9.3.14  Random variables: random_variable randvar

The randvar command produces an object representing a random variable. The value(s) can be generated subsequently by calling sample (see Section 9.3.1), rand (see Section 9.3.1), randvector (see Section 9.3.15) or randmatrix (see Section 9.3.16).
random_variable is a synonym for randvar.

• randvar takes a sequence of arguments:
  distspec, which can specify a probability distribution. The distributions and their arguments are:
  • Uniform distribution (see Section 9.4.2):
    Arguments:
    • uniform or uniformd.
    • Either:
      • Optionally, mean=µ, to specify a mean of µ.
      • Optionally, stddev=σ, to specify a standard deviation of σ.
      • Optionally, variance=σ², to specify a variance of σ².
      or:
      • a and b, two numbers specifying the end points of a range.
        The range can also be specified by a..b or range=a..b.
  • Binomial distribution (see Section 9.4.3):
    Arguments:
    • binomial.
    • Either:
      • n, a positive integer.
      • p, a probability (a number between 0 and 1).
      or:
      • Optionally, mean=µ, to specify a mean of µ.
      • Optionally, stddev=σ, to specify a standard deviation of σ.
      • Optionally, variance=σ², to specify a variance of σ².
  • Negative binomial distribution (see Section 9.4.4):
    Arguments:
    • negbinomial.
    • n, a positive integer.
    • p, a probability (a number between 0 and 1).
  • Multinomial distribution (see Section 9.3.4):
    Arguments:
    • randmultinomial.
    • [p₀,p₁,…,p_j], a list of probabilities with p₀+⋯ + p_j=1.
    • Optionally, [a₀,a₁,…,a_j], a list of possible return values.
  • Poisson distribution (see Section 9.4.6):
    Arguments:
    • poisson.
    • λ, a real number
      or
      mean=µ, the mean value.
  • Standard normal distribution (see Section 9.3.6):
    Argument:
    • randnorm or randNorm.
  • Normal distribution (see Section 9.4.7):
    Arguments:
    • normald.
    • Either no arguments (for the standard normal distribution) or one of:
      • Optionally, mean=µ, to specify a mean of µ.
      • Optionally, stddev=σ, to specify a standard deviation of σ.
      • Optionally, variance=σ², to specify a variance of σ².
      or:
      • a and b, two numbers specifying the end points of a range.
        The range can also be specified by a..b or range=a..b.
  • Student’s distribution (see Section 9.4.8):
    Arguments:
    • student.
    • n, an integer (the degrees of freedom).
  • χ² distribution (see Section 9.4.9):
    Arguments:
    • chisquare.
    • n, an integer (the degrees of freedom).
  • Fisher-Snédécor distribution (see Section 9.4.10):
    Arguments:
    • fisher, fisherd, or snedecor.
    • n₁ and n₂, integers (the degrees of freedom).
  • Gamma distribution (see Section 9.4.11):
    Arguments:
    • gammad.
    • a and b, real numbers.
  • Beta distribution (see Section 9.4.12):
    Arguments:
    • betaa.
    • a and b, real numbers.
  • Geometric distribution (see Section 9.4.13):
    Arguments:
    • geometric.
    • p, a probability (number between 0 and 1).
  • Cauchy distribution (see Section 9.4.14):
    Arguments:
    • cauchy or cauchyd.
    • a and b, real numbers.
  • Exponential distribution (see Section 9.4.15):
    Arguments:
    • exponential or exponentiald.
    • λ, a real number.
  • Weibull’s distribution (see Section 9.4.16):
    Arguments:
    • k, an integer.
    • λ, a real number.
  • Discrete distribution:
    Arguments:
    • W=[w₁,w₂,…,w_n], a list of nonnegative weights.
    • Optionally, V=[v₁,v₂,…,v_n], a list of values.
    or:
    • [[v₁,w₁],[v₂,w₂],…,[v_n,w_n]], a list of of object-weight pairs.
    or:
    • f, a nonnegative function.
    • a..b or range=a..b with real numbers a and b, a range specification.
    • Optionally, N, a positive integer or V=[v₀,v₁,v₂,…,v_n], a list of values with n=b−a (here a and b have to be integers).
    The weights are automatically scaled by the inverse of their sum to obtain the values of the probability mass function. If a function f is given instead of a list of weights, then w_k=f(a+k) for k=0,1,…,b−a unless N is given, in which case w_k=f(x_k) where x_k=a+(k−1) b−a/N and k=1,2,…,N. The resulting random variable X has values in {0,1,…,n−1} for 0-based modes (e.g. Xcas) resp. in {1,2…,n} for 1-based modes (e.g. Maple). If the list V of custom objects is given, then V[X] is returned instead of X. If N is given, then v_k=x_k for k=1,2,…,N.
• randvar(distspec) returns an object representing a random variable.

Examples.

