遗传算法代写 ASSIGNMENT TWO: GENETIC ALGORITHM

ASSIGNMENT TWO: GENETIC ALGORITHM

ADVANCED PROGRAMMING TECHNIQUES – SEMESTER 2, 2018

SUMMARY

In this assignment, you will use your C programming skills to build a simple Genetic Algorithm.

A genetic algorithm (GA) is a heuristic search technique inspired by natural selection. A population of candidate solutions to a problem is iteratively evolved towards an optimal solution (as measured by an evaluation function) by:

  •   Evaluating the ‘fitness’ of candidate solutions according to the evaluation function
  •   Probabilistically selecting ‘fitter’ candidate solutions from the population to act as

    ‘parents’ for the next generation of candidate solutions

  •   Introducing constrained novelty to the new generation via operations analogous to

    reproduction and mutation. (Note: the selection function will then tend to remove ‘bad’ novelty).

    The course Canvas assignment 2 page has links to descriptions of the above operations.

    Genetic algorithms are used in computing and industry in areas as diverse as hardware design optimization, scheduling and machine learning. The genetic algorithm of this assignment is applied to two problems:

    minfn: determine variables to satisfy an equation
    pcbmill: minimise the drill head movement (and thus the time taken to drill) of a mill

    for drilling holes in a Printed Circuit Board (PCB).

    The course Canvas assignment 2 page has links to specifications for minfn and pcbmill and sample data files that they can use.

    In the following sections of this document, the details of your assignment are explained. Read them thoroughly and ask your questions from the teaching team. You are expected to understand every requirement explained in this document and implement all of them accordingly.

    A start-up code is provided along with this document. Just producing code to implement the requirements is not enough. You must use the start-up code in your assignment, developing and using the given functions in a reasonable way to implement the requirements. You are not permitted to change the given function prototypes. The prototypes have been written in such a way to make you demonstrate your ability to use certain concepts taught in this course. You are required to submit the exact files provided in the start-up code after completing the specified tasks.

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

SUBMISSION

Total mark: 30% of the final mark
Assignment demo: During tutorial sessions of week 10 (24th – 27th September 2018). Final submission: End of week 12 – Friday, 12 October – 10:00 PM

PLAGIARISM NOTICE

Plagiarism is a very serious offence. The minimum first time penalty for plagiarised assignments is zero marks for that assignment. Please keep in mind that RMIT University uses plagiarism detection software to detect plagiarism and that all assignments will be tested using this software. More information about plagiarism can be found here: https://www.rmit.edu.au/students/student-essentials/rights-and-responsibilities/academic- integrity.

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

DESCRIPTION OF THE PROGRAM CORE REQUIREMENTS

REQ1: Command-line arguments

The user of your program must be able to execute it by passing in the following arguments:

./ga geneType alleleSize popSize numGen inputFile [outputFile]

where:

  •   ./ga is the name of the executable
  •   geneType is a string, of either minfn or pcbmill – indicating which problem the GA is

    being applied to

  •   alleleSize is a positive non-zero integer specifying the length of a candidate solution

    (chromosome)

  •   popSize is a positive non-zero integer specifying the number of candidate solutions in

    the population

  •   numGen is a positive integer specifying the number of generations to run the GA for
  •   inputFile is a string specifying the name of a data file to read
  •   outputFile is an optional string specifying the name of an output file to write

    The main program must verify the number of command-line arguments is as expected and ensure they can be converted to valid types and values.

    REQ2: Loading the vector input data file

    Your program needs to load into memory the data from the input data file specified in the command-line argument. You will need to tokenise this data before you can load it into the system. In this requirement, you can assume the specified input file is completely valid. Dealing with invalid input files is covered in a separate section of the assignment.

    You will need to use the InVTable structure defined in the start-up code to store your input data. The input data is organised as a series of input vectors, one per line. The syntax of a line is:

    InputVector:vector-number(vector-element1,vector-element2….vector-elementN) Where

  •   vector-number is a positive integer, incrementing sequentially from 0
  •   The vector is a sequence of at least one (comma-separated) integer vector-element
  •   Each InputVector in the file must have the same number of vector-elements

    For example:

         InputVector:0(1,3,4,2,40)
         InputVector:1(2,7,1,2,9)
    

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

REQ3: Implementing and calling function test_minfn

Function test_minfn() tests the operation of functions gene_create_rand_gene, mutate_minfn, crossover_minfn, gene_print and gene_free. function test_minfn() is only compiled and called if macro DEBUG is defined. The start-up code includes a description of the function calls necessary to implement test_minfn().

