);
}
static floating_pt distance (uint8_t m, uint8_t n) {
const uint8_t im = m / 10; const uint8_t jm = m % 10; const uint8_t in = n / 10; const uint8_t jn = n % 10; const int di = (int) im - (int) in; const int dj = (int) jm - (int) jn; return SQRT ((di * di) + (dj * dj));
}
static floating_pt path_length (uint8_t vec[VECSZ]) {
floating_pt len = distance (vec[0], vec[VECSZ - 1]); for (size_t j = 0; j != VECSZ - 1; j += 1) len += distance (vec[j], vec[j + 1]); return len;
}
static void swap_s_elements (uint8_t vec[], uint8_t u, uint8_t v) {
size_t j = 1; size_t iu = 0; size_t iv = 0; while (iu == 0 || iv == 0) { if (vec[j] == u) iu = j; else if (vec[j] == v) iv = j; j += 1; } vec[iu] = v; vec[iv] = u;
}
static void update_s (uint8_t vec[]) {
const uint8_t u = 1 + (random () % (VECSZ - 1)); uint8_t neighbors[8]; unsigned int num_neighbors; neighborhood (neighbors, &num_neighbors, u); const uint8_t v = neighbors[random () % num_neighbors]; swap_s_elements (vec, u, v);
}
static inline void copy_s (uint8_t dst[VECSZ], uint8_t src[VECSZ]) {
memcpy (dst, src, VECSZ * (sizeof src[0]));
}
static void trial_s (uint8_t trial[VECSZ], uint8_t vec[VECSZ]) {
copy_s (trial, vec); update_s (trial);
}
static floating_pt temperature (floating_pt kT, unsigned int kmax, unsigned int k) {
return kT * (1 - ((floating_pt) k / (floating_pt) kmax));
}
static floating_pt probability (floating_pt delta_E, floating_pt T) {
floating_pt prob; if (delta_E < 0) prob = 1; else if (T == 0) prob = 0; else prob = EXP (-(delta_E / T)); return prob;
}
static void show (unsigned int k, floating_pt T, floating_pt E) {
printf (" %7u %7.1f %13.5f\n", k, (double) T, (double) E);
}
static void simulate_annealing (floating_pt kT,
unsigned int kmax, uint8_t s[VECSZ])
{
uint8_t trial[VECSZ];
unsigned int kshow = kmax / 10; floating_pt E = path_length (s); for (unsigned int k = 0; k != kmax + 1; k += 1) { const floating_pt T = temperature (kT, kmax, k); if (k % kshow == 0) show (k, T, E); trial_s (trial, s); const floating_pt E_trial = path_length (trial); const floating_pt delta_E = E_trial - E; const floating_pt P = probability (delta_E, T); if (P == 1 || randnum () <= P) { copy_s (s, trial); E = E_trial; } }
}
static void display_path (uint8_t vec[VECSZ]) {
for (size_t i = 0; i != VECSZ; i += 1) { const uint8_t x = vec[i]; printf ("%2u ->", (unsigned int) x); if ((i % 8) == 7) printf ("\n"); else printf (" "); } printf ("%2u\n", vec[0]);
}
int main (void) {
char state[STATESZ]; uint32_t seed[1]; int status = getentropy (&seed[0], sizeof seed[0]); if (status < 0) seed[0] = 1; initstate (seed[0], state, STATESZ);
floating_pt kT = 1.0; unsigned int kmax = 1000000;
uint8_t s[VECSZ]; init_s (s);
printf ("\n"); printf (" kT: %f\n", (double) kT); printf (" kmax: %u\n", kmax); printf ("\n"); printf (" k T E(s)\n"); printf (" -----------------------------\n"); simulate_annealing (kT, kmax, s); printf ("\n"); display_path (s); printf ("\n"); printf ("Final E(s): %.5f\n", (double) path_length (s)); printf ("\n");
return 0;
}</lang>
- Output:
An example run:
$ cc -Ofast -march=native simanneal.c -lm && ./a.out kT: 1.000000 kmax: 1000000 k T E(s) ----------------------------- 0 1.0 383.25223 100000 0.9 195.81190 200000 0.8 186.58963 300000 0.7 152.46564 400000 0.6 143.59039 500000 0.5 130.91815 600000 0.4 126.53572 700000 0.3 112.85691 800000 0.2 103.72134 900000 0.1 103.07108 1000000 0.0 102.24265 0 -> 10 -> 20 -> 21 -> 31 -> 30 -> 40 -> 50 -> 60 -> 61 -> 62 -> 72 -> 82 -> 81 -> 71 -> 70 -> 80 -> 90 -> 91 -> 92 -> 93 -> 94 -> 84 -> 83 -> 73 -> 63 -> 64 -> 65 -> 66 -> 76 -> 86 -> 87 -> 77 -> 67 -> 68 -> 58 -> 57 -> 56 -> 55 -> 45 -> 35 -> 26 -> 36 -> 46 -> 47 -> 48 -> 38 -> 37 -> 27 -> 28 -> 29 -> 39 -> 49 -> 59 -> 69 -> 79 -> 78 -> 88 -> 89 -> 99 -> 98 -> 97 -> 96 -> 95 -> 85 -> 75 -> 74 -> 54 -> 53 -> 52 -> 51 -> 41 -> 42 -> 43 -> 44 -> 34 -> 33 -> 32 -> 22 -> 23 -> 14 -> 4 -> 5 -> 6 -> 7 -> 8 -> 9 -> 19 -> 18 -> 17 -> 16 -> 15 -> 25 -> 24 -> 13 -> 3 -> 2 -> 12 -> 11 -> 1 -> 0 Final E(s): 102.24265
C++
Compiler: MSVC (19.27.29111 for x64) <lang c++>
- include<array>
- include<utility>
- include<cmath>
- include<random>
- include<iostream>
using coord = std::pair<int,int>; constexpr size_t numCities = 100;
// CityID with member functions to get position struct CityID{
int v{-1}; CityID() = default; constexpr explicit CityID(int i) noexcept : v(i){} constexpr explicit CityID(coord ij) : v(ij.first * 10 + ij.second) { if(ij.first < 0 || ij.first > 9 || ij.second < 0 || ij.second > 9){ throw std::logic_error("Cannot construct CityID from invalid coordinates!"); } }
constexpr coord get_pos() const noexcept { return {v/10,v%10}; }
}; bool operator==(CityID const& lhs, CityID const& rhs) {return lhs.v == rhs.v;}
// Function for distance between two cities double dist(coord city1, coord city2){
double diffx = city1.first - city2.first; double diffy = city1.second - city2.second; return std::sqrt(std::pow(diffx, 2) + std::pow(diffy,2));
}
// Function for total distance travelled template<size_t N> double dist(std::array<CityID,N> cities){
double sum = 0; for(auto it = cities.begin(); it < cities.end() - 1; ++it){ sum += dist(it->get_pos(),(it+1)->get_pos()); } sum += dist((cities.end()-1)->get_pos(), cities.begin()->get_pos()); return sum;
}
// 8 nearest cities, id cannot be at the border and has to have 8 valid neighbors constexpr std::array<CityID,8> get_nearest(CityID id){
auto const ij = id.get_pos(); auto const i = ij.first; auto const j = ij.second; return { CityID({i-1,j-1}), CityID({i ,j-1}), CityID({i+1,j-1}), CityID({i-1,j }), CityID({i+1,j }), CityID({i-1,j+1}), CityID({i ,j+1}), CityID({i+1,j+1}), };
}
// Function for formating of results constexpr int get_num_digits(int num){
int digits = 1; while(num /= 10){ ++digits; } return digits;
}
// Function for shuffeling of initial state template<typename It, typename RandomEngine> void shuffle(It first, It last, RandomEngine& rand_eng){
for(auto i=(last-first)-1; i>0; --i){ std::uniform_int_distribution<int> dist(0,i); std::swap(first[i], first[dist(rand_eng)]); }
}
class SA{
int kT{1}; int kmax{1'000'000}; std::array<CityID,numCities> s; std::default_random_engine rand_engine{0};
// Temperature double temperature(int k) const { return kT * (1.0 - static_cast<double>(k) / kmax); }
// Probabilty of acceptance between 0.0 and 1.0 double P(double dE, double T){ if(dE < 0){ return 1; } else{ return std::exp(-dE/T); } }
// Permutation of state through swapping of cities in travel path std::array<CityID,numCities> next_permut(std::array<CityID,numCities> cities){ std::uniform_int_distribution<> disx(1,8); std::uniform_int_distribution<> disy(1,8); auto randCity = CityID({disx(rand_engine),disy(rand_engine)}); // Select city which is not at the border, since all neighbors are valid under this condition and all permutations are still possible auto neighbors = get_nearest(randCity); // Get list of nearest neighbors std::uniform_int_distribution<> selector(0,neighbors.size()-1); // [0,7] const auto [i,j] = randCity.get_pos(); auto randNeighbor = neighbors[selector(rand_engine)]; // Since randCity is not at the border, all 8 neighbors are valid auto cityit1 = std::find(cities.begin(),cities.end(),randCity); // Find selected city in state auto cityit2 = std::find(cities.begin(), cities.end(),randNeighbor);// Find selected neighbor in state std::swap(*cityit1, *cityit2); // Swap city and neighbor return cities; }
// Logging function for progress output void log_progress(int k, double T, double E) const { auto nk = get_num_digits(kmax); auto nt = get_num_digits(kT); std::printf("k: %*i | T: %*.3f | E(s): %*.4f\n", nk, k, nt, T, 3, E); }
public:
// Initialize state with integers from 0 to 99 SA() { int i = 0; for(auto it = s.begin(); it != s.end(); ++it){ *it = CityID(i); ++i; } shuffle(s.begin(),s.end(),rand_engine); }
// Logging function for final path void log_path(){ for(size_t idx = 0; idx < s.size(); ++idx){ std::printf("%*i -> ", 2, s[idx].v); if((idx + 1)%20 == 0){ std::printf("\n"); } } std::printf("%*i", 2, s[0].v); }
// Core simulated annealing algorithm std::array<CityID,numCities> run(){ std::cout << "kT == " << kT << "\n" << "kmax == " << kmax << "\n" << "E(s0) == " << dist(s) << "\n"; for(int k = 0; k < kmax; ++k){ auto T = temperature(k); auto const E1 = dist(s); auto s_next{next_permut(s)}; auto const E2 = dist(s_next); auto const dE = E2 - E1; // lower is better std::uniform_real_distribution dist(0.0, 1.0); auto E = E1; if(P(dE,T) >= dist(rand_engine)){ s = s_next; E = E2; } if(k%100000 == 0){ log_progress(k,T,E1); } } log_progress(kmax,0.0,dist(s)); std::cout << "\nFinal path: \n"; log_path(); return s; }
};
int main(){
SA sa{}; auto result = sa.run(); // Run simulated annealing and log progress and result std::cin.get(); return 0;
} </lang>
- Output:
kT == 1 kmax == 1000000 E(s0) == 529.423 k: 0 | T: 1.000 | E(s): 529.4231 k: 100000 | T: 0.900 | E(s): 197.5111 k: 200000 | T: 0.800 | E(s): 183.7467 k: 300000 | T: 0.700 | E(s): 165.8442 k: 400000 | T: 0.600 | E(s): 143.8588 k: 500000 | T: 0.500 | E(s): 133.9247 k: 600000 | T: 0.400 | E(s): 125.9499 k: 700000 | T: 0.300 | E(s): 115.8657 k: 800000 | T: 0.200 | E(s): 107.8635 k: 900000 | T: 0.100 | E(s): 102.4853 k: 1000000 | T: 0.000 | E(s): 102.4853 Final path: 71 -> 61 -> 51 -> 50 -> 60 -> 70 -> 80 -> 90 -> 91 -> 92 -> 82 -> 83 -> 93 -> 94 -> 84 -> 85 -> 95 -> 96 -> 86 -> 76 -> 75 -> 74 -> 64 -> 65 -> 55 -> 45 -> 44 -> 54 -> 53 -> 43 -> 33 -> 34 -> 35 -> 26 -> 16 -> 6 -> 7 -> 17 -> 27 -> 37 -> 47 -> 57 -> 58 -> 48 -> 38 -> 28 -> 18 -> 8 -> 9 -> 19 -> 29 -> 39 -> 49 -> 59 -> 69 -> 68 -> 78 -> 79 -> 88 -> 89 -> 99 -> 98 -> 97 -> 87 -> 77 -> 67 -> 66 -> 56 -> 46 -> 36 -> 25 -> 24 -> 14 -> 15 -> 5 -> 4 -> 3 -> 2 -> 11 -> 21 -> 31 -> 41 -> 40 -> 30 -> 20 -> 10 -> 0 -> 1 -> 12 -> 13 -> 23 -> 22 -> 32 -> 42 -> 52 -> 62 -> 63 -> 73 -> 72 -> 81 -> 71
EchoLisp
<lang scheme> (lib 'math)
- distances
(define (d ci cj) (distance (% ci 10) (quotient ci 10) (% cj 10) (quotient cj 10))) (define _dists (build-vector 10000 (lambda (ij) (d (quotient ij 100) (% ij 100))))) (define-syntax-rule (dist ci cj) [_dists (+ ci (* 100 cj))])
- E(s) = length(path)
(define (Es path) (define lpath (vector->list path)) (for/sum ((ci lpath) (cj (rest lpath))) (dist ci cj)))
- temperature() function
(define (T k kmax kT) (* kT (- 1 (// k kmax))))
- |
- alternative temperature()
- must be decreasing with k increasing and → 0
(define (T k kmax kT) (* kT (- 1 (sin (* PI/2 (// k kmax)))))) |#
- ∆E = Es_new - Es_old > 0
- probability to move if ∆E > 0, → 0 when T → 0 (frozen state)
(define (P ∆E k kmax kT) (exp (// (- ∆E ) (T k kmax kT))))
- ∆E from path ( .. a u b .. c v d ..) to (.. a v b ... c u d ..)
