Non Sequitur: An exploration of Python's random module

An exploration of Python's random module for the curious programmer, this talk will give a little background in statistics and pseudorandom number generation, explain the properties of python's choice of pseudorandom generator and explore through visualizations the different distributions provided by the module.

Transcript

1. Non Sequitur An exploration of Python's random module Jair Trejo, EuroPython 2014
2. Jair Trejo Director of Operations at Vinco Orbis
8. Random sample
9. Simulation
10. 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048 552433147702632768969101985048
11. 14159265358979323846 26433832795028841971 69399375105820974944 59230781640628620899 86280348253421170679
12. 4.77 Counter({ '9': 14, '2': 12, '8': 12, '3': 11, '4': 10, '6': 9, '1': 8, '0': 8, '5': 8, '7': 8})
16. 14159265358979323846 26433832795028841971 69399375105820974944 59230781640628620899 86280348253421170679
17. 3.1415926535897932384 626433832795028841971 693993751058209749445 923078164062862089986 280348253421170679
23. 7638, 3390, 4921, 2162, 6742, 4545, 6570, 1649, 7192, 7248, 5335, 4622, 3628, 1623, 6341, 2082, 3347, 2024, 965, 3122, 7468, 7710, 4441, 7224, 1861, 4633, 4646, 5853, 2576, 6357, 4114, 9249, 5440, 5936, 2360, 5696, 4444, 7491, 1150, 3225, 4006, 480, 3040, 2416, 8370, 569, 2376, 6453, 6412, 1137, 2927, 5673, 1829, 3452, 9163, 9605, 2560, 5536, 6472, 8867, 6236, 8876, 7833, 3558, 6593, 4676, 8649, 8052, 8347, 6724, 2121, 4986, 8601, 9772, 4919, 1965, 8612, 1665, 7722, 6292, 5892, 7156, 2083, 3388, 4785, 8962, 3174, 742, 5056, 5631, 7081, 1405, 9740, 8676, 2729, 4474, 166, 2755, 5900, 8100
26. 7, 23, 41, 59, 89
36. Python’s random module source code
39. import random
40. r = random.Random(seed) r.random()
41. r.randrange(max) r.random() * max
42. r.randrange(start, stop) start + r.randrange(stop - start)
43. r.randrange(start, stop, step)
44. r.random() r.randrange(start, stop, step) r.randint(a, b)
45. r.choice(a_sequence)
46. r.sample(population, how_many)
47. r.shuffle(a_list)
48. sorted(a_list, key=lambda e: r.random())
54. r.triangular r.gammavariate r.betavariate r.paretovariate r.weibullvariate
56. import random ! random.random()
57. random.WichmannHill random.SystemRandom randomdotorg.RandomDotOrg
58. Conclusions • The definition of randomness is more a philosophical than a mathematical problem. • But we can use mathematical definitions that are useful for our purposes. • If we need sequences that are deterministic, but behave as if random, we can use pseudo-random number generation. • If we need numbers that are completely unpredictable, we need sources of entropy like input devices, noise measurements or other external sources. • For most of our random number needs, python provides more than adequate capabilities.
59. References
60. 10 PRINT CHR$(205.5+RND(1)); : GOTO 10 By Nick Montfort, Patsy Baudoin, John Bell, Ian Bogost, Jeremy Douglass, Mark C. Marino, Michael Mateas, Casey Reas, Mark Sample and Noah Vawter
61. The art of computer programming, Volume 2 Seminumerical Algorithms
62. On randomness in cryptography! http://blog.cloudflare.com/why-randomness-matters On random number generators testing! http://www.fourmilab.ch/hotbits/statistical_testing/stattest.html On the possible NSA backdoor into RSA’s random number generator! http://arstechnica.com/security/2014/01/how-the-nsa-may-have-put-a-backdoor rsas-cryptography-a-technical-primer/
63. Thank you!