SSA.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Salp Swarm Algorithm Implementation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Necessary Libraries\n",
    "import numpy  as np\n",
    "import math\n",
    "import random\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Iniatializing the population coordinates\n",
    "\n",
    "Iniatializing variable coordinates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def InitialPositions(populationSize, xmin, xmax, iterations, obj_func):\n",
    "    \n",
    "    # dimensions of variable\n",
    "    dimensions = len(xmin)\n",
    "    \n",
    "    \n",
    "    # initialize the population coordinates all 0 using numpy----> size = (populationSize, dimensions+1)\n",
    "    # In extra column is used to store the objective function value\n",
    "    \n",
    "    \n",
    "    pos = np.zeros((populationSize, dimensions))\n",
    "                   \n",
    "    \n",
    "    \n",
    "    # initialize the population coordinates randomly in the given search space\n",
    "                   \n",
    "    for i in range(populationSize):\n",
    "        for j in range(dimensions):\n",
    "            \n",
    "            # position of ith variable in jth dimension is no between xmin and xmax in jth dimension\n",
    "            # number is generated randomly \n",
    "            pos[i,j] = random.uniform(xmin[j], xmax[j])\n",
    "            \n",
    "               \n",
    "        #  objective function value of the generated coordinates of ith varaiable and store it in last column\n",
    "        pos[i,-1] = obj_func(pos[i,0:-1])\n",
    "        \n",
    "    return pos\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Initializing the food source \n",
    "\n",
    "In our case randomly assigning the optimum value \n",
    "\n",
    "or\n",
    "\n",
    "Randomly assigning coordinates to food source"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def InitialFood(dimensions, obj_func):\n",
    "    \n",
    "    # initialize the food coordinates all 0 using numpy----> size = (1, dimensions+1)\n",
    "    # extra column is used to store the objective function value\n",
    "    \n",
    "    fpos = np.zeros((1, dimensions+1))\n",
    "    \n",
    "    # Assigning values to all dimensions (except last col)\n",
    "    for j in range(dimensions):\n",
    "        fpos[0,j] = 0.0\n",
    "        \n",
    "    # calculating the objective value for this coordinates\n",
    "    fpos[0,-1] = obj_func(fpos[0,0:-1])\n",
    "    \n",
    "    return fpos\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Update Food Position or coordinates\n",
    "\n",
    "i.e updating optima by selecting the best coordinates whose obj function values is min/max depending upon quesiton default here is minimum"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def updateFood(pos, fpos):\n",
    "    \n",
    "    for i in range(pos.shape[0]):\n",
    "        \n",
    "        # check the given variable obj fun val is min\n",
    "        if (fpos[0,-1] > pos[i,-1]):\n",
    "            \n",
    "            # Update the coordinates of food position with salp coordinates\n",
    "            for j in range(pos.shape[1]):\n",
    "                fpos[0,j] = pos[i,j]\n",
    "    return fpos"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Update Salps coordinates\n",
    "\n",
    "c2, c3 are constants which are randomly generated using rand function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_position(pos, fpos, c1, xmin, xmax, obj_func):\n",