- ∆E before swapping (u,v)
- Quicker than Es(s_next) - Es(s)
(define (dE s u v)
- old
(define a (dist [s (1- u)] [s u])) (define b (dist [s (1+ u)] [s u])) (define c (dist [s (1- v)] [s v])) (define d (dist [s (1+ v)] [s v]))
- new
(define na (dist [s (1- u)] [s v])) (define nb (dist [s (1+ u)] [s v])) (define nc (dist [s (1- v)] [s u])) (define nd (dist [s (1+ v)] [s u]))
(cond ((= v (1+ u)) (- (+ na nd) (+ a d))) ((= u (1+ v)) (- (+ nc nb) (+ c b))) (else (- (+ na nb nc nd) (+ a b c d)))))
- all 8 neighbours
(define dirs #(1 -1 10 -10 9 11 -11 -9))
(define (sa kmax (kT 10)) (define s (list->vector (cons 0 (append (shuffle (range 1 100)) 0)))) (printf "E(s0) %d" (Es s)) ;; random starter (define Emin (Es s)) ;; E0
(for ((k kmax)) (when (zero? (% k (/ kmax 10))) (printf "k: %10d T: %8.4d Es: %8.4d" k (T k kmax kT) (Es s)) )
(define u (1+ (random 99))) ;; city index 1 99 (define cv (+ [s u] [dirs (random 8)])) ;; city number #:continue (or (> cv 99) (<= cv 0)) #:continue (> (dist [s u] cv) 5) ;; check true neighbour (eg 0 9) (define v (vector-index cv s 1)) ;; city index
(define ∆e (dE s u v)) (when (or (< ∆e 0) ;; always move if negative (>= (P ∆e k kmax kT) (random))) (vector-swap! s u v) (+= Emin ∆e))
;; (assert (= (round Emin) (round (Es s)))) ) ;; for
(printf "k: %10d T: %8.4d Es: %8.4d" kmax (T (1- kmax) kmax kT) (Es s)) (s-plot s 0) (printf "E(s_final) %d" Emin) (writeln 'Path s)) </lang>
- Output:
(sa 1000000 1) E(s0) 501.0909 k: 0 T: 1 Es: 501.0909 k: 100000 T: 0.9 Es: 167.3632 k: 200000 T: 0.8 Es: 160.7791 k: 300000 T: 0.7 Es: 166.8746 k: 400000 T: 0.6 Es: 142.579 k: 500000 T: 0.5 Es: 131.0657 k: 600000 T: 0.4 Es: 116.9214 k: 700000 T: 0.3 Es: 110.8569 k: 800000 T: 0.2 Es: 103.3137 k: 900000 T: 0.1 Es: 102.4853 k: 1000000 T: 0 Es: 102.4853 E(s_final) 102.4853 Path #( 0 10 20 30 40 50 60 70 71 61 62 53 63 64 54 44 45 55 65 74 84 83 73 72 82 81 80 90 91 92 93 94 95 85 75 76 86 96 97 98 99 88 89 79 69 59 49 48 47 57 58 68 78 87 77 67 66 56 46 36 35 25 24 34 33 32 43 42 52 51 41 31 21 11 12 22 23 13 14 15 16 17 26 27 37 38 39 29 28 18 19 9 8 7 6 5 4 3 2 1 0)
Fortran
<lang fortran>module simanneal_support
implicit none
! ! The following two integer kinds are meant to be treated as ! synonyms. ! ! selected_int_kind (2) = integers in the range of at least -100 to ! +100. ! integer, parameter :: city_location_kind = selected_int_kind (2) integer, parameter :: path_index_kind = city_location_kind
! ! selected_int_kind (1) = integers in the range of at least -10 to ! +10. ! integer, parameter :: coordinate_kind = selected_int_kind(1)
! ! selected_real_kind (6) = floating point with at least 6 decimal ! digits of precision. ! integer, parameter :: float_kind = selected_real_kind (6)
! ! Shorthand notations. ! integer, parameter :: clk = city_location_kind integer, parameter :: pik = path_index_kind integer, parameter :: cok = coordinate_kind integer, parameter :: flk = float_kind
type path_vector integer(kind = clk) :: elem(0:99) end type path_vector
contains
function random_integer (imin, imax) result (n) integer, intent(in) :: imin, imax integer :: n
real(kind = flk) :: randnum
call random_number (randnum) n = imin + floor ((imax - imin + 1) * randnum) end function random_integer
function i_coord (loc) result (i) integer(kind = clk), intent(in) :: loc integer(kind = cok) :: i
i = loc / 10_clk end function i_coord function j_coord (loc) result (j) integer(kind = clk), intent(in) :: loc integer(kind = cok) :: j
j = mod (loc, 10_clk) end function j_coord
function location (i, j) result (loc) integer(kind = cok), intent(in) :: i, j integer(kind = clk) :: loc
loc = (10_clk * i) + j end function location
subroutine randomize_path_vector (path) type(path_vector), intent(out) :: path
integer(kind = pik) :: i, j integer(kind = clk) :: xi, xj
do i = 0_pik, 99_pik path%elem(i) = i end do
! Do a Fisher-Yates shuffle of elements 1 .. 99. do i = 1_pik, 98_pik j = int (random_integer (i + 1, 99), kind = pik) xi = path%elem(i) xj = path%elem(j) path%elem(i) = xj path%elem(j) = xi end do end subroutine randomize_path_vector
function distance (loc1, loc2) result (dist) integer(kind = clk), intent(in) :: loc1, loc2 real(kind = flk) :: dist
integer(kind = cok) :: i1, j1 integer(kind = cok) :: i2, j2 integer :: di, dj
i1 = i_coord (loc1) j1 = j_coord (loc1) i2 = i_coord (loc2) j2 = j_coord (loc2) di = i1 - i2 dj = j1 - j2 dist = sqrt (real ((di * di) + (dj * dj), kind = flk)) end function distance
function path_length (path) result (len) type(path_vector), intent(in) :: path real(kind = flk) :: len
integer(kind = pik) :: i
len = distance (path%elem(0_pik), path%elem(99_pik)) do i = 0_pik, 98_pik len = len + distance (path%elem(i), path%elem(i + 1_pik)) end do end function path_length
subroutine find_neighbors (loc, neighbors, num_neighbors) integer(kind = clk), intent(in) :: loc integer(kind = clk), intent(out) :: neighbors(1:8) integer, intent(out) :: num_neighbors
integer(kind = cok) :: i, j integer(kind = clk) :: c1, c2, c3, c4, c5, c6, c7, c8
c1 = 0_clk c2 = 0_clk c3 = 0_clk c4 = 0_clk c5 = 0_clk c6 = 0_clk c7 = 0_clk c8 = 0_clk
i = i_coord (loc) j = j_coord (loc)
if (i < 9_cok) then c1 = location (i + 1_cok, j) if (j < 9_cok) then c2 = location (i + 1_cok, j + 1_cok) end if if (0_cok < j) then c3 = location (i + 1_cok, j - 1_cok) end if end if if (0_cok < i) then c4 = location (i - 1_cok, j) if (j < 9_cok) then c5 = location (i - 1_cok, j + 1_cok) end if if (0_cok < j) then c6 = location (i - 1_cok, j - 1_cok) end if end if if (j < 9_cok) then c7 = location (i, j + 1_cok) end if if (0_cok < j) then c8 = location (i, j - 1_cok) end if
num_neighbors = 0 call add_neighbor (c1) call add_neighbor (c2) call add_neighbor (c3) call add_neighbor (c4) call add_neighbor (c5) call add_neighbor (c6) call add_neighbor (c7) call add_neighbor (c8)
contains
subroutine add_neighbor (neighbor) integer(kind = clk), intent(in) :: neighbor
if (neighbor /= 0_clk) then num_neighbors = num_neighbors + 1 neighbors(num_neighbors) = neighbor end if end subroutine add_neighbor
end subroutine find_neighbors
function make_neighbor_path (path) result (neighbor_path) type(path_vector), intent(in) :: path type(path_vector) :: neighbor_path
integer(kind = clk) :: u, v integer(kind = clk) :: neighbors(1:8) integer :: num_neighbors integer(kind = pik) :: j, iu, iv
neighbor_path = path
u = int (random_integer (1, 99), kind = clk) call find_neighbors (u, neighbors, num_neighbors) v = neighbors (random_integer (1, num_neighbors))
j = 0_pik iu = 0_pik iv = 0_pik do while (iu == 0_pik .or. iv == 0_pik) if (neighbor_path%elem(j + 1) == u) then iu = j + 1 else if (neighbor_path%elem(j + 1) == v) then iv = j + 1 end if j = j + 1 end do neighbor_path%elem(iu) = v neighbor_path%elem(iv) = u end function make_neighbor_path
function temperature (kT, kmax, k) result (temp) real(kind = flk), intent(in) :: kT integer, intent(in) :: kmax, k real(kind = flk) :: temp
real(kind = flk) :: kf, kmaxf
kf = real (k, kind = flk) kmaxf = real (kmax, kind = flk) temp = kT * (1.0_flk - (kf / kmaxf)) end function temperature
function probability (delta_E, T) result (prob) real(kind = flk), intent(in) :: delta_E, T real(kind = flk) :: prob
if (T == 0.0_flk) then prob = 0.0_flk else prob = exp (-(delta_E / T)) end if end function probability
subroutine show (k, T, E) integer, intent(in) :: k real(kind = flk), intent(in) :: T, E
write (*, 10) k, T, E
10 format (1X, I7, 1X, F7.1, 1X, F10.2)