Functions mutate_minfn, crossover_minfn and gene_print also use conditional compilation, including debug statements if macro DEBUG is defined. In particular, if DEBUG is defined:

  •   mutate_minfn will use a mutation point of chromosome index 2
  •   crossover_minfn will use a crossover point of chromosome index 2

    An example of output from a call of test_minfn() is:

    MINFN gene:
    chrom:13,16,27,25,23,25 fit: 0.000 raw: 0.000 Mutate: index 2
    chrom:13,16,12,25,23,25 fit: 0.000 raw: 0.000 MINFN genes:
    chrom: 9, 1, 2, 7,20,19 fit: 0.000 raw: 0.000 chrom:23,16, 0, 6,22,16 fit: 0.000 raw: 0.000 Crossover: index 2
    chrom: 9, 1, 2, 6,22,16 fit: 0.000 raw: 0.000 Since some functions call rand(), output may vary from above.

    REQ4: Implementing and calling function test_pcbmill

    Function test_pcbmill() tests the operation of functions gene_create_rand_gene, mutate_pcbmill, crossover_pcbmill, gene_print and gene_free. function test_pcbmill() is only compiled and called if macro DEBUG is defined. The start-up code includes a description of the function calls necessary to implement test_pcbmill().

    Functions mutate_pcbmill, crossover_pcbmill and gene_print also use conditional compilation, including debug statements if macro DEBUG is defined. In particular, if DEBUG is defined:

  •   mutate_pcbmill will use a mutation points of chromosome indexes 2 and 4
  •   crossover_pcbmill will use a crossover point of chromosome indexes 2 and 4

    An example of output from a call of test_pcbmill() is:

PCBMILL gene:
chrom: 1, 4, 0, 3, 5, 2 fit:
Mutate: index1 2 index2 4
chrom: 1, 4, 5, 3, 0, 2 fit:
PCBMILL genes:
chrom: 5, 1, 2, 3, 4, 0 fit:
chrom: 0, 1, 2, 3, 5, 4 fit:
Crossover: index1 2 index2 4
chrom: 2, 3, 4, 0, 1, 5 fit: 0.000 raw: 0.000 Since some functions call rand(), output may vary from above.

Advanced Programming Techniques – Semester 2, 2018

Assignment Two (v1.0, 3 Sept 2018)

0.000 raw:  0.000
0.000 raw:  0.000
0.000 raw:  0.000
0.000 raw:  0.000

REQ5: Conditional compilation

Functions test_minfn and test_pcbmill should only be compiled and called if the macro DEBUG has been defined. Debug statements in functions mutate_pcbmill, mutate_minfn, crossover_pcbmill and crossover_minfn should also be compiled only when the macro DEBUG has been defined. This macro should be defined when the user calls make with the argument DEBUG.

REQ6: Makefile

We should be able to compile your program using a makefile, which should be submitted along with the code of your assignment. All compile commands must include the “-ansi -Wall – pedantic” compile options. You program should compile cleanly with these options; no error
or warning is acceptable even during the demo.

Your makefile must compile your program incrementally. That is, it should use object files as an intermediate form of compilation.

Your makefile should permit the defining of macro DEBUG from the command line, for use in conditional compilation of test_pcbmill() and test_minfn().

You should also include a target called “clean” that deletes unnecessary files from your working directory such as object files, executable files, core dump files etc. This directive should only be executed when the user types “make clean” at the command prompt.

REQ7: Implementing and calling pop_print_fittest

Function pop_print_fittest should be called each time a population has been evaluated and ready for reproduction. It should cause the the ‘fittest’ gene in the population to be displayed, along with the number of the current generation. As specified in the start-up code, the function must not be able to access any generation value outside of the function.

A sample output for running minfn for 4 generations is:

Gen: 0 chrom: 9, 1, 2, 7 fit: 0.437 raw: 6.000
Gen: 1 chrom: 9, 1, 2, 7 fit: 0.163 raw: 6.000
Gen: 2 chrom: 2, 1, 2,13 fit: 0.218 raw: 1.000
Gen: 3 chrom: 9, 3, 2, 7 fit: 0.220 raw: 0.000
Since some functions call rand(), and different input vector files may be used, output may vary from above.

REQ8: Evolution of the population

The program should correctly create an appropriate initial population and ‘evolve’ it for a specified number of generations, where each generation is evaluated and reproduced to create the next generation, displaying details of the fittest gene of each generation as described above.

