CS152 Lab 6: Searching and Optimization

Java Python CS152

Lab 6: Searching and Optimization

The purpose of this lab/project is to get you thinking about automating a scientific experiment and optimizing the parameters of a process.

This is the second part of our study of the elephant population in Kruger National Park in South  Africa. This week we'll be focusing on how varying the parameters will change the management strategy.

Rather than manually exploring the parameter space, we will automate the process and use a search method to find the optimal darting percentage given the simulation parameters.

Setting up the Lab/Project

Create a Lab06 folder and open VS Code in it.

L1. Implement binary search on a sorted list

In your Lab06 folder create a new file, search.py. Then create a function searchSortedList that takes in two parameters, a list and a target value. You can use the following template.

def searchSortedList ( mylist, value ):

# assign to the variable done, the value False

# assign to the variable found, the value False

# assign to the variable count, the value 0

# assign to the variable maxIdx, the one less than the

length of mylist

# assign to the variable minIdx, the value 0

# start a while loop that executes while done is not True

# increment count  (which keeps track of how many times

the loop executes

# assign to testIndex the average of maxIdx and minIdx

(use integer math)

# if the myList value at testIndex is less than

value

# assign to minIdx the value testIndex + 1

# el if the myList value at testIndex is greater

than value

# assign to maxIdx the value testIndex - 1

# else

# set done to True

# set found to True

# if maxIdx is less than minIdx

# set done to True

# set found to False

return (found, count)

Once you have coded that up, import the random package, copy the following test function, and run it to see if your function finds the value 42 in the list (it should). Think about why it can take a different number of steps each time you run the program. Experiment with different values of N and relate the number of steps that the function takes to complete to log2(N).

def test ():

a = []

N = 10**6

for i in range (N):

a.append ( random.rand int (0,N) )

a.append (42)

a.sort ()

print (searchSortedList ( a, 42 ))

if name == " main ":

test ()

This function is practice for creating the optimization function for the elephant simulation. That function will have a similar structure, and it will also execute a binary search, but it will have more parameters and be more flexible in how it works.

L2. Write a function to generate a default parameter list

Create a Project06 folder and copy your elephant.py file from the last project.

Next, add a new function to your elephant.py file. The function should be&nb CS152 Lab 6: Searching and Optimization sp;called

defaultParameters (). It should take no arguments and return a list with all of the necessary parameters for the simulation with their default values.

calvingIn t = 3.1

probDart = 0.0

juvAge = 12

maxAge = 60

probCalfSurvival = 0.85

probAdultSurvival = 0.996

probSeniorSurvival = 0.2

carryingCapacity = 1000

numYears = 200

NOTE: For this project, set the default carrying capacity to 1000.

L3. Write an elephant simulation function

Next, also in your elephant.py file, create a function elephantSim. The arguments to the function should be probDart and a parameter list, which should have a default value of None. In other words, set it up as the following.

def elephantSim ( probDart, inputParameters = None ):

First, set up the parameters. If inputParameters is equal to the value None, then assign to parameters the result of calling your defaultParameters function. Otherwise, assign to parameters the value of inputParameters.

Next, store the parameter value for probDart in the correct location in the parameters list.

Then, declare an empty list named results where we will store the return from the call to runSimulation as follows:

results = results + runSimulation (parameters)

Place this in a loop which will call runSimulation five times and collect all of the results in a single list. The structure of the results list is that of a list of lists. Each sublist consists of the total population and other statistics for a given simulation year.

Finally, loop over the results list and calculate the average total population. Remember, the total population for each year is the first element in each sublist.

elephantSim will return the following crucial piece of information: (carrying capacity) - (average total population of the five simulations). Cast this value as an integer before returning. Just like our cost function in all our search algorithms, this metric will guide the search for the optimal darting probability.

You can think of this return value as specifying whether too many or too few elephants are being darted. If the return value is negative, then the population is too big and we need to tweak the darting probability higher so that the elephant population shrinks; and if the return value is positive, then the population is too small and we need to tweak the darting probability lower so that the elephant population grows.

L4. Test your elephantSim function

Once you have written elephantSim, use the testfile test_elephantSim.py to evaluate its  performance. It tests the percent darted for five different values. The difference should go from negative (darting probability too low) to positive (darting probability too high) for the default simulation parameters.

Example output: (your numbers may be different because of the random and your elephant parameter boundaries are different)

probDarting 0.405 diff -385

probDarting 0.415 diff -331

probDarting 0.425 diff -172

probDarting 0.435 diff -8

probDarting 0