Slide 1

Slide 1 text

AVMf An Open-Source Implementation of the Alternating Variable Method Gregory M. Kapfhammer Phil McMinn SSBSE 2016 October 9, 2016

Slide 2

Slide 2 text

AVM is Everywhere Application Domains

Slide 3

Slide 3 text

AVM is Everywhere Application Domains Workloads

Slide 4

Slide 4 text

AVM is Everywhere Application Domains Workloads Testing

Slide 5

Slide 5 text

AVM is Everywhere Application Domains Workloads Testing SoftwareProductLines

Slide 6

Slide 6 text

AVM is Everywhere Application Domains Workloads Testing SoftwareProductLines AVM is used in varied domains

Slide 7

Slide 7 text

Exploring AVM Input Vector

Slide 8

Slide 8 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn)

Slide 9

Slide 9 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)

Slide 10

Slide 10 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Objective Function

Slide 11

Slide 11 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)

Slide 12

Slide 12 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratory Moves

Slide 13

Slide 13 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratory Moves Positive or negative “direction”?

Slide 14

Slide 14 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves

Slide 15

Slide 15 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value?

Slide 16

Slide 16 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value? Yes! pattern or No! explore

Slide 17

Slide 17 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value? Yes! pattern or No! explore

Slide 18

Slide 18 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)

Slide 19

Slide 19 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves

Slide 20

Slide 20 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves Consider all input vector variables

Slide 21

Slide 21 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves Consider all input vector variables x = (x1, x2, . . . , xn)

Slide 22

Slide 22 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves x = (x1, x2, . . . , xn) Revisit each xi in the input vector

Slide 23

Slide 23 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves x = (x1, x2, . . . , xn) Restart for local optimum

Slide 24

Slide 24 text

Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves x = (x1, x2, . . . , xn) Continue until termination condition

Slide 25

Slide 25 text

AVM Innovations Search Algorithms

Slide 26

Slide 26 text

AVM Innovations Search Algorithms Geometric

Slide 27

Slide 27 text

AVM Innovations Search Algorithms Geometric Lattice

Slide 28

Slide 28 text

AVM Innovations Search Algorithms Geometric Lattice IteratedPattern

Slide 29

Slide 29 text

AVM Innovations Search Algorithms Geometric Lattice IteratedPattern Better search for many landscapes

Slide 30

Slide 30 text

AVM Innovations Search Algorithms Geometric Lattice IteratedPattern Provably faster for unimodal

Slide 31

Slide 31 text

AVM Innovations Representations

Slide 32

Slide 32 text

AVM Innovations Representations Decimals

Slide 33

Slide 33 text

AVM Innovations Representations Decimals Strings

Slide 34

Slide 34 text

AVM Innovations Representations Decimals Strings Integers

Slide 35

Slide 35 text

AVM Innovations Representations Decimals Strings Integers Handle real-world programs

Slide 36

Slide 36 text

Missing Features Data? Method?

Slide 37

Slide 37 text

Missing Features Data? Method?

Slide 38

Slide 38 text

Key Challenge Prior AVMs lack provably faster methods!

Slide 39

Slide 39 text

Tools Using AVM Test Generation

Slide 40

Slide 40 text

Tools Using AVM Test Generation AUSTIN

Slide 41

Slide 41 text

Tools Using AVM Test Generation AUSTIN EvoSuite

Slide 42

Slide 42 text

Tools Using AVM Test Generation AUSTIN EvoSuite SchemaAnalyst

Slide 43

Slide 43 text

Tools Using AVM Test Generation AUSTIN EvoSuite SchemaAnalyst AVM is used in many tools

Slide 44

Slide 44 text

Extracting AVM Fitness? Search?

Slide 45

Slide 45 text

Extracting AVM Fitness? Search?

Slide 46

Slide 46 text

Extracting AVM Fitness? Search?

Slide 47

Slide 47 text

Key Challenge Hard to extract AVM from custom software!

Slide 48

Slide 48 text

Rescued by AVMf

Slide 49

Slide 49 text

Rescued by AVMf OriginalAVMplusenhance- mentsfordataandsearch

Slide 50

Slide 50 text

Rescued by AVMf Clearimplementa- tionofcorealgorithms

Slide 51

Slide 51 text

Rescued by AVMf Adherestotheprinciples ofobject-orienteddesign

Slide 52

Slide 52 text

Rescued by AVMf Freeandopen-sourcesoft- warefromavmframework.org

Slide 53

Slide 53 text

Rescued by AVMf

Slide 54

Slide 54 text

Rescued by AVMf NewApplicationDomain

Slide 55

Slide 55 text

Rescued by AVMf

Slide 56

Slide 56 text

Rescued by AVMf NewSearchAlgorithm

Slide 57

Slide 57 text

Design of AVMf

Slide 58

Slide 58 text

Design of AVMf Con gure

Slide 59

Slide 59 text

Design of AVMf Con gure Represent

Slide 60

Slide 60 text

Design of AVMf

Slide 61

Slide 61 text

Design of AVMf Objective

Slide 62

Slide 62 text

Design of AVMf Objective Report

Slide 63

Slide 63 text

Design of AVMf

Slide 64

Slide 64 text

Design of AVMf Seethepaperformoredesign andimplementationdetails

Slide 65

Slide 65 text

Design of AVMf Thetool’swebsitecontains extensivedocumentation

Slide 66

Slide 66 text

AVMf Demonstration

Slide 67

Slide 67 text

AVMf Demonstration javaorg.avmframework. examples.Quadratic

Slide 68

Slide 68 text

AVMf Demonstration javaorg.avmframework. examples.StringOptimization

Slide 69

Slide 69 text

AVMf Demonstration javaorg.avmframework. examples.GenerateInputData

Slide 70

Slide 70 text

AVMf Demonstration Input→Output StochasticBehavior

Slide 71

Slide 71 text

AVMf Demonstration Alreadyrun: gitclone&mvnpackage

Slide 72

Slide 72 text

AVMf’s Contributions

Slide 73

Slide 73 text

AVMf’s Contributions Overcomesthe challengesofusingAVM

Slide 74

Slide 74 text

AVMf’s Contributions Provablyfastersearches andnewdatatypes

Slide 75

Slide 75 text

AVMf’s Contributions Accessibleobject-oriented andalgorithmicdesign

Slide 76

Slide 76 text

AVMf’s Contributions Open-sourcedownload fromavmframework.org