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

REQ9: Implementation of function pointers

The Pop_list data type has three pointers to functions:

  •   create_rand_chrom which can be set to point to a function for creating a random chromosome. Prototypes for two such functions have been provided: create_pcbmill_chrom for creating a valid pcbmill chromosome, and create_minfn_chrom for creating a valid minfn chromosome.
  •   mutate_gene which can be set to point to a function for mutating a gene. Prototypes for two such functions have been provided: mutate_pcbmill for mutating a pcbmill gene, and mutate_minfn for mutating a minfn gene.
  •   crossover_genes which can be set to point to a function for performing crossover on

two genes. Prototypes for two such functions have been provided: crossover_pcbmill for performing crossover on pcbmill genes, and crossover_minfn for performing crossover on minfn genes.

 evaluate_fn which can be set to point to a function for evaluating the raw fitness of a gene. Prototypes for two such functions have been provided: eval_pcbmill for evaluating a pcbmill gene, and eval_minfn for evaluating a minfn gene.

Your program should appropriately set and use these functions pointers.

REQ10: Validating the input file

In the previous requirements, it was assumed that the input vector data file that you have loaded into memory were valid and contained no error. However, in real life this is not necessarily the case and you will be required to detect errors that may arise from reading a corrupt or invalid data file.

Examples of the corrupt and/or invalid data files include the following:

  •   There are lines of data in the file which does not contain enough fields to constitute a

    valid input vector;

  •   There are lines of data in the file which contain too many fields for a valid input vector;
  •   The number of fields is correct; however, the type of data available for one or more of

    the fields don’t match the expectation of the program;

  •   The data types match the expectation, but they are out of the acceptable range or have

    otherwise invalid values;

  •   The data file is empty;
  •   The data file doesn’t exist

    REQ11: Producing the output file

    If command-line argument outputFile is provided, a file with the same name as the argument should be created/truncated to zero length and all output directed to the file.

    REQ12: Memory leaks and abuses

    The start-up code requires the use of dynamic memory allocation. Therefore, you will need to check that your program does not contain memory leaks. Use the following valgrind command to check for memory leaks on problem minfn with a population 20 run for 50 generations and

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

submit the report in a text file named Requirement11a.txt along with the rest of the files in your project.

valgrind --leak-check=full --show-reachable=yes <command> <arguments>

valgrind ––leak-check=full –show-reachable=yes <command> <arguments>

Another common problem is memory abuses, which refers to situations such as reading from uninitialised memory, writing to memory addresses you should not have access to, and conditional statements that depend on uninitialised values. You can test for these again by using valgrind:

valgrind –track-origins=yes <command> <arguments>

The report generated by this command must be stored in a new file named Requirement11b.txt and included in your submission. Marks will only be awarded for this requirement if the reports generated by valgrind indicate that there are no memory leaks nor any other memory related problems.

REQ13: Abstract Data Types

In this assignment, you are implementing three Abstract Data Types; gene, population and InVTable. A number of function prototypes for gene have been provided, but the interface for the population and the InVTable are largely for you to develop. For this requirement, you will need to propose a list of interface functions for the population and InVTable ADTs and implement these. All references to these types should be via these interface functions.

REQ14: General requirements

You must read and follow the “Buffer Handling”, “Input Validation” and “Coding Conventions and Practices” requirements written in the “General Requirements” section of this specification.

REQ15: Demo

A demonstration is required for this assignment in order to show your progress. This will occur in your scheduled lab classes on the Unix machines used in this course. Demonstrations will be very brief, and you must be ready to demonstrate your work in a two-minute period when it is your turn. You must also have already uploaded your files to the assignment 2 submission link as proof of your work.

As part of the assignment demo, the tutor may ask you random questions about your source code. You may be asked to open your source files and explain the operation of a randomly picked section. In the event that you will not be able to answer the questions, we will have to make sure that you are the genuine author of the program.

Input validation will not be assessed during demonstrations, so you are advised to incorporate these only after you have implemented the demonstration requirements. Coding conventions and practices too will not be scrutinised, but it is recommended that you adhere to such requirements at all times.

During the demonstration you will be marked for implementing the requirements REQ3, REQ4 and REQ5 which should be completed and functional.

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

GENERAL REQUIREMENTS

GR1: Buffer handling

Your program must not suffer from buffer overflow if input is larger than expected.

GR2: Coding conventions and practices

Marks are awarded for good coding conventions/practices such as:

  •   Completing the header comment sections in each source file included in the submission.
  •   Avoiding global variables. If you still do not know what global variables are, do not

    assume. Ask the teaching team about it.