    "    \n",
    "    \n",
    "    # size of slaps population\n",
    "    n = len(pos[0])\n",
    "    \n",
    "    # dimensions of variable\n",
    "    dimensions = len(xmin)\n",
    "    \n",
    "    #updating coordinates...........\n",
    "    \n",
    "    for i in range(n):\n",
    "        \n",
    "        if(i<=n/2):\n",
    "            \n",
    "            \n",
    "            #updating the dimension of ith variable\n",
    "            \n",
    "            for j in range(dimensions):\n",
    "                \n",
    "                \n",
    "                # generate c2, c3 randomly\n",
    "                c2 = int.from_bytes(os.urandom(8), byteorder = \"big\") / ((1 << 64) - 1)\n",
    "                c3 = int.from_bytes(os.urandom(8), byteorder = \"big\") / ((1 << 64) - 1)\n",
    "                \n",
    "                if (c3 >= 0.5): #c from eqn in paper\n",
    "                    \n",
    "                    # here clip function is used to map the coordinates which are out of search space\n",
    "                    \n",
    "                    pos[i,j] = np.clip((fpos[0,j] + c1*((xmax[j] - xmin[j])*c2 + xmin[j])), xmin[j], xmax[j])\n",
    "                else:\n",
    "                    \n",
    "                    pos[i,j] = np.clip((fpos[0,j] - c1*((xmax[j] - xmin[j])*c2 + xmin[j])), xmin[j], xmax[j])\n",
    "                    \n",
    "            \n",
    "            \n",
    "        elif(i > n/2 and i < n + 1):\n",
    "            \n",
    "            for j in range(dimensions):\n",
    "                \n",
    "                # average of ith and i-1 th variable coordinates\n",
    "                pos[i,j] = np.clip(((pos[i - 1,j] + pos[i,j])/2), xmin[j], xmax[j]) \n",
    "                \n",
    "                \n",
    "        # calculating the objective value for this updated coordinates\n",
    "        pos[i,-1] = obj_func(pos[i,0:-1])         \n",
    "    return pos\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###   Salp Swarm Function\n",
    "\n",
    "populationSize is the number of salps\n",
    "\n",
    "xmin is the array of minimum values of the variables\n",
    "\n",
    "xmax is the array of maximum values of the variables\n",
    "\n",
    "obj_func is the objective function\n",
    "\n",
    "pos contains the coordiantes of all salps presnt in the population\n",
    "\n",
    "l is the current iteration \n",
    "\n",
    "L is the maximum number of iterations.\n",
    "\n",
    "fpos contains the coordiantes of food\n",
    "\n",
    "c1 is a constant"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ssa(populationSize, xmin, xmax, iterations, obj_func):\n",
    "    \n",
    "    L = iterations\n",
    "    # dimensions of variable\n",
    "    dimensions = len(xmin) \n",
    "    \n",
    "    # Initialize the population\n",
    "    pos  =  InitialPositions(populationSize, xmin, xmax, iterations, obj_func)\n",
    "    \n",
    "    # Initialize the global optimum in our case that is food of salps\n",
    "    fpos =  InitialFood(dimensions, obj_func)\n",
    "    \n",
    "    \n",
    "    # Iterations or Next Generations\n",
    "    for l in range(L): \n",
    "        \n",
    "        \n",
    "        # Printing the optimal or food positon after every Genration\n",
    "        print(\"Generation ={} f(x) = {}\".format(l, fpos[0-1]))\n",
    "        \n",
    "        \n",