• Define a random variable with a Fisher-Snedecor distribution (two degrees of freedom):
  Input:
  X:=random_variable(fisher,2,3)
  Output:

  fisherd

  ⎛ ⎝

  2,3

  ⎞ ⎠

  To generate some values of X:
  Input:
  rand(X)
  or:
  sample(X)
  Output:

  0.30584514472

  Input:
  randvector(5,X)
  or:
  sample(X,5)
  Output:

  ⎡ ⎣

  2.2652,0.1397,6.3320,1.0556,0.2995

  ⎤ ⎦

• Define a random variable with multinomial distribution:
  Input:
  M:=randvar(multinomial,[1/2,1/3,1/6],[a,b,c])
  Output:

  multinomial,

  ⎡ ⎢ ⎢ ⎣

  1 2

  ,

  1 3

  ,

  1 6

  ⎤ ⎥ ⎥ ⎦

  ,

  ⎡ ⎣

  a,b,c

  ⎤ ⎦

  Input:
  randvector(10,M)
  Output:

  ⎡ ⎣

  b,b,b,b,b,b,a,a,b,b

  ⎤ ⎦

Some continuous distributions can be defined by specifying their first and/or second moment.
Examples.

• Input:
  randvector(10,randvar(poisson,mean=5))
  Output:

  ⎡ ⎣

  7,2,5,6,7,9,8,4,3,4

  ⎤ ⎦

• Input:
  randvector(5,randvar(weibull,mean=5.0,stddev=1.5))
  Output:

  ⎡ ⎣

  1.6124,3.2720,7.02627,5.5360,3.1929

  ⎤ ⎦

• Input:
  X:=randvar(binomial,mean=18,stddev=4)
  Output:

  ⎛ ⎜ ⎜ ⎜ ⎜ ⎝ 162 1 9 ⎞ ⎟ ⎟ ⎟ ⎟ ⎠

• Input:
  X:=randvar(weibull,mean=12.5,variance=1)
  Output:

  weibulld

  ⎛ ⎝

  3.08574940721,13.9803128143

  ⎞ ⎠

  Input:
  mean(randvector(1000,X))
  Output:

  12.5728578447

• Input:
  G:=randvar(geometric,stddev=2.5)
  Output:

  geometric

  ⎛ ⎝

  0.327921561087

  ⎞ ⎠

  Input:
  evalf(stddev(randvector(1000,G)))
  Output:

  2.57913473863

• Input:
  randvar(gammad,mean=12,variance=4)
  Output:

  gammad

  ⎛ ⎝

  36,3

  ⎞ ⎠

Uniformly distributed random variables can be defined by specifying the support as an interval.
Examples:

• Input:
  randvector(5,randvar(uniform,range=15..81))
  Output:

  ⎡ ⎣

  77.0025,77.7644,63.2414,52.0707,66.3837

  ⎤ ⎦

• Input:
  rand(randvar(uniform,e..pi))
  Output:

  3.1010453504

The following examples demonstrate various ways to define a discrete random variable.
Examples:

• Input:

  X:=randvar([["apple",1/3],["orange",1/4], ["pear",1/5],["plum",13/60]]):;

  randvector(5,X)

  Output (for example):

  ⎡ ⎣

  "orange","apple","apple","plum","apple"

  ⎤ ⎦

• Input:

  W:=[1,4,5,3,1,1,1,2]:; X:=randvar(W):;

  approx(W/sum(W))

  Output (for example):

  ⎡ ⎣

  0.0556,0.2222,0.2778,0.1667,0.0556,0.0556,0.0556,0.1111

  ⎤ ⎦

• Input:
  frequencies(randvector(10000,X))
  (See Section 9.1.12.)
  Output:

  ⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣

  0 0.0527 1 0.2189 2 0.2791 3 0.1698 4 0.0546 5 0.0557 6 0.059 7 0.1102

  ⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦

• Input:

  X:=randvar(k->1-(k/10)^2,range=-10..10):;

  histogram(randvector(10000,X),-10,0.33,display=filled)

  Output:
  
• Input:

  X:=randvar([3,1,2,5],[alpha,beta,gamma,delta]):; randmatrix(5,4,X)

  Output:

  ⎛ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ δ δ β δ δ γ γ β α δ α δ α α γ α δ δ β δ ⎞ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠

Discrete random variables can be used to approximate custom continuous random variables. For example, consider a probability density function f as a mixture of two normal distributions on the support S=[−10,10]. You can sample f in N=10000 points in S.
Input:

Now generate 25000 values of X and plot a histogram:
Input:

Output:

Sampling from discrete distributions is fast: for instance, generating 25 million samples from the distribution of X which has about 10000 outcomes takes only couple of seconds. In fact, the sampling complexity is constant. Also, observe that the process isn’t slowed down by spreading it across multiple calls of randvector.
Input:

Evaluation time: 2.12

Independent random variables can be combined in an expression, yielding a new random variable. In the example below, you define a log-normally distributed variable Y from a variable X with standard normal distribution.
Input:

Output:

It is known that