end subroutine show
subroutine display_path (path) type(path_vector), intent(in) :: path
integer(kind = pik) :: i
999 format () 100 format (' ->') 110 format (' ') 120 format (I2)
do i = 0_pik, 99_pik write (*, 120, advance = 'no') path%elem(i) write (*, 100, advance = 'no') if (mod (i, 8_pik) == 7_pik) then write (*, 999, advance = 'yes') else write (*, 110, advance = 'no') end if end do write (*, 120, advance = 'no') path%elem(0_pik) end subroutine display_path
subroutine simulate_annealing (kT, kmax, initial_path, final_path) real(kind = flk), intent(in) :: kT integer, intent(in) :: kmax type(path_vector), intent(in) :: initial_path type(path_vector), intent(inout) :: final_path
integer :: kshow integer :: k real(kind = flk) :: E, E_trial, T type(path_vector) :: path, trial real(kind = flk) :: randnum
kshow = kmax / 10
path = initial_path E = path_length (path) do k = 0, kmax T = temperature (kT, kmax, k) if (mod (k, kshow) == 0) call show (k, T, E) trial = make_neighbor_path (path) E_trial = path_length (trial) if (E_trial <= E) then path = trial E = E_trial else call random_number (randnum) if (randnum <= probability (E_trial - E, T)) then path = trial E = E_trial end if end if end do final_path = path end subroutine simulate_annealing
end module simanneal_support
program simanneal
use, non_intrinsic :: simanneal_support implicit none
real(kind = flk), parameter :: kT = 1.0_flk integer, parameter :: kmax = 1000000
type(path_vector) :: initial_path type(path_vector) :: final_path
call random_seed
call randomize_path_vector (initial_path)
10 format () 20 format (' kT: ', F0.2) 30 format (' kmax: ', I0) 40 format (' k T E(s)') 50 format (' --------------------------') 60 format ('Final E(s): ', F0.2)
write (*, 10) write (*, 20) kT write (*, 30) kmax write (*, 10) write (*, 40) write (*, 50) call simulate_annealing (kT, kmax, initial_path, final_path) write (*, 10) call display_path (final_path) write (*, 10) write (*, 10) write (*, 60) path_length (final_path) write (*, 10)
end program simanneal</lang>
Go
<lang go>package main
import (
"fmt" "math" "math/rand" "time"
)
var (
dists = calcDists() dirs = [8]int{1, -1, 10, -10, 9, 11, -11, -9} // all 8 neighbors
)
// distances func calcDists() []float64 {
dists := make([]float64, 10000) for i := 0; i < 10000; i++ { ab, cd := math.Floor(float64(i)/100), float64(i%100) a, b := math.Floor(ab/10), float64(int(ab)%10) c, d := math.Floor(cd/10), float64(int(cd)%10) dists[i] = math.Hypot(a-c, b-d) } return dists
}
// index into lookup table of float64s func dist(ci, cj int) float64 {
return dists[cj*100+ci]
}
// energy at s, to be minimized func Es(path []int) float64 {
d := 0.0 for i := 0; i < len(path)-1; i++ { d += dist(path[i], path[i+1]) } return d
}
// temperature function, decreases to 0 func T(k, kmax, kT int) float64 {
return (1 - float64(k)/float64(kmax)) * float64(kT)
}
// variation of E, from state s to state s_next func dE(s []int, u, v int) float64 {
su, sv := s[u], s[v] // old a, b, c, d := dist(s[u-1], su), dist(s[u+1], su), dist(s[v-1], sv), dist(s[v+1], sv) // new na, nb, nc, nd := dist(s[u-1], sv), dist(s[u+1], sv), dist(s[v-1], su), dist(s[v+1], su) if v == u+1 { return (na + nd) - (a + d) } else if u == v+1 { return (nc + nb) - (c + b) } else { return (na + nb + nc + nd) - (a + b + c + d) }
}
// probability to move from s to s_next func P(deltaE float64, k, kmax, kT int) float64 {
return math.Exp(-deltaE / T(k, kmax, kT))
}
func sa(kmax, kT int) {
rand.Seed(time.Now().UnixNano()) temp := make([]int, 99) for i := 0; i < 99; i++ { temp[i] = i + 1 } rand.Shuffle(len(temp), func(i, j int) { temp[i], temp[j] = temp[j], temp[i] }) s := make([]int, 101) // all 0 by default copy(s[1:], temp) // random path from 0 to 0 fmt.Println("kT =", kT) fmt.Printf("E(s0) %f\n\n", Es(s)) // random starter Emin := Es(s) // E0 for k := 0; k <= kmax; k++ { if k%(kmax/10) == 0 { fmt.Printf("k:%10d T: %8.4f Es: %8.4f\n", k, T(k, kmax, kT), Es(s)) } u := 1 + rand.Intn(99) // city index 1 to 99 cv := s[u] + dirs[rand.Intn(8)] // city number if cv <= 0 || cv >= 100 { // bogus city continue } if dist(s[u], cv) > 5 { // check true neighbor (eg 0 9) continue } v := s[cv] // city index deltae := dE(s, u, v) if deltae < 0 || // always move if negative P(deltae, k, kmax, kT) >= rand.Float64() { s[u], s[v] = s[v], s[u] Emin += deltae } } fmt.Printf("\nE(s_final) %f\n", Emin) fmt.Println("Path:") // output final state for i := 0; i < len(s); i++ { if i > 0 && i%10 == 0 { fmt.Println() } fmt.Printf("%4d", s[i]) } fmt.Println()
}
func main() {
sa(1e6, 1)
}</lang>
- Output:
Sample run:
kT = 1 E(s0) 520.932463 k: 0 T: 1.0000 Es: 520.9325 k: 100000 T: 0.9000 Es: 185.1279 k: 200000 T: 0.8000 Es: 167.7657 k: 300000 T: 0.7000 Es: 158.6923 k: 400000 T: 0.6000 Es: 151.6564 k: 500000 T: 0.5000 Es: 139.9185 k: 600000 T: 0.4000 Es: 132.9964 k: 700000 T: 0.3000 Es: 121.8962 k: 800000 T: 0.2000 Es: 120.0445 k: 900000 T: 0.1000 Es: 116.8476 k: 1000000 T: 0.0000 Es: 116.5565 E(s_final) 116.556509 Path: 0 11 21 31 41 51 52 61 62 72 82 73 74 64 44 45 55 54 63 53 42 32 43 33 35 34 24 23 22 13 12 2 3 4 14 25 26 7 6 16 15 5 17 27 36 46 56 66 65 75 77 78 68 69 59 49 39 38 37 28 29 19 9 8 18 47 48 58 57 67 76 86 85 95 96 97 87 88 79 89 99 98 84 94 83 93 92 91 90 80 81 71 70 60 50 40 30 20 10 1 0
J
Implementation:
<lang J>dist=: +/&.:*:@:-"1/~10 10#:i.100
satsp=:4 :0
kT=. 1 pathcost=. [: +/ 2 {&y@<\ 0 , ] , 0: neighbors=. 0 (0}"1) y e. 1 2{/:~~.,y s=. (?~#y)-.0 d=. pathcost s step=. x%10 for_k. i.x+1 do. T=. kT*1-k%x u=. ({~ ?@#)s v=. ({~ ?@#)I.u{neighbors sk=. (?0 do. s=.sk d=.dk end. if. 0=step|k do. echo k,T,d end. end. 0,s,0
)</lang>
Notes:
E(s_final) gets displayed on the kmax progress line.
We do not do anything special for negative deltaE because the exponential will be greater than 1 for that case and that will always be greater than our random number from the range 0..1.
Also, while we leave connection distances (and, thus, number of cities) as a parameter, some other aspects of this problem made more sense when included in the implementation:
We leave city 0 out of our data structure, since it can't appear in the middle of our path. But we bring it back in when computing path distance.
Neighbors are any city which have one of the two closest non-zero distances from the current city (and specifically excluding city 0, since that is anchored as our start and end city).
Sample run:
<lang J> 1e6 satsp dist 0 1 538.409 100000 0.9 174.525 200000 0.8 165.541 300000 0.7 173.348 400000 0.6 168.188 500000 0.5 134.983 600000 0.4 121.585 700000 0.3 111.443 800000 0.2 101.657 900000 0.1 101.657 1e6 0 101.657 0 1 2 3 4 13 23 24 34 44 43 33 32 31 41 42 52 51 61 62 53 54 64 65 55 45 35 25 15 14 5 6 7 17 16 26 27 37 36 46 47 48 38 28 18 8 9 19 29 39 49 59 69 79 78 68 58 57 56 66 67 77 76 75 85 86 87 88 89 99 98 97 96 95 94 84 74 73 63 72 82 83 93 92 91 90 80 81 71 70 60 50 40 30 20 21 22 12 11 10 0</lang>
Julia
Module: <lang julia>module TravelingSalesman
using Random, Printf
- Eₛ: length(path)
Eₛ(distances, path) = sum(distances[ci, cj] for (ci, cj) in zip(path, Iterators.drop(path, 1)))
- T: temperature
T(k, kmax, kT) = kT * (1 - k / kmax)
- Alternative temperature:
- T(k, kmax, kT) = kT * (1 - sin(π / 2 * k / kmax))
- ΔE = Eₛ_new - Eₛ_old > 0
- Prob. to move if ΔE > 0, → 0 when T → 0 (fronzen state)
P(ΔE, k, kmax, kT) = exp(-ΔE / T(k, kmax, kT))
- ∆E from path ( .. a u b .. c v d ..) to (.. a v b ... c u d ..)