  •   Avoiding goto statements.
  •   Consistent use of spaces or tabs for indentation. Either consistently use tabs or three

    spaces for indentation. Be careful not to mix tabs and spaces. Each “block” of code

    should be indented one level.

  •   Keeping line lengths of code to a reasonable maximum such that they fit into 80

    columns.

  •   Appropriate commenting to explain non-trivial sections of the code.
  •   Writing header descriptions for all functions.
  •   Appropriate identifier names.
  •   Avoiding magic numbers (remember, it is not only about numbers).
  •   Avoiding platform specific code with system().
  •   General code readability.

    GR3: Functional abstraction

    We encourage the use of functional abstraction throughout your code. This is considered to be a good practice when developing software with multiple benefits. Abstracting code into separate functions reduces the possibility of bugs in your project, simplifies programming logic and make the debugging less complicated. We will look for some evidence of functional abstraction throughout your code.

    As a rule, if you notice that you have the same or very similar block of code in two different locations of your source, you can abstract this into a separate function. Another easy rule is that you should not have functions with more than approximately 50 lines of code. If you have noticeably longer functions, you are expected to break that function into multiple logical parts and create new functions accordingly.

    START-UP CODE

    We provide you with a start-up code that you must use in implementing your program. The start-up code includes several files which you must carefully read, understand and complete.

    STRUCTURE AND FILES

    The start-up code contains the following files:
    Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

  •   drive.h and drive.c: This is the main header and source file. The main method should be implemented in drive.c file. This is also the only place that srand should be called.
    gene.h and gene.c: These files include the declaration and body of the functions related to genes.
  •   pop.h and pop.c: These files contain the body of the logic for creating, managing and running a population of genes to implement a genetic algorithm.
  •   invector.h and invector.c: These files contain functions related to reading, validating, storing and accessing input vectors.
  •   utility.h: This file is used to declare the Boolean type used by multiple other modules.

    In addition to the provided functions in the files, you may need to use functional abstraction to break down the logic of some functions into calling additional smaller functions.

    MARKING GUIDE

    The following table contains the detailed breakdown of marks allocated for each requirement.

    Requirement Mark

    REQ1: Command-line arguments 6 REQ2: Loading the vector input data file 6 REQ3: Implementing and calling test_minfn 5 REQ4: Implementing and calling test_pcbmill 5 REQ5: Conditional compilation 6 REQ6: Makefile 5 REQ7: Implementing and calling pop_print_fittest 3 REQ8: Evolution of the population 20 REQ9: Implementation of function pointers 6

REQ10: Validating the input file REQ11: Producing the output file REQ12: Memory leaks and abuses REQ13: Abstract data types REQ14: General requirements REQ15: Demonstration in lab Total

3
2
5
6
12 (4+4+4) 10

100

PENALTIES

Marks will be deducted for the following:

  •   Compile errors and warnings (especially ensure it compiles with the -ansi flag!!).
  •   Fatal run-time errors such as segmentation faults, bus errors, infinite loops, etc.
  •   Missing files (affected components get zero marks).
  •   Files incorrectly named, or whole directories submitted.
  •   Not using start-up code or not filling-in the readme file.

    Programs with compile errors that cannot be easily corrected by the marker will result in a maximum possible score of 40% of the total available marks for the assignment.

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)

Any sections of the assignment that cause the program to terminate unexpectedly (i.e., segmentation fault, bus error, infinite loop, etc.) will result in a maximum possible score of 40% of the total available marks for those sections. Any sections that cannot be tested as a result of problems in other sections will also receive a maximum possible score of 40%.

It is not possible to resubmit your assignment after the deadline due to mistakes.

SUBMISSION INFORMATION

Create a zip archive using the following command on saturn, jupiter or titan:

This will create a .zip file named “username-a2.zip” with your username substituted for “username”. Note that you must test your archive file before submission to make sure that it is a valid archive, contains the necessary files, and can be extracted without any errors.

Submit the zip file before the submission deadline.

LATE SUBMISSION PENALTY

Late submissions attract a marking deduction of 10% of the maximum mark attainable per day for the first 5 days, including weekdays and weekends. After this time, a 100% deduction is applied.

zip ${USER}-a2 \
drive.h drive.c \
gene.h gene.c \
invector.h invector.c \
pop.h pop.c \

makefile \
Requirement11a.txt Requirement11b.txt

Advanced Programming Techniques – Semester 2, 2018 Assignment Two (v1.0, 3 Sept 2018)