AVMf: An Open-Source Implementation of the Alternating Variable Method

Interested in learning more about this topic? Visit this web site to read the paper: https://www.gregorykapfhammer.com/research/papers/McMinn2016b/

Transcript

1. AVMf An Open-Source Implementation of the Alternating Variable Method Gregory M. Kapfhammer Phil McMinn SSBSE 2016 October 9, 2016
2. AVM is Everywhere Application Domains
3. AVM is Everywhere Application Domains Workloads
4. AVM is Everywhere Application Domains Workloads Testing
5. AVM is Everywhere Application Domains Workloads Testing SoftwareProductLines
6. AVM is Everywhere Application Domains Workloads Testing SoftwareProductLines AVM is used in varied domains
7. Exploring AVM Input Vector
8. Exploring AVM Input Vector x = (x1, x2, . . . , xn)
9. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)
10. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Objective Function
11. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)
12. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratory Moves
13. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratory Moves Positive or negative “direction”?
14. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves
15. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value?
16. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value? Yes! pattern or No! explore
17. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Pattern Moves Improve objective value? Yes! pattern or No! explore
18. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn)
19. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves
20. Exploring AVM Input Vector x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) x = (x1, x2, . . . , xn) Exploratoryand PatternMoves Consider all input vector variables
21. 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)
22. 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
23. 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
24. 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
25. AVM Innovations Search Algorithms
26. AVM Innovations Search Algorithms Geometric
27. AVM Innovations Search Algorithms Geometric Lattice
28. AVM Innovations Search Algorithms Geometric Lattice IteratedPattern
29. AVM Innovations Search Algorithms Geometric Lattice IteratedPattern Better search for many landscapes
30. AVM Innovations Search Algorithms Geometric Lattice IteratedPattern Provably faster for unimodal
31. AVM Innovations Representations
32. AVM Innovations Representations Decimals
33. AVM Innovations Representations Decimals Strings
34. AVM Innovations Representations Decimals Strings Integers
35. AVM Innovations Representations Decimals Strings Integers Handle real-world programs
36. Missing Features Data? Method?
37. Missing Features Data? Method?
38. Key Challenge Prior AVMs lack provably faster methods!
39. Tools Using AVM Test Generation
40. Tools Using AVM Test Generation AUSTIN
41. Tools Using AVM Test Generation AUSTIN EvoSuite
42. Tools Using AVM Test Generation AUSTIN EvoSuite SchemaAnalyst
43. Tools Using AVM Test Generation AUSTIN EvoSuite SchemaAnalyst AVM is used in many tools
44. Extracting AVM Fitness? Search?
45. Extracting AVM Fitness? Search?
46. Extracting AVM Fitness? Search?
47. Key Challenge Hard to extract AVM from custom software!
48. Rescued by AVMf
49. Rescued by AVMf OriginalAVMplusenhance- mentsfordataandsearch
50. Rescued by AVMf Clearimplementa- tionofcorealgorithms
51. Rescued by AVMf Adherestotheprinciples ofobject-orienteddesign
52. Rescued by AVMf Freeandopen-sourcesoft- warefromavmframework.org
53. Rescued by AVMf
54. Rescued by AVMf NewApplicationDomain
55. Rescued by AVMf
56. Rescued by AVMf NewSearchAlgorithm
57. Design of AVMf
58. Design of AVMf Con gure
59. Design of AVMf Con gure Represent
60. Design of AVMf
61. Design of AVMf Objective
62. Design of AVMf Objective Report
63. Design of AVMf
64. Design of AVMf Seethepaperformoredesign andimplementationdetails
65. Design of AVMf Thetool’swebsitecontains extensivedocumentation
66. AVMf Demonstration
67. AVMf Demonstration javaorg.avmframework. examples.Quadratic
68. AVMf Demonstration javaorg.avmframework. examples.StringOptimization
69. AVMf Demonstration javaorg.avmframework. examples.GenerateInputData
70. AVMf Demonstration Input→Output StochasticBehavior
71. AVMf Demonstration Alreadyrun: gitclone&mvnpackage
72. AVMf’s Contributions
73. AVMf’s Contributions Overcomesthe challengesofusingAVM
74. AVMf’s Contributions Provablyfastersearches andnewdatatypes
75. AVMf’s Contributions Accessibleobject-oriented andalgorithmicdesign
76. AVMf’s Contributions Open-sourcedownload fromavmframework.org