- ∆E before swapping (u,v)
- Quicker than Eₛ(s_next) - Eₛ(path)
function dE(distances, path, u, v)
a = distances[path[u - 1], path[u]] b = distances[path[u + 1], path[u]] c = distances[path[v - 1], path[v]] d = distances[path[v + 1], path[v]]
na = distances[path[u - 1], path[v]] nb = distances[path[u + 1], path[v]] nc = distances[path[v - 1], path[u]] nd = distances[path[v + 1], path[u]]
if v == u + 1 return (na + nd) - (a + d) elseif u == v + 1 return (nc + nb) - (c + b) else return (na + nb + nc + nd) - (a + b + c + d) end
end
const dirs = [1, -1, 10, -10, 9, 11, -11, -9]
function _prettypath(path)
r = IOBuffer() for g in Iterators.partition(path, 10) println(r, join(lpad.(g, 3), ", ")) end return String(take!(r))
end
function findpath(distances, kmax, kT)
n = size(distances, 1) path = vcat(1, shuffle(2:n), 1) Emin = Eₛ(distances, path) @printf("\n# Entropy(s₀) = %10.2f\n", Emin) println("# Random path: \n", _prettypath(path))
for k in Base.OneTo(kmax) if iszero(k % (kmax ÷ 10)) @printf("k: %10d | T: %8.4f | Eₛ: %8.4f\n", k, T(k, kmax, kT), Eₛ(distances, path)) end u = rand(2:n) v = path[u] + rand(dirs) v ∈ 2:n || continue
δE = dE(distances, path, u, v) if δE < 0 || P(δE, k, kmax, kT) ≥ rand() path[u], path[v] = path[v], path[u] Emin += δE end end
@printf("k: %10d | T: %8.4f | Eₛ: %8.4f\n", kmax, T(kmax, kmax, kT), Eₛ(distances, path)) println("\n# Found path:\n", _prettypath(path)) return path
end
end # module TravelingSalesman</lang>
Main: <lang julia>distance(a, b) = sqrt(sum((a .- b) .^ 2)) const _citydist = collect(distance((ci % 10, ci ÷ 10), (cj % 10, cj ÷ 10)) for ci in 1:100, cj in 1:100)
TravelingSalesman.findpath(_citydist, 1_000_000, 1)</lang>
- Output:
# Entropy(s₀) = 521.86 # Random path: 1, 2, 11, 80, 78, 73, 68, 19, 43, 69 86, 79, 66, 67, 77, 96, 26, 62, 60, 98 71, 3, 59, 37, 18, 40, 34, 92, 97, 6 84, 94, 29, 63, 36, 50, 87, 45, 83, 90 76, 28, 15, 38, 91, 58, 47, 44, 85, 17 25, 33, 31, 99, 27, 74, 53, 95, 16, 13 42, 88, 8, 4, 7, 64, 54, 9, 14, 41 5, 81, 65, 23, 75, 100, 89, 51, 20, 48 82, 12, 21, 55, 24, 70, 49, 10, 35, 72 52, 22, 61, 32, 46, 57, 30, 93, 39, 56 1 k: 100000 | T: 0.9000 | Eₛ: 184.4448 k: 200000 | T: 0.8000 | Eₛ: 175.3662 k: 300000 | T: 0.7000 | Eₛ: 169.0505 k: 400000 | T: 0.6000 | Eₛ: 160.8328 k: 500000 | T: 0.5000 | Eₛ: 147.1973 k: 600000 | T: 0.4000 | Eₛ: 132.9186 k: 700000 | T: 0.3000 | Eₛ: 126.9931 k: 800000 | T: 0.2000 | Eₛ: 122.0656 k: 900000 | T: 0.1000 | Eₛ: 119.7924 k: 1000000 | T: 0.0000 | Eₛ: 119.7924 k: 1000000 | T: 0.0000 | Eₛ: 119.7924 # Found path: 1, 2, 12, 13, 3, 4, 6, 7, 8, 9 19, 18, 17, 5, 14, 15, 16, 27, 28, 29 39, 38, 26, 25, 24, 23, 22, 10, 21, 20 30, 31, 32, 33, 34, 35, 36, 37, 49, 48 47, 46, 45, 44, 43, 42, 41, 40, 50, 51 52, 53, 54, 55, 56, 57, 58, 59, 69, 68 67, 65, 64, 63, 62, 61, 71, 60, 70, 80 81, 82, 72, 73, 74, 66, 78, 79, 89, 99 98, 97, 96, 95, 94, 85, 86, 87, 88, 77 76, 75, 84, 83, 93, 92, 91, 100, 90, 11 1
Nim
<lang Nim>import math, random, sequtils, strformat
const
kT = 1 kMax = 1_000_000
proc randomNeighbor(x: int): int =
case x of 0: sample([1, 10, 11]) of 9: sample([8, 18, 19]) of 90: sample([80, 81, 91]) of 99: sample([88, 89, 98]) elif x > 0 and x < 9: # top ceiling sample [x-1, x+1, x+9, x+10, x+11] elif x > 90 and x < 99: # bottom floor sample [x-11, x-10, x-9, x-1, x+1] elif x mod 10 == 0: # left wall sample([x-10, x-9, x+1, x+10, x+11]) elif (x+1) mod 10 == 0: # right wall sample([x-11, x-10, x-1, x+9, x+10]) else: # center sample([x-11, x-10, x-9, x-1, x+1, x+9, x+10, x+11])
proc neighbor(s: seq[int]): seq[int] =
result = s var city = sample(s) var cityNeighbor = city.randomNeighbor while cityNeighbor == 0 or city == 0: city = sample(s) cityNeighbor = city.randomNeighbor result[s.find city].swap result[s.find cityNeighbor]
func distNeighbor(a, b: int): float =
template divmod(a: int): (int, int) = (a div 10, a mod 10) let (diva, moda) = a.divmod (divb, modb) = b.divmod hypot((diva-divb).float, (moda-modb).float)
func temperature(k, kmax: float): float =
kT * (1 - (k / kmax))
func pdelta(eDelta, temp: float): float =
if eDelta < 0: 1.0 else: exp(-eDelta / temp)
func energy(path: seq[int]): float =
var sum = 0.distNeighbor path[0] for i in 1 ..< path.len: sum += path[i-1].distNeighbor(path[i]) sum + path[^1].distNeighbor 0
proc main =
randomize() var s = block: var x = toSeq(0..99) template shuffler: int = rand(1 .. x.high) for i in 1 .. x.high: x[i].swap x[shuffler()] x echo fmt"E(s0): {energy s:6.4f}" for k in 0 .. kMax: var temp = temperature(float k, float kMax) lastenergy = energy s newneighbor = s.neighbor newenergy = newneighbor.energy if k mod (kMax div 10) == 0: echo fmt"k: {k:7} T: {temp:6.2f} Es: {lastenergy:6.4f}" var deltaEnergy = newenergy - lastenergy if pDelta(deltaEnergy, temp) >= rand(1.0): s = newneighbor
s.add 0 echo fmt"E(sFinal): {energy s:6.4f}" echo fmt"path: {s}"
main()</lang>
Compile and run:
nim c -r -d:release --opt:speed travel_sa.nim
- Output:
Sample run:
E(s0): 505.1591 k: 0 T: 1.00 Es: 505.1591 k: 100000 T: 0.90 Es: 196.5216 k: 200000 T: 0.80 Es: 165.6735 k: 300000 T: 0.70 Es: 159.3411 k: 400000 T: 0.60 Es: 144.8330 k: 500000 T: 0.50 Es: 131.7888 k: 600000 T: 0.40 Es: 127.6914 k: 700000 T: 0.30 Es: 113.9280 k: 800000 T: 0.20 Es: 104.7279 k: 900000 T: 0.10 Es: 103.3137 k: 1000000 T: 0.00 Es: 103.3137 E(sFinal): 103.3137 path: @[0, 10, 11, 22, 21, 20, 30, 31, 41, 40, 50, 51, 61, 60, 70, 71, 81, 80, 90, 91, 92, 93, 82, 83, 73, 72, 62, 63, 53, 52, 42, 32, 33, 23, 13, 14, 24, 34, 35, 25, 15, 16, 26, 36, 47, 48, 38, 39, 49, 59, 58, 57, 68, 69, 79, 89, 99, 98, 97, 96, 95, 94, 84, 74, 75, 85, 86, 87, 88, 78, 77, 67, 76, 66, 65, 64, 54, 43, 44, 45, 55, 56, 46, 37, 27, 28, 29, 19, 9, 8, 18, 17, 7, 6, 5, 4, 3, 2, 12, 1, 0]
Perl
<lang perl>use utf8; use strict; use warnings; use List::Util qw(shuffle sum);
- simulation setup
my $cities = 100; # number of cities my $kmax = 1e6; # iterations to run my $kT = 1; # initial 'temperature'
die 'City count must be a perfect square.' if sqrt($cities) != int sqrt($cities);
- locations of (up to) 8 neighbors, with grid size derived from number of cities
my $gs = sqrt $cities; my @neighbors = (1, -1, $gs, -$gs, $gs-1, $gs+1, -($gs+1), -($gs-1));
- matrix of distances between cities
my @D; for my $j (0 .. $cities**2 - 1) {
my ($ab, $cd) = (int($j/$cities), int($j%$cities)); my ($a, $b, $c, $d) = (int($ab/$gs), int($ab%$gs), int($cd/$gs), int($cd%$gs)); $D[$ab][$cd] = sqrt(($a - $c)**2 + ($b - $d)**2);
}
- temperature function, decreases to 0
sub T {
my($k, $kmax, $kT) = @_; (1 - $k/$kmax) * $kT
}
- probability to move from s to s_next
sub P {
my($ΔE, $k, $kmax, $kT) = @_; exp -$ΔE / T($k, $kmax, $kT)
}
- energy at s, to be minimized
sub Es {
my(@path) = @_; sum map { $D[ $path[$_] ] [ $path[$_+1] ] } 0 .. @path-2
}
- variation of E, from state s to state s_next
sub delta_E {
my($u, $v, @s) = @_; my ($a, $b, $c, $d) = ($D[$s[$u-1]][$s[$u]], $D[$s[$u+1]][$s[$u]], $D[$s[$v-1]][$s[$v]], $D[$s[$v+1]][$s[$v]]); my ($na, $nb, $nc, $nd) = ($D[$s[$u-1]][$s[$v]], $D[$s[$u+1]][$s[$v]], $D[$s[$v-1]][$s[$u]], $D[$s[$v+1]][$s[$u]]); if ($v == $u+1) { return ($na + $nd) - ($a + $d) } elsif ($u == $v+1) { return ($nc + $nb) - ($c + $b) } else { return ($na + $nb + $nc + $nd) - ($a + $b + $c + $d) }
}
- E(s0), initial state
my @s = 0; map { push @s, $_ } shuffle 1..$cities-1; push @s, 0; my $E_min_global = my $E_min = Es(@s);
for my $k (0 .. $kmax-1) {
printf "k:%8u T:%4.1f Es: %3.1f\n" , $k, T($k, $kmax, $kT), Es(@s) if $k % ($kmax/10) == 0;
# valid candidate cities (exist, adjacent) my $u = 1 + int rand $cities-1; my $cv = $neighbors[int rand 8] + $s[$u]; next if $cv <= 0 or $cv >= $cities or $D[$s[$u]][$cv] > sqrt(2);
my $v = $s[$cv]; my $ΔE = delta_E($u, $v, @s); if ($ΔE < 0 or P($ΔE, $k, $kmax, $kT) >= rand) { # always move if negative ($s[$u], $s[$v]) = ($s[$v], $s[$u]); $E_min += $ΔE; $E_min_global = $E_min if $E_min < $E_min_global; }
}
printf "\nE(s_final): %.1f\n", $E_min_global; for my $l (0..4) {
printf "@{['%4d' x 20]}\n", @s[$l*20 .. ($l+1)*20 - 1];
} printf " 0\n";</lang>
- Output:
k: 0 T: 1.0 Es: 519.2 k: 100000 T: 0.9 Es: 188.2 k: 200000 T: 0.8 Es: 178.5 k: 300000 T: 0.7 Es: 162.3 k: 400000 T: 0.6 Es: 157.0 k: 500000 T: 0.5 Es: 148.9 k: 600000 T: 0.4 Es: 128.7 k: 700000 T: 0.3 Es: 129.5 k: 800000 T: 0.2 Es: 119.8 k: 900000 T: 0.1 Es: 119.5 E(s_final): 119.1 0 12 23 24 35 36 26 27 16 7 8 9 19 29 28 18 17 6 14 13 22 32 33 34 25 15 5 4 3 2 1 11 20 21 31 30 40 51 50 60 61 62 53 43 44 54 56 57 48 49 39 38 37 46 45 55 65 64 63 74 84 83 82 81 80 90 91 92 93 94 95 85 66 47 58 59 69 89 88 87 77 67 68 78 79 99 98 97 96 86 76 75 73 72 70 71 52 42 41 10 0
Phix
