Page - 54 - in Programming for Computations – Python - A Gentle Introduction to Numerical Simulations with Python 3.6, Volume Second Edition

54 2 AFewMoreSteps 2.4 RandomNumbers Programming languages usually offer ways to produce (apparently) random num- bers, referred to as pseudo-randomnumbers. These numbers are not truly random, since theyareproducedinapredictablewayoncea“seed”hasbeenset (theseed is anumber,whichgenerationdependson the current time). Drawing One Random Number at a Time Pseudo-random numbers come in handy if your code is to deal with phenomena characterized by some randomness. For example, your code could simulate a throw of dice by generating pseudo- random integers between 1 and 6. A Python program (throw_2_dice.py) that mimics the throwof two dicecould read import random a = 1; b = 6 r1 = random.randint(a, b) # first die r2 = random.randint(a, b) # second die print(’The dice gave: {:d} and {:d}’.format(r1, r2)) The function randint is available from the imported module random, which is part of the standard Python library, and returns a pseudo-random integer on the interval [a,b], a ≤ b. Each number on the interval has equal probability of being picked. It should be clear that, when numbers are generated pseudo- randomly, we can not tell in advance what numbers will be produced (unless we happen to have detailed knowledge about the number generation process). Also, runningthecode twice,generallygivesdifferent results, asyoumightconfirmwith throw_2_dice.py. Note that, since the seed depends on the current time, this applieseven ifyourestart yourcomputer inbetween the tworuns. When debugging programs that involve pseudo-random number generation, it is a great advantage to fix the seed, which ensures that the very same sequence of numbers will be generated each time the code is run. This simply means that you pick the seed yourself and tell Python what that seed should be. For our little programthrow_2_dice.py, we could choose, e.g., 10 as our seed and insert the line random.seed(10) after theimportstatement(andbeforerandint iscalled).Test thismodificationand confirmthat it causeseachrun toprint thesame twonumberswitheveryexecution. In fact, it is a good idea to fix the seed from the outset when you write the program.Later,when(youthink) itworksforafixedseed,youchangeit so that the numbergeneratorsets its ownseed, afterwhichyouproceedwith further testing. Analternative tothrow_2_dice.py, couldbe touse Python interactivelyas In [1]: import random In [2]: random.randint(1, 6) Out[2]: 6