FAC_LOC: fixed exercises 3.1 text which was duplicate of 3.2!

health-data-science-OR · Jan 27, 2021 · 795ffdf · 795ffdf
1 parent c2c6d62
commit 795ffdf
Show file tree

Hide file tree

Showing 2 changed files with 100 additions and 802 deletions.
diff --git a/optimisation/02_facility_location.ipynb b/optimisation/02_facility_location.ipynb
@@ -395,32 +395,40 @@
    "source": [
     "## Exercise 3.1 Generating an initial population.\n",
     "\n",
-    "Basic evolutionary strategies work by mutating the most promising solutions in the population.  There are many ways to implement mutation.  Here you will use `BasicFacilityLocationMutator`.  Each element in a solution has a constant probability of mutation (by default 1 / no. of facilities in a solution, but you may wish to set this higher.).  If a facility is chosen then it is replaced by a random facility current not in the solution.\n",
-    "\n",
-    "You can create a `BasicFacilityLocationMutator` as follows:\n",
+    "The first task when using a population based method is to create an initial random population of solutions!  For our purposes, this is a multi-dimensional array.  We can use an object of type `FacilityLocationPopulationGenerator` to do the work for us here.\n",
     "\n",
     "```python\n",
-    "mutator = FacilityLocationMutator(n_candidates=28, solution_size=4)\n",
-    "solution = np.array([1, 2, 3, 4])\n",
-    "\n",
-    "mutant = mutator.mutate(solution)\n",
-    "print(mutant)\n",
+    "from metapy.evolutionary.evolutionary import FacilityLocationPopulationGenerator\n",
     "```\n",
     "\n",
-    "To use a higher mutation rate:\n",
+    "`FacilityLocationPopulationGenerator` accepts three arguments when it is created:\n",
+    "\n",
+    "* n_candidates: int.  This is $P$ the number of candidate locations\n",
+    "* n_facilities: int. This is $p$ the number of facilities to place.\n",
+    "* random_seed: int, optional (default=None).  Set if you want a reproducible result.  For example = 42.\n",
+    "\n",
+    "`FacilityLocationPopulationGenerator` has a single method `generate` that accepts a parameter specifying the population size.  It returns a multi-dimensional numpy array.\n",
+    "\n",
+    "Let's assume you want have a problem with $P$ = 28, $p$ = 8 and we want to create a population of size 10.\n",
     "\n",
     "```python\n",
-    "mutator = FacilityLocationMutator(n_candidates=28, solution_size=4\n",
-    "                                  mutation_rate=0.6)\n",
-    "solution = np.array([1, 2, 3, 4])\n",
+    "#example solution\n",
+    "N_CANDIDATES = 28\n",
+    "N_FACILITIES = 8\n",
+    "SEED = 42\n",
+    "POPULATION_SIZE = 10\n",
     "\n",
-    "mutant = mutator.mutate(solution)\n",
-    "print(mutant)\n",
+    "gen = FacilityLocationPopulationGenerator(n_candidates=N_CANDIDATES,\n",
+    "                                          n_facilities=N_FACILITIES,\n",
+    "                                          random_seed=SEED)\n",
+    "\n",
+    "\n",
+    "gen.generate(population_size=POPULATION_SIZE)\n",
     "```\n",
     "\n",
-    "**Task**:\n",
-    "* Create a `FacilityLocationMutator` with 50 candidate solutions and a solution size of 10\n",
-    "* mutate the solution a total of 10 times and printout the generations."
+    "**Task:**\n",
+    "\n",
+    "* Run the example code given above to create an initial population.  You"
    ]
   },
   {