with javascript_semantics function hypot(atom a,b) return sqrt(a*a+b*b) end function function calc_dists() sequence dists = repeat(0,10000) for abcd=1 to 10000 do integer {ab,cd} = {floor(abcd/100),mod(abcd,100)}, {a,b,c,d} = {floor(ab/10),mod(ab,10), floor(cd/10),mod(cd,10)} dists[abcd] = hypot(a-c,b-d) end for return dists end function constant dists = calc_dists() function dist(integer ci,cj) return dists[cj*100+ci] end function function Es(sequence path) atom d = 0 for i=1 to length(path)-1 do d += dist(path[i],path[i+1]) end for return d end function -- temperature() function function T(integer k, kmax, kT) return (1-k/kmax)*kT end function -- deltaE = Es_new - Es_old > 0 -- probability to move if deltaE > 0, -->0 when T --> 0 (frozen state) function P(atom deltaE, integer k, kmax, kT) return exp(-deltaE/T(k,kmax,kT)) end function -- deltaE from path ( .. a u b .. c v d ..) to (.. a v b ... c u d ..) function dE(sequence s, integer u,v) -- (note that u,v are 0-based, but 1..99 here) -- integer sum1 = s[u-1], su = s[u], sup1 = s[u+1], -- svm1 = s[v-1], sv = s[v], svp1 = s[v+1] integer sum1 = s[u], su = s[u+1], sup1 = s[u+2], svm1 = s[v], sv = s[v+1], svp1 = s[v+2] -- old atom {a,b,c,d}:={dist(sum1,su), dist(su,sup1), dist(svm1,sv), dist(sv,svp1)}, -- new {na,nb,nc,nd}:={dist(sum1,sv), dist(sv,sup1), dist(svm1,su), dist(su,svp1)} return iff(v==u+1?(na+nd)-(a+d): iff(u==v+1?(nc+nb)-(c+b): (na+nb+nc+nd)-(a+b+c+d))) end function -- all 8 neighbours constant dirs = {1, -1, 10, -10, 9, 11, -11, -9} procedure sa(integer kmax, kT=10) sequence s = 0&shuffle(tagset(99))&0 atom Emin:=Es(s) -- E0 printf(1,"E(s0) %f\n",Emin) -- random starter for k=0 to kmax do if mod(k,kmax/10)=0 then printf(1,"k:%,10d T: %8.4f Es: %8.4f\n",{k,T(k,kmax,kT),Es(s)}) if k=kmax then exit end if -- avoid exp(x,-inf) end if integer u = rand(99), -- city index 1 99 cv = s[u+1]+dirs[rand(8)] -- city number if cv>0 and cv<100 -- not bogus city and dist(s[u+1],cv)<5 then -- and true neighbour integer v = s[cv+1] -- city index atom deltae := dE(s,u,v); if deltae<0 -- always move if negative or P(deltae,k,kmax,kT)>=rnd() then {s[u+1],s[v+1]} = {s[v+1],s[u+1]} Emin += deltae end if end if end for printf(1,"E(s_final) %f\n",Emin) printf(1,"Path:\n") pp(s,{pp_IntFmt,"%2d",pp_IntCh,false}) end procedure sa(1_000_000,1)
- Output:
E(s0) 515.164811 k: 0 T: 1.0000 Es: 515.1648 k: 100,000 T: 0.9000 Es: 189.3123 k: 200,000 T: 0.8000 Es: 198.7498 k: 300,000 T: 0.7000 Es: 158.2189 k: 400,000 T: 0.6000 Es: 165.4813 k: 500,000 T: 0.5000 Es: 156.3467 k: 600,000 T: 0.4000 Es: 142.7928 k: 700,000 T: 0.3000 Es: 128.0352 k: 800,000 T: 0.2000 Es: 121.7794 k: 900,000 T: 0.1000 Es: 121.2328 k: 1,000,000 T: 0.0000 Es: 121.1291 E(s_final) 121.129115 Path: { 0,10,62,63,64,65,76,75,84,85,95,86,96,97,87,77,67,66,56,46,47,48,49,59,69, 79,89,99,98,88,78,68,58,57,37,38,27,26,36,35,45,55,54,53,52,43,33,23,22,32, 42,41,51,61,60,50,40,30,31,21,20,11,12, 2, 3, 4, 5, 6,17,18,28,39,29,19, 9, 8, 7,16,15,24,44,74,83,93,94,92,91,71,70,90,80,81,82,72,73,34,25,14,13, 1, 0}
Racket
<lang racket>
- lang racket
(require racket/fixnum)
(define current-dim (make-parameter 10)) (define current-dim-- (make-parameter 9)) (define current-dim² (make-parameter 100)) (define current-kT (make-parameter 1)) (define current-k-max (make-parameter 1000000)) (define current-monitor-frequency (make-parameter 100000)) (define current-monitor (make-parameter
(λ (s k T E) (when (zero? (modulo k (current-monitor-frequency))) (printf "T:~a E:~a~%" (~r T) E)))))
- Simulated Annealing Solver
(define (P ΔE T)
(if (negative? ΔE) 1 (exp (- (/ ΔE T)))))
(define (solve/SA s₀ next-s k-max temperature E monitor)
(for*/fold ((s s₀) (E_s (E s₀))) ((k (in-range k-max))) (define T (temperature k k-max)) (when monitor (monitor s k T E_s)) (let* ((s´ (next-s s k)) (E_s´ (E s´)) (ΔE (- E_s´ E_s))) (if (>= (P ΔE T) (random)) (values s´ E_s´) (values s E_s)))))
(define (temperature k k-max)
(* (current-kT) (- 1 (/ k k-max))))
- TSP Problem
(struct tsp (path indices Es ΣE) #:transparent)
(define (y/x i d) (quotient/remainder i d))
(define (dist a b (d (current-dim)))
(let-values (([ay ax] (y/x a d)) ([by bx] (y/x b d))) (sqrt (+ (sqr (- ay by)) (sqr (- ax bx))))))
(define (indices->tsp indices)
(define path (make-fxvector (current-dim²))) (for ((i indices) (n (current-dim²))) (fxvector-set! path i n)) (define Es (for/vector #:length (fxvector-length path) ((a (in-fxvector path)) (b (in-sequences (in-fxvector path 1) (in-value (fxvector-ref path 0))))) (dist a b))) (tsp path indices Es (for/sum ((E Es)) E)))
(define (dir->delta dir (dim (current-dim))) (case dir [(l) -1] [(r) +1] [(u) (- dim)] [(d) dim]))
(define (invalid-direction? x y d (mx (current-dim--)))
(match* (x y d) ((0 _ 'l) #t) (((== mx) _ 'r) #t) ((_ 0 'u) #t) ((_ (== mx) 'd) #t) ((_ _ _) #f)))
- extended to take k to reset numerical drift from the Δ calculation
(define (tsp:swap-one-neighbour t k)
(define dim (current-dim)) (define dim² (current-dim²)) (define candidate (random dim²)) (define-values [cy cx] (quotient/remainder candidate dim)) (define dir (vector-ref #(l r u d) (random 4))) (cond [(invalid-direction? cx cy dir) (tsp:swap-one-neighbour t k)] [else (define delta (dir->delta dir)) (define neighbour (+ candidate delta)) (define path´ (fxvector-copy (tsp-path t))) (define indices´ (fxvector-copy (tsp-indices t))) (define cand-idx (fxvector-ref (tsp-indices t) candidate)) (define ngbr-idx (fxvector-ref (tsp-indices t) neighbour)) (fxvector-set! path´ cand-idx neighbour) (fxvector-set! path´ ngbr-idx candidate) (fxvector-set! indices´ candidate ngbr-idx) (fxvector-set! indices´ neighbour cand-idx) (define Es (tsp-Es t)) (define Es´ (vector-copy Es))
(let* ((cand-idx++ (modulo (add1 cand-idx) dim²)) (cand-idx-- (modulo (sub1 cand-idx) dim²)) (ngbr-idx++ (modulo (add1 ngbr-idx) dim²)) (ngbr-idx-- (modulo (sub1 ngbr-idx) dim²))) (define Σold-E-around-nodes (+ (vector-ref Es cand-idx) (vector-ref Es cand-idx--) (vector-ref Es ngbr-idx) (vector-ref Es ngbr-idx--))) (define E´-at-cand (dist (fxvector-ref path´ cand-idx) (fxvector-ref path´ cand-idx++))) (define E´-pre-cand (dist (fxvector-ref path´ cand-idx) (fxvector-ref path´ cand-idx--))) (define E´-at-ngbr (dist (fxvector-ref path´ ngbr-idx) (fxvector-ref path´ ngbr-idx++))) (define E´-pre-ngbr (dist (fxvector-ref path´ ngbr-idx) (fxvector-ref path´ ngbr-idx--))) (vector-set! Es´ cand-idx E´-at-cand) (vector-set! Es´ cand-idx-- E´-pre-cand) (vector-set! Es´ ngbr-idx E´-at-ngbr) (vector-set! Es´ ngbr-idx-- E´-pre-ngbr)
(define ΔE (- (+ E´-at-cand E´-pre-cand E´-at-ngbr E´-pre-ngbr) Σold-E-around-nodes)) (tsp path´ indices´ Es´ (if (zero? (modulo k 1000)) (for/sum ((e Es´)) e) (+ (tsp-ΣE t) ΔE))))]))
(define (tsp:random-state)
(indices->tsp (for/fxvector ((i (shuffle (range (current-dim²))))) i)))
(define (Simulated-annealing)
(define-values (solution solution-E) (solve/SA (tsp:random-state) tsp:swap-one-neighbour (current-k-max) temperature tsp-ΣE (current-monitor))) (displayln (tsp-path solution)) (displayln solution-E))
(module+ main
(Simulated-annealing))</lang>
- Output:
T:1 E:552.4249706051347 T:0.9 E:204.89460292101052 T:0.8 E:178.6926191428981 T:0.7 E:157.77681824512447 T:0.6 E:145.91227208091533 T:0.5 E:127.16624235784029 T:0.4 E:119.56239369288322 T:0.3 E:111.92798007771523 T:0.2 E:102.24264068711928 T:0.1 E:101.65685424949237 #fx(67 68 78 88 98 99 89 79 69 59 49 48 38 39 29 19 9 8 7 17 18 28 27 37 36 26 25 15 16 6 5 4 3 12 13 14 24 34 44 54 53 43 33 23 22 32 31 21 11 2 1 0 10 20 30 40 41 42 52 62 61 51 50 60 70 71 81 80 90 91 92 93 94 84 83 82 72 73 63 64 74 75 65 55 45 35 46 47 58 57 56 66 76 86 85 95 96 97 87 77) 101.65685424949237
Raku
(formerly Perl 6)
<lang perl6># simulation setup my \cities = 100; # number of cities my \kmax = 1e6; # iterations to run my \kT = 1; # initial 'temperature'
die 'City count must be a perfect square.' if cities.sqrt != cities.sqrt.Int;
- locations of (up to) 8 neighbors, with grid size derived from number of cities
my \gs = cities.sqrt; my \neighbors = [1, -1, gs, -gs, gs-1, gs+1, -(gs+1), -(gs-1)];
- matrix of distances between cities
my \D = [;]; for 0 ..^ cities² -> \j {
my (\ab, \cd) = (j/cities, j%cities)».Int; my (\a, \b, \c, \d) = (ab/gs, ab%gs, cd/gs, cd%gs)».Int; D[ab;cd] = sqrt (a - c)² + (b - d)²
}
sub T(\k, \kmax, \kT) { (1 - k/kmax) × kT } # temperature function, decreases to 0 sub P(\ΔE, \k, \kmax, \kT) { exp( -ΔE / T(k, kmax, kT)) } # probability to move from s to s_next sub Es(\path) { sum (D[ path[$_]; path[$_+1] ] for 0 ..^ +path-1) } # energy at s, to be minimized
- variation of E, from state s to state s_next
sub delta-E(\s, \u, \v) {
my (\a, \b, \c, \d) = D[s[u-1];s[u]], D[s[u+1];s[u]], D[s[v-1];s[v]], D[s[v+1];s[v]]; my (\na, \nb, \nc, \nd) = D[s[u-1];s[v]], D[s[u+1];s[v]], D[s[v-1];s[u]], D[s[v+1];s[u]]; if v == u+1 { return (na + nd) - (a + d) } elsif u == v+1 { return (nc + nb) - (c + b) } else { return (na + nb + nc + nd) - (a + b + c + d) }
}
- E(s0), initial state
my \s = @ = flat 0, (1 ..^ cities).pick(*), 0; my \E-min-global = my \E-min = $ = Es(s);
for 0 ..^ kmax -> \k {
printf "k:%8u T:%4.1f Es: %3.1f\n" , k, T(k, kmax, kT), Es(s) if k % (kmax/10) == 0;
# valid candidate cities (exist, adjacent) my \cv = neighbors[(^8).roll] + s[ my \u = 1 + (^(cities-1)).roll ]; next if cv ≤ 0 or cv ≥ cities or D[s[u];cv] > sqrt(2);
my \v = s[cv]; my \ΔE = delta-E(s, u, v); if ΔE < 0 or P(ΔE, k, kmax, kT) ≥ rand { # always move if negative (s[u], s[v]) = (s[v], s[u]); E-min += ΔE; E-min-global = E-min if E-min < E-min-global; }
}
say "\nE(s_final): " ~ E-min-global.fmt('%.1f'); say "Path:\n" ~ s».fmt('%2d').rotor(20,:partial).join: "\n";</lang>
- Output:
k: 0 T: 1.0 Es: 522.0 k: 100000 T: 0.9 Es: 185.3 k: 200000 T: 0.8 Es: 166.1 k: 300000 T: 0.7 Es: 174.7 k: 400000 T: 0.6 Es: 146.9 k: 500000 T: 0.5 Es: 140.2 k: 600000 T: 0.4 Es: 127.5 k: 700000 T: 0.3 Es: 115.9 k: 800000 T: 0.2 Es: 111.9 k: 900000 T: 0.1 Es: 109.4 E(s_final): 109.4 Path: 0 10 20 30 40 50 60 84 85 86 96 97 87 88 98 99 89 79 78 77 67 68 69 59 58 57 56 66 76 95 94 93 92 91 90 80 70 81 82 83 73 72 71 62 63 64 74 75 65 55 54 53 52 61 51 41 31 21 22 32 42 43 44 45 46 35 34 24 23 33 25 15 16 26 36 47 37 27 17 18 28 38 48 49 39 29 19 9 8 7 6 5 4 14 13 12 11 2 3 1 0
Scheme
Note: Each line of the printed table gives E(s) before the next path is chosen. Thus the top line describes the initial state.