    "        # updating the c1 constant\n",
    "        c1 = 2*math.exp(-(4*(l/L))**2)\n",
    "        \n",
    "        \n",
    "        # Update optima i.e food position\n",
    "        fpos = updateFood(pos, fpos)\n",
    "        \n",
    "        # Update positions of salps\n",
    "        \n",
    "        pos = update_position(pos, fpos, c1, xmin, xmax, obj_func)  \n",
    "        \n",
    "    \n",
    "    \n",
    "    # print the optimal answer or food position\n",
    "    print(fpos)\n",
    "    \n",
    "    return fpos"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bench Mark Function\n",
    "\n",
    "1. Sphere Model\n",
    "2. rastrigen\n",
    "3. ackley\n",
    "\n",
    "To test the bench mark function undo the comment and run the script"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generation =0 f(x) = [0. 0. 0.]\n",
      "Generation =1 f(x) = [0. 0. 0.]\n",
      "Generation =2 f(x) = [0. 0. 0.]\n",
      "Generation =3 f(x) = [0. 0. 0.]\n",
      "Generation =4 f(x) = [0. 0. 0.]\n",
      "Generation =5 f(x) = [0. 0. 0.]\n",
      "Generation =6 f(x) = [0. 0. 0.]\n",
      "Generation =7 f(x) = [0. 0. 0.]\n",
      "Generation =8 f(x) = [0. 0. 0.]\n",
      "Generation =9 f(x) = [0. 0. 0.]\n",
      "Generation =10 f(x) = [0. 0. 0.]\n",
      "Generation =11 f(x) = [0. 0. 0.]\n",
      "Generation =12 f(x) = [0. 0. 0.]\n",
      "Generation =13 f(x) = [0. 0. 0.]\n",
      "Generation =14 f(x) = [0. 0. 0.]\n",
      "Generation =15 f(x) = [0. 0. 0.]\n",
      "Generation =16 f(x) = [0. 0. 0.]\n",
      "Generation =17 f(x) = [0. 0. 0.]\n",
      "Generation =18 f(x) = [0. 0. 0.]\n",
      "Generation =19 f(x) = [0. 0. 0.]\n",
      "Generation =20 f(x) = [0. 0. 0.]\n",
      "Generation =21 f(x) = [0. 0. 0.]\n",
      "Generation =22 f(x) = [0. 0. 0.]\n",
      "Generation =23 f(x) = [0. 0. 0.]\n",
      "Generation =24 f(x) = [0. 0. 0.]\n",
      "Generation =25 f(x) = [0. 0. 0.]\n",
      "Generation =26 f(x) = [0. 0. 0.]\n",
      "Generation =27 f(x) = [0. 0. 0.]\n",
      "Generation =28 f(x) = [0. 0. 0.]\n",
      "Generation =29 f(x) = [0. 0. 0.]\n",
      "Generation =30 f(x) = [0. 0. 0.]\n",
      "Generation =31 f(x) = [0. 0. 0.]\n",
      "Generation =32 f(x) = [0. 0. 0.]\n",
      "Generation =33 f(x) = [0. 0. 0.]\n",
      "Generation =34 f(x) = [0. 0. 0.]\n",
      "Generation =35 f(x) = [0. 0. 0.]\n",
      "Generation =36 f(x) = [0. 0. 0.]\n",
      "Generation =37 f(x) = [0. 0. 0.]\n",
      "Generation =38 f(x) = [0. 0. 0.]\n",
      "Generation =39 f(x) = [0. 0. 0.]\n",
      "Generation =40 f(x) = [0. 0. 0.]\n",
      "Generation =41 f(x) = [0. 0. 0.]\n",
      "Generation =42 f(x) = [0. 0. 0.]\n",
      "Generation =43 f(x) = [0. 0. 0.]\n",
      "Generation =44 f(x) = [0. 0. 0.]\n",
      "Generation =45 f(x) = [0. 0. 0.]\n",
      "Generation =46 f(x) = [0. 0. 0.]\n",
      "Generation =47 f(x) = [0. 0. 0.]\n",
      "Generation =48 f(x) = [0. 0. 0.]\n",
      "Generation =49 f(x) = [0. 0. 0.]\n",
      "Generation =50 f(x) = [0. 0. 0.]\n",
      "Generation =51 f(x) = [0. 0. 0.]\n",
      "Generation =52 f(x) = [0. 0. 0.]\n",
      "Generation =53 f(x) = [0. 0. 0.]\n",
      "Generation =54 f(x) = [0. 0. 0.]\n",
      "Generation =55 f(x) = [0. 0. 0.]\n",
      "Generation =56 f(x) = [0. 0. 0.]\n",
      "Generation =57 f(x) = [0. 0. 0.]\n",
      "Generation =58 f(x) = [0. 0. 0.]\n",
      "Generation =59 f(x) = [0. 0. 0.]\n",
      "Generation =60 f(x) = [0. 0. 0.]\n",
      "Generation =61 f(x) = [0. 0. 0.]\n",
      "Generation =62 f(x) = [0. 0. 0.]\n",
      "Generation =63 f(x) = [0. 0. 0.]\n",
      "Generation =64 f(x) = [0. 0. 0.]\n",
      "Generation =65 f(x) = [0. 0. 0.]\n",
      "Generation =66 f(x) = [0. 0. 0.]\n",
      "Generation =67 f(x) = [0. 0. 0.]\n",
      "Generation =68 f(x) = [0. 0. 0.]\n",
      "Generation =69 f(x) = [0. 0. 0.]\n",
      "Generation =70 f(x) = [0. 0. 0.]\n",
      "Generation =71 f(x) = [0. 0. 0.]\n",
      "Generation =72 f(x) = [0. 0. 0.]\n",
      "Generation =73 f(x) = [0. 0. 0.]\n",
      "Generation =74 f(x) = [0. 0. 0.]\n",
      "Generation =75 f(x) = [0. 0. 0.]\n",
      "Generation =76 f(x) = [0. 0. 0.]\n",
      "Generation =77 f(x) = [0. 0. 0.]\n",
      "Generation =78 f(x) = [0. 0. 0.]\n",
      "Generation =79 f(x) = [0. 0. 0.]\n",
      "Generation =80 f(x) = [0. 0. 0.]\n",
      "Generation =81 f(x) = [0. 0. 0.]\n",
      "Generation =82 f(x) = [0. 0. 0.]\n",
      "Generation =83 f(x) = [0. 0. 0.]\n",
      "Generation =84 f(x) = [0. 0. 0.]\n",
      "Generation =85 f(x) = [0. 0. 0.]\n",
      "Generation =86 f(x) = [0. 0. 0.]\n",
      "Generation =87 f(x) = [0. 0. 0.]\n",
      "Generation =88 f(x) = [0. 0. 0.]\n",
      "Generation =89 f(x) = [0. 0. 0.]\n",
      "Generation =90 f(x) = [0. 0. 0.]\n",
      "Generation =91 f(x) = [0. 0. 0.]\n",
      "Generation =92 f(x) = [0. 0. 0.]\n",
      "Generation =93 f(x) = [0. 0. 0.]\n",
      "Generation =94 f(x) = [0. 0. 0.]\n",
      "Generation =95 f(x) = [0. 0. 0.]\n",
      "Generation =96 f(x) = [0. 0. 0.]\n",
      "Generation =97 f(x) = [0. 0. 0.]\n",
      "Generation =98 f(x) = [0. 0. 0.]\n",
      "Generation =99 f(x) = [0. 0. 0.]\n",
      "[[0. 0. 0.]]\n"
     ]
    }
   ],
   "source": [
    "def sphereModel(variables_values = [0,0]):\n",
    "    func_value = 0\n",
    "    \n",
    "    for i in range(len(variables_values)):\n",
    "        func_value += variables_values[i]**2\n",
    "    return func_value\n",
    "\n",
    "f1 = ssa(populationSize = 15, xmin = [-5.12,-5.12], xmax = [5.12,5.12], iterations = 100, obj_func = sphereModel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rastrigen(variables_values = [0,0]):\n",
    "    func_value = 0\n",
    "    \n",
    "    for i in range(len(variables_values)):\n",
    "        func_value += (variables_values[i]**2-10 * math.cos(2* math.pi *variables_values[i]))\n",
    "        \n",
    "    return func_value +10 * len(variables_values)\n",
    "\n",
    "# f2 = ssa(populationSize = 15, xmin = [-5.12,-5.12], xmax = [5.12,5.12], iterations = 100, obj_func = rastrigen) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ackley(variables_values = [0,0]):\n",
    "    func_value = 0\n",
    "    \n",
    "    a=20\n",
    "    b=0.2\n",
    "    c=2 * math.pi\n",
    "    \n",
    "    part1 = 0\n",
    "    \n",
    "    part2 = 0\n",
    "    \n",
    "    n = len(variables_values)\n",
    "    \n",
    "    for i in range(n):\n",
    "        part1 += variables_values[i]**2\n",
    "        part2 += math.cos(c * variables_values[i])\n",
    "        \n",
    "    part1 = math.exp(-b * math.sqrt(part1/n))\n",
    "    \n",
    "    part2 = math.exp(part2/n)\n",
    "    \n",
    "    \n",
    "    func_value = -a * part1 - part2 + a + math.exp(1)\n",
    "    \n",
    "    return func_value\n",
    "\n",
    "# f3 = ssa(populationSize = 15, xmin = [-32.768,-32.768], xmax = [32.768,32.768], iterations = 100, obj_func = ackley) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}