<lang scheme>(cond-expand
(r7rs) (chicken (import r7rs)))
(import (scheme base)) (import (scheme inexact)) (import (scheme write)) (import (only (srfi 1) delete)) (import (only (srfi 1) iota)) (import (srfi 27)) ; Random numbers.
- You can do without SRFI-144 by changing fl+ to +, etc.
(import (srfi 144)) ; Optimizations for flonums.
(cond-expand
(chicken (import (format))) (else))
(random-source-randomize! default-random-source)
(define (n->ij n)
(truncate/ n 10))
(define (ij->n i j)
(+ (* 10 i) j))
(define neighbor-offsets
'((0 . 1) (1 . 0) (1 . 1) (0 . -1) (-1 . 0) (-1 . -1) (1 . -1) (-1 . 1)))
(define (neighborhood n)
(let-values (((i j) (n->ij n))) (let recurs ((offsets neighbor-offsets)) (if (null? offsets) '() (let* ((offs (car offsets)) (i^ (+ i (car offs))) (j^ (+ j (cdr offs)))) (if (and (not (negative? i^)) (not (negative? j^)) (< i^ 10) (< j^ 10)) (cons (ij->n i^ j^) (recurs (cdr offsets))) (recurs (cdr offsets))))))))
(define (distance m n)
(let-values (((im jm) (n->ij m)) ((in jn) (n->ij n))) (flsqrt (inexact (+ (square (- im in)) (square (- jm jn)))))))
(define (shuffle! vec i n)
;; A Fisher-Yates shuffle of n elements of vec, starting at index i. (do ((j 0 (+ j 1))) ((= j n)) (let* ((k (+ i j (random-integer (- n j)))) (xi (vector-ref vec i)) (xk (vector-ref vec k))) (vector-set! vec i xk) (vector-set! vec k xi))))
(define (make-s0)
(let ((vec (list->vector (iota 100)))) (shuffle! vec 1 99) vec))
(define (swap-s-elements! vec u v)
(let loop ((j 1) (iu 0) (iv 0)) (cond ((positive? iu) (if (= (vector-ref vec j) v) (begin (vector-set! vec iu v) (vector-set! vec j u)) (loop (+ j 1) iu iv))) ((positive? iv) (if (= (vector-ref vec j) u) (begin (vector-set! vec j v) (vector-set! vec iv u)) (loop (+ j 1) iu iv))) ((= (vector-ref vec j) u) (loop (+ j 1) j 0)) ((= (vector-ref vec j) v) (loop (+ j 1) 0 j)) (else (loop (+ j 1) 0 0)))))
(define (update-s! vec)
(let* ((u (+ 1 (random-integer 99))) (neighbors (delete 0 (neighborhood u) =)) (n (length neighbors)) (v (list-ref neighbors (random-integer n)))) (swap-s-elements! vec u v)))
(define (s->s vec) ; s_k -> s_(k + 1)
(let ((vec^ (vector-copy vec))) (update-s! vec^) vec^))
(define (path-length vec) ; E(s)
(let loop ((plen (distance (vector-ref vec 0) (vector-ref vec 99))) (x (vector-ref vec 0)) (i 1)) (if (= i 100) plen (let ((y (vector-ref vec i))) (loop (fl+ plen (distance x y)) y (+ i 1))))))
(define (make-temperature-procedure kT kmax)
(let ((kT (inexact kT)) (kmax (inexact kmax))) (lambda (k) (fl* kT (fl- 1.0 (fl/ (inexact k) kmax))))))
(define (probability delta-E T)
(if (flnegative? delta-E) 1.0 (if (flzero? T) 0.0 (flexp (fl- (fl/ delta-E T))))))
(define fmt10 (string-append " k T E(s)~%"
" -----------------------------~%"))
(define fmt20 " ~7D ~3,1F ~12,5F~%")
(define (simulate-annealing kT kmax)
(let* ((temperature (make-temperature-procedure kT kmax)) (s0 (make-s0)) (E0 (path-length s0)) (kmax/10 (truncate-quotient kmax 10)) (show (lambda (k T E) (if (zero? (truncate-remainder k kmax/10)) (cond-expand (chicken (format #t fmt20 k T E)) (else (display k) (display " ") (display T) (display " ") (display E) (newline))))))) (cond-expand (chicken (format #t fmt10)) (else)) (let loop ((k 0) (s s0) (E E0)) (if (= k (+ 1 kmax)) s (let* ((T (temperature k)) (_ (show k T E)) (s^ (s->s s)) (E^ (path-length s^)) (delta-E (fl- E^ E)) (P (probability delta-E T))) (if (or (fl=? P 1.0) (fl<=? (random-real) P)) (loop (+ k 1) s^ E^) (loop (+ k 1) s E)))))))
(define (display-path vec)
(do ((i 0 (+ i 1))) ((= i 100)) (let ((x (vector-ref vec i))) (when (< x 10) (display " ")) (display x) (display " -> ") (when (= 7 (truncate-remainder i 8)) (newline)))) (let ((x (vector-ref vec 0))) (when (< x 10) (display " ")) (display x)))
(define kT 1) (define kmax 1000000)
(newline) (display " kT: ") (display kT) (newline) (display " kmax: ") (display kmax) (newline) (newline) (define s-final (simulate-annealing kT kmax)) (newline) (display "Final path:") (newline) (display-path s-final) (newline) (newline) (cond-expand
(chicken (format #t "Final E(s): ~,5F~%" (path-length s-final))) (else (display "Final E(s): ") (display (path-length s-final)) (newline)))
(newline)</lang>
- Output:
An example run:
$ csc -O5 -X r7rs -R r7rs sa.scm && ./sa kT: 1 kmax: 1000000 k T E(s) ----------------------------- 0 1.0 422.71361 100000 0.9 185.28073 200000 0.8 171.70817 300000 0.7 156.66104 400000 0.6 145.07621 500000 0.5 130.88759 600000 0.4 115.34219 700000 0.3 112.27113 800000 0.2 105.37820 900000 0.1 103.89949 1000000 0.0 103.89949 Final path: 0 -> 1 -> 2 -> 3 -> 4 -> 6 -> 7 -> 8 -> 9 -> 19 -> 29 -> 39 -> 38 -> 37 -> 47 -> 48 -> 49 -> 58 -> 59 -> 69 -> 79 -> 89 -> 99 -> 98 -> 97 -> 96 -> 95 -> 94 -> 84 -> 83 -> 93 -> 92 -> 82 -> 72 -> 62 -> 71 -> 81 -> 91 -> 90 -> 80 -> 70 -> 60 -> 61 -> 50 -> 40 -> 41 -> 51 -> 52 -> 63 -> 73 -> 74 -> 75 -> 85 -> 86 -> 76 -> 77 -> 87 -> 88 -> 78 -> 68 -> 67 -> 57 -> 56 -> 66 -> 65 -> 64 -> 55 -> 45 -> 46 -> 36 -> 35 -> 25 -> 26 -> 27 -> 28 -> 18 -> 17 -> 16 -> 15 -> 5 -> 14 -> 13 -> 23 -> 24 -> 34 -> 44 -> 54 -> 53 -> 43 -> 33 -> 42 -> 32 -> 31 -> 30 -> 20 -> 21 -> 22 -> 12 -> 11 -> 10 -> 0 Final E(s): 103.89949
A different E(s)
Here E(s) is the sum of squares of differences, so that E(s) is an integer. Also I use kT=1.5 and twice as large a kmax.
(I also demonstrate some of SRFI 143. Note that CHICKEN has type annotations as an alternative to using SRFI 143 and SRFI 144, but the SRFI extensions are more portable.)
<lang scheme>(cond-expand
(r7rs) (chicken (import r7rs)))
(import (scheme base)) (import (scheme inexact)) (import (scheme write)) (import (only (srfi 1) delete)) (import (only (srfi 1) iota)) (import (srfi 27)) ; Random numbers.
- The following import is CHICKEN-specific, but your Scheme likely
- has Common Lisp formatting somewhere.
(import (format)) ; Common Lisp formatting.
- You can do without SRFI-143 or SRFI-144 by changing fx+ or fl+ to
- +, etc.
(import (srfi 143)) ; Optimizations for fixnums. (import (srfi 144)) ; Optimizations for flonums.
(random-source-randomize! default-random-source)
(define (n->ij n)
(values (fxquotient n 10) (fxremainder n 10)))
(define (ij->n i j)
(fx+ (fx* 10 i) j))
(define neighbor-offsets
'((0 . 1) (1 . 0) (1 . 1) (0 . -1) (-1 . 0) (-1 . -1) (1 . -1) (-1 . 1)))
(define (neighborhood n)
(let-values (((i j) (n->ij n))) (let recurs ((offsets neighbor-offsets)) (if (null? offsets) '() (let* ((offs (car offsets)) (i^ (fx+ i (car offs))) (j^ (fx+ j (cdr offs)))) (if (and (not (fxnegative? i^)) (not (fxnegative? j^)) (fx<? i^ 10) (fx<? j^ 10)) (cons (ij->n i^ j^) (recurs (cdr offsets))) (recurs (cdr offsets))))))))
(define (distance**2 m n)
(let-values (((im jm) (n->ij m)) ((in jn) (n->ij n))) (fx+ (fxsquare (fx- im in)) (fxsquare (fx- jm jn)))))
(define (shuffle! vec i n)
;; A Fisher-Yates shuffle of n elements of vec, starting at index i. (do ((j 0 (+ j 1))) ((= j n)) (let* ((k (+ i j (random-integer (- n j)))) (xi (vector-ref vec i)) (xk (vector-ref vec k))) (vector-set! vec i xk) (vector-set! vec k xi))))
(define (make-s0)
(let ((vec (list->vector (iota 100)))) (shuffle! vec 1 99) vec))
(define (swap-s-elements! vec u v)
(let loop ((j 1) (iu 0) (iv 0)) (cond ((fxpositive? iu) (if (fx=? (vector-ref vec j) v) (begin (vector-set! vec iu v) (vector-set! vec j u)) (loop (fx+ j 1) iu iv))) ((fxpositive? iv) (if (fx=? (vector-ref vec j) u) (begin (vector-set! vec j v) (vector-set! vec iv u)) (loop (fx+ j 1) iu iv))) ((fx=? (vector-ref vec j) u) (loop (fx+ j 1) j 0)) ((fx=? (vector-ref vec j) v) (loop (fx+ j 1) 0 j)) (else (loop (fx+ j 1) 0 0)))))
(define (update-s! vec)
(let* ((u (fx+ 1 (random-integer 99))) (neighbors (delete 0 (neighborhood u) fx=?)) (n (length neighbors)) (v (list-ref neighbors (random-integer n)))) (swap-s-elements! vec u v)))
(define (s->s vec) ; s_k -> s_(k + 1)
(let ((vec^ (vector-copy vec))) (update-s! vec^) vec^))
(define (path-length vec)
(let loop ((plen (flsqrt (inexact (distance**2 (vector-ref vec 0) (vector-ref vec 99))))) (x (vector-ref vec 0)) (i 1)) (if (fx=? i 100) plen (let ((y (vector-ref vec i))) (loop (fl+ plen (flsqrt (inexact (distance**2 x y)))) y (fx+ i 1))))))
(define (E_s vec)
(let loop ((E (distance**2 (vector-ref vec 0) (vector-ref vec 99))) (x (vector-ref vec 0)) (i 1)) (if (fx=? i 100) E (let ((y (vector-ref vec i))) (loop (fx+ E (distance**2 x y)) y (fx+ i 1))))))
(define (make-temperature-procedure kT kmax)
(let ((kT (inexact kT)) (kmax (inexact kmax))) (lambda (k) (fl* kT (fl- 1.0 (fl/ (inexact k) kmax))))))
(define (probability delta-E T)
(if (fxnegative? delta-E) 1.0 (if (flzero? T) 0.0 (flexp (fl- (fl/ (inexact delta-E) T))))))
(define fmt10 (string-append
" k T E(s) path length~%" " ---------------------------------------~%"))
(define fmt20 " ~7D ~7,2F ~8D ~14,5F~%")
(define (simulate-annealing kT kmax)
(let* ((temperature (make-temperature-procedure kT kmax)) (s0 (make-s0)) (E0 (E_s s0)) (kmax/10 (fxquotient kmax 10)) (show (lambda (k T E s) (when (fxzero? (fxremainder k kmax/10)) (format #t fmt20 k T E (path-length s)))))) (format #t fmt10) (let loop ((k 0) (s s0) (E E0)) (if (fx=? k (fx+ 1 kmax)) s (let* ((T (temperature k)) (_ (show k T E s)) (s^ (s->s s)) (E^ (E_s s^)) (delta-E (fx- E^ E)) (P (probability delta-E T))) (if (or (fl=? P 1.0) (fl<=? (random-real) P)) (loop (fx+ k 1) s^ E^) (loop (fx+ k 1) s E)))))))
(define (display-path vec)
(do ((i 0 (+ i 1))) ((= i 100)) (let ((x (vector-ref vec i))) (when (< x 10) (display " ")) (display x) (display " -> ") (when (= 7 (truncate-remainder i 8)) (newline)))) (let ((x (vector-ref vec 0))) (when (< x 10) (display " ")) (display x)))
(define kT 1.5) (define kmax 2000000)
(newline) (display " kT: ") (display kT) (newline) (display " kmax: ") (display kmax) (newline) (newline) (define s-final (simulate-annealing kT kmax)) (newline) (display "Final path:") (newline) (display-path s-final) (newline) (newline) (format #t "Final E(s): ~,5F~%" (E_s s-final)) (format #t "Final path length: ~,5F~%" (path-length s-final)) (newline)</lang>
- Output:
$ csc -O5 -X r7rs -R r7rs sa2.scm && ./sa2 kT: 1.5 kmax: 2000000 k T E(s) path length --------------------------------------- 0 1.50 2298 384.59396 200000 1.35 180 127.94332 400000 1.20 168 124.60675 600000 1.05 148 118.07012 800000 0.90 130 111.94113 1000000 0.75 116 106.62742 1200000 0.60 114 105.55635 1400000 0.45 112 104.97056 1600000 0.30 104 101.65685 1800000 0.15 104 101.65685 2000000 0.00 104 101.65685 Final path: 0 -> 1 -> 2 -> 3 -> 4 -> 5 -> 15 -> 24 -> 34 -> 43 -> 44 -> 54 -> 53 -> 63 -> 64 -> 65 -> 66 -> 67 -> 77 -> 76 -> 75 -> 85 -> 84 -> 74 -> 73 -> 72 -> 62 -> 52 -> 42 -> 41 -> 31 -> 32 -> 33 -> 23 -> 13 -> 14 -> 25 -> 26 -> 27 -> 17 -> 16 -> 6 -> 7 -> 8 -> 9 -> 19 -> 18 -> 28 -> 29 -> 39 -> 49 -> 59 -> 58 -> 48 -> 38 -> 37 -> 47 -> 46 -> 36 -> 35 -> 45 -> 55 -> 56 -> 57 -> 68 -> 69 -> 79 -> 78 -> 88 -> 89 -> 99 -> 98 -> 97 -> 87 -> 86 -> 96 -> 95 -> 94 -> 93 -> 83 -> 82 -> 92 -> 91 -> 90 -> 80 -> 81 -> 71 -> 70 -> 60 -> 61 -> 51 -> 50 -> 40 -> 30 -> 20 -> 21 -> 22 -> 12 -> 11 -> 10 -> 0 Final E(s): 104.00000 Final path length: 101.65685
A second run shows E(s) temporarily increasing:
kT: 1.5 kmax: 2000000 k T E(s) path length --------------------------------------- 0 1.50 2132 368.24125 200000 1.35 142 115.58483 400000 1.20 146 116.75641 600000 1.05 148 117.40669 800000 0.90 124 109.27770 1000000 0.75 112 104.97056 1200000 0.60 124 109.45584 1400000 0.45 114 105.55635 1600000 0.30 108 103.31371 1800000 0.15 108 103.31371 2000000 0.00 108 103.31371 Final path: 0 -> 1 -> 11 -> 21 -> 31 -> 42 -> 52 -> 62 -> 72 -> 73 -> 63 -> 74 -> 64 -> 65 -> 55 -> 45 -> 54 -> 53 -> 43 -> 44 -> 34 -> 35 -> 36 -> 26 -> 25 -> 24 -> 23 -> 33 -> 32 -> 22 -> 12 -> 2 -> 3 -> 13 -> 14 -> 4 -> 5 -> 15 -> 16 -> 6 -> 7 -> 17 -> 27 -> 28 -> 18 -> 8 -> 9 -> 19 -> 29 -> 39 -> 38 -> 37 -> 47 -> 48 -> 49 -> 58 -> 59 -> 69 -> 68 -> 67 -> 57 -> 46 -> 56 -> 66 -> 76 -> 86 -> 87 -> 77 -> 78 -> 79 -> 89 -> 99 -> 88 -> 98 -> 97 -> 96 -> 95 -> 85 -> 75 -> 84 -> 94 -> 93 -> 83 -> 82 -> 92 -> 91 -> 90 -> 80 -> 81 -> 71 -> 70 -> 60 -> 61 -> 51 -> 50 -> 41 -> 40 -> 30 -> 20 -> 10 -> 0 Final E(s): 108.00000 Final path length: 103.31371
A third run shows E(s) temporarily increasing, and also achieves a path length less than 101:
kT: 1.5 kmax: 2000000 k T E(s) path length --------------------------------------- 0 1.50 2246 388.77550 200000 1.35 176 124.95651 400000 1.20 160 121.22854 600000 1.05 148 117.29304 800000 0.90 126 109.86348 1000000 0.75 118 107.45584 1200000 0.60 120 108.04163 1400000 0.45 108 103.31371 1600000 0.30 106 102.48528 1800000 0.15 102 100.82843 2000000 0.00 102 100.82843 Final path: 0 -> 1 -> 11 -> 12 -> 2 -> 3 -> 13 -> 14 -> 4 -> 5 -> 15 -> 16 -> 6 -> 7 -> 17 -> 27 -> 28 -> 18 -> 8 -> 9 -> 19 -> 29 -> 39 -> 38 -> 48 -> 49 -> 59 -> 69 -> 79 -> 78 -> 77 -> 67 -> 68 -> 58 -> 57 -> 47 -> 37 -> 36 -> 26 -> 25 -> 24 -> 23 -> 22 -> 21 -> 31 -> 32 -> 43 -> 44 -> 54 -> 53 -> 52 -> 62 -> 61 -> 51 -> 41 -> 42 -> 33 -> 34 -> 35 -> 45 -> 46 -> 56 -> 55 -> 65 -> 66 -> 76 -> 86 -> 87 -> 88 -> 89 -> 99 -> 98 -> 97 -> 96 -> 95 -> 94 -> 84 -> 85 -> 75 -> 74 -> 64 -> 63 -> 73 -> 83 -> 93 -> 92 -> 82 -> 72 -> 71 -> 81 -> 91 -> 90 -> 80 -> 70 -> 60 -> 50 -> 40 -> 30 -> 20 -> 10 -> 0 Final E(s): 102.00000 Final path length: 100.82843
Sidef
<lang ruby>module TravelingSalesman {
# Eₛ: length(path) func Eₛ(distances, path) { var total = 0 [path, path.slice(1)].zip {|ci,cj| total += distances[ci-1][cj-1] } total }
# T: temperature func T(k, kmax, kT) { kT * (1 - k/kmax) }
# ΔE = Eₛ_new - Eₛ_old > 0 # Prob. to move if ΔE > 0, → 0 when T → 0 (fronzen state) func P(ΔE, k, kmax, kT) { exp(-ΔE / T(k, kmax, kT)) }
# ∆E from path ( .. a u b .. c v d ..) to (.. a v b ... c u d ..) # ∆E before swapping (u,v) # Quicker than Eₛ(s_next) - Eₛ(path) func dE(distances, path, u, v) {
var a = distances[path[u-1]-1][path[u]-1] var b = distances[path[u+1]-1][path[u]-1] var c = distances[path[v-1]-1][path[v]-1] var d = distances[path[v+1]-1][path[v]-1]
var na = distances[path[u-1]-1][path[v]-1] var nb = distances[path[u+1]-1][path[v]-1] var nc = distances[path[v-1]-1][path[u]-1] var nd = distances[path[v+1]-1][path[u]-1]
if (v == u+1) { return ((na+nd) - (a+d)) }
if (u == v+1) { return ((nc+nb) - (c+b)) }
return ((na+nb+nc+nd) - (a+b+c+d)) }
const dirs = [1, -1, 10, -10, 9, 11, -11, -9]
func _prettypath(path) { path.slices(10).map { .map{ "%3s" % _ }.join(', ') }.join("\n") }
func findpath(distances, kmax, kT) {
const n = distances.len const R = 2..n
var path = [1, R.shuffle..., 1] var Emin = Eₛ(distances, path)
printf("# Entropy(s₀) = s%10.2f\n", Emin) printf("# Random path:\n%s\n\n", _prettypath(path))
for k in (1 .. kmax) {
if (k % (kmax//10) == 0) { printf("k: %10d | T: %8.4f | Eₛ: %8.4f\n", k, T(k, kmax, kT), Eₛ(distances, path)) }
var u = R.rand var v = (path[u-1] + dirs.rand) v ~~ R || next
var δE = dE(distances, path, u-1, v-1) if ((δE < 0) || (P(δE, k, kmax, kT) >= 1.rand)) { path.swap(u-1, v-1) Emin += δE } }
printf("k: %10d | T: %8.4f | Eₛ: %8.4f\n", kmax, T(kmax, kmax, kT), Eₛ(distances, path)) say ("\n# Found path:\n", _prettypath(path)) return path }
}
var citydist = {|ci|
{ |cj| var v1 = Vec(ci%10, ci//10) var v2 = Vec(cj%10, cj//10) v1.dist(v2) }.map(1..100)
}.map(1..100)
TravelingSalesman::findpath(citydist, 1e6, 1)</lang>
- Output:
# Entropy(s₀) = 520.29 # Random path: 1, 10, 79, 52, 24, 9, 58, 11, 42, 4 15, 87, 62, 88, 21, 91, 99, 84, 61, 14 5, 17, 33, 95, 74, 31, 40, 13, 37, 69 6, 22, 97, 45, 56, 63, 75, 83, 53, 41 3, 47, 89, 80, 78, 98, 46, 18, 25, 51 93, 16, 50, 30, 48, 8, 66, 68, 59, 73 49, 96, 36, 32, 100, 27, 76, 44, 64, 39 90, 82, 20, 12, 54, 86, 29, 81, 26, 72 60, 94, 35, 92, 43, 7, 85, 55, 28, 57 23, 34, 65, 71, 38, 2, 77, 70, 19, 67 1 k: 100000 | T: 0.9000 | Eₛ: 185.1809 k: 200000 | T: 0.8000 | Eₛ: 168.6262 k: 300000 | T: 0.7000 | Eₛ: 146.5948 k: 400000 | T: 0.6000 | Eₛ: 140.1441 k: 500000 | T: 0.5000 | Eₛ: 129.5132 k: 600000 | T: 0.4000 | Eₛ: 132.8942 k: 700000 | T: 0.3000 | Eₛ: 124.2865 k: 800000 | T: 0.2000 | Eₛ: 120.0859 k: 900000 | T: 0.1000 | Eₛ: 115.0771 k: 1000000 | T: 0.0000 | Eₛ: 114.9728 k: 1000000 | T: 0.0000 | Eₛ: 114.9728 # Found path: 1, 2, 13, 3, 4, 5, 6, 7, 8, 9 19, 29, 18, 28, 27, 17, 16, 26, 25, 15 14, 24, 23, 12, 11, 10, 20, 21, 30, 40 41, 31, 32, 44, 45, 46, 47, 48, 49, 39 38, 37, 36, 35, 34, 42, 51, 50, 60, 61 52, 53, 54, 55, 56, 57, 58, 59, 69, 68 77, 67, 66, 65, 64, 62, 72, 71, 70, 80 81, 82, 74, 75, 76, 87, 88, 78, 79, 89 99, 98, 97, 96, 86, 85, 83, 91, 90, 100 92, 93, 94, 95, 84, 73, 63, 43, 33, 22 1
Wren
<lang ecmascript>import "random" for Random import "/math" for Math import "/fmt" for Fmt
// distances var calcDists = Fn.new {
var dists = List.filled(10000, 0) for (i in 0..9999) { var ab = (i/100).floor var cd = i % 100 var a = (ab/10).floor var b = ab % 10 var c = (cd/10).floor var d = cd % 10 dists[i] = Math.hypot(a-c, b-d) } return dists
}
var dists = calcDists.call() var dirs = [1, -1, 10, -10, 9, 11, -11, -9] // all 8 neighbors var rand = Random.new()
// index into lookup table of Nums var dist = Fn.new { |ci, cj| dists[cj*100 + ci] }
// energy at s, to be minimized var Es = Fn.new { |path|
var d = 0 for (i in 0...path.count-1) d = d + dist.call(path[i], path[i+1]) return d
}
// temperature function, decreases to 0 var T = Fn.new { |k, kmax, kT| (1 - k / kmax) * kT }
// variation of E, from state s to state s_next var dE = Fn.new { |s, u, v|
var su = s[u] var sv = s[v] // old var a = dist.call(s[u-1], su) var b = dist.call(s[u+1], su) var c = dist.call(s[v-1], sv) var d = dist.call(s[v+1], sv) // new var na = dist.call(s[u-1], sv) var nb = dist.call(s[u+1], sv) var nc = dist.call(s[v-1], su) var nd = dist.call(s[v+1], su) if (v == u+1) return (na + nd) - (a + d) if (u == v+1) return (nc + nb) - (c + b) return (na + nb + nc + nd) - (a + b + c + d)
}
// probability to move from s to s_next var P = Fn.new { |deltaE, k, kmax, kT| (-deltaE / T.call(k, kmax, kT)).exp }
// Simulated annealing var sa = Fn.new { |kmax, kT|
var temp = List.filled(99, 0) for (i in 0..98) temp[i] = i + 1 rand.shuffle(temp) var s = List.filled(101, 0) for (i in 0..98) s[i+1] = temp[i] // random path from 0 to 0 System.print("kT = %(kT)") System.print("E(s0) %(Es.call(s))\n") // random starter var Emin = Es.call(s) // E0 for (k in 0..kmax) { if (k % (kmax/10).floor == 0) { Fmt.print("k:$10d T: $8.4f Es: $8.4f", k, T.call(k, kmax, kT), Es.call(s)) } var u = rand.int(1, 100) // city index 1 to 99 var cv = s[u] + dirs[rand.int(8)] // city number if (cv <= 0 || cv >= 100) { // bogus city continue } if (dist.call(s[u], cv) > 5) { // check true neighbor (eg 0 9) continue } var v = s[cv] // city index var deltae = dE.call(s, u, v) if (deltae < 0 || // always move if negative P.call(deltae, k, kmax, kT) >= rand.float()) { s.swap(u, v) Emin = Emin + deltae } } System.print("\nE(s_final) %(Emin)") System.print("Path:") // output final state for (i in 0...s.count) { if (i > 0 && i % 10 == 0) System.print() Fmt.write("$4d", s[i]) } System.print()
}
sa.call(1e6, 1)</lang>
- Output:
Sample run:
kT = 1 E(s0) 541.82779520458 k: 0 T: 1.0000 Es: 541.8278 k: 100000 T: 0.9000 Es: 187.1429 k: 200000 T: 0.8000 Es: 191.0983 k: 300000 T: 0.7000 Es: 171.7284 k: 400000 T: 0.6000 Es: 154.0549 k: 500000 T: 0.5000 Es: 147.0249 k: 600000 T: 0.4000 Es: 123.5822 k: 700000 T: 0.3000 Es: 121.5808 k: 800000 T: 0.2000 Es: 114.0930 k: 900000 T: 0.1000 Es: 112.6788 k: 1000000 T: 0.0000 Es: 112.6788 E(s_final) 112.67876668098 Path: 0 11 10 1 2 12 13 24 25 15 14 3 4 5 6 8 9 19 18 28 29 39 38 27 17 7 16 26 36 37 47 46 45 35 34 44 43 33 23 22 32 53 52 51 41 31 21 20 30 40 50 60 61 70 80 90 91 92 93 73 63 62 72 71 81 82 83 84 94 95 85 86 96 97 98 99 89 79 69 59 49 48 58 57 68 78 88 87 77 67 56 55 54 65 66 76 75 74 64 42 0
zkl
<lang zkl>var [const] _dists=(0d10_000).pump(List,fcn(abcd){ // two points (a,b) & (c,d), calc distance
ab,cd,a,b,c,d:=abcd/100, abcd%100, ab/10,ab%10, cd/10,cd%10; (a-c).toFloat().hypot(b-d)
}); fcn dist(ci,cj){ _dists[cj*100 + ci] } // index into lookup table of floats
fcn Es(path) // E(s) = length(path): E(a,b,c)--> dist(a,b) + dist(b,c)
{ d:=Ref(0.0); path.reduce('wrap(a,b){ d.apply('+,dist(a,b)); b }); d.value }
// temperature() function fcn T(k,kmax,kT){ (1.0 - k.toFloat()/kmax)*kT }
// deltaE = Es_new - Es_old > 0 // probability to move if deltaE > 0, -->0 when T --> 0 (frozen state) fcn P(deltaE,k,kmax,kT){ (-deltaE/T(k,kmax,kT)).exp() } //-->Float
// deltaE from path ( .. a u b .. c v d ..) to (.. a v b ... c u d ..) // deltaE before swapping (u,v) fcn dE(s,u,v){ su,sv:=s[u],s[v]; //-->Float
// old a,b,c,d:=dist(s[u-1],su), dist(s[u+1],su), dist(s[v-1],sv), dist(s[v+1],sv); // new na,nb,nc,nd:=dist(s[u-1],sv), dist(s[u+1],sv), dist(s[v-1],su), dist(s[v+1],su);
if (v==u+1) (na+nd) - (a+d); else if(u==v+1) (nc+nb) - (c+b); else (na+nb+nc+nd) - (a+b+c+d);
}
// all 8 neighbours var [const] dirs=ROList(1, -1, 10, -10, 9, 11, -11, -9),
fmt="k:%10,d T: %8.4f Es: %8.4f".fmt; // since we use it twice
fcn sa(kmax,kT=10){
s:=List(0, [1..99].walk().shuffle().xplode(), 0); // random path from 0 to 0 println("E(s0) %f".fmt(Es(s))); // random starter Emin:=Es(s); // E0 foreach k in (kmax){ if(0==k%(kmax/10)) println(fmt(k,T(k,kmax,kT),Es(s))); u:=(1).random(100); // city index 1 99 cv:=s[u] + dirs[(0).random(8)]; // city number if(not (0<cv<100)) continue; // bogus city if(dist(s[u],cv)>5) continue; // check true neighbour (eg 0 9) v:=s.index(cv,1); // city index deltae:=dE(s,u,v); if(deltae<0 or // always move if negative
P(deltae,k,kmax,kT)>=(0.0).random(1)){ s.swap(u,v); Emin+=deltae;
} // (assert (= (round Emin) (round (Es s)))) }//foreach println(fmt(kmax,T(kmax-1,kmax,kT),Es(s))); println("E(s_final) %f".fmt(Emin)); println("Path: ",s.toString(*));
}</lang> <lang zkl>sa(0d1_000_000,1);</lang>
- Output:
E(s0) 540.897080 k: 0 T: 1.0000 Es: 540.8971 k: 100,000 T: 0.9000 Es: 181.5102 k: 200,000 T: 0.8000 Es: 167.1944 k: 300,000 T: 0.7000 Es: 159.0975 k: 400,000 T: 0.6000 Es: 170.2344 k: 500,000 T: 0.5000 Es: 130.9919 k: 600,000 T: 0.4000 Es: 115.3422 k: 700,000 T: 0.3000 Es: 113.9280 k: 800,000 T: 0.2000 Es: 106.7924 k: 900,000 T: 0.1000 Es: 103.7213 k: 1,000,000 T: 0.0000 Es: 103.7213 E(s_final) 103.721349 Path: L(0,10,11,21,20,30,40,50,60,70,80,81,71,72,73,63,52,62,61,51,41,31,32,22,12,13,14,15,25,16,17,18,28,27,26,36,35,45,34,24,23,33,42,43,44,54,53,64,74,84,83,82,90,91,92,93,94,95,85,86,96,97,87,88,98,99,89,79,69,68,78,77,67,66,76,75,65,55,56,46,37,38,48,47,57,58,59,49,39,29,19,9,8,7,6,5,4,3,2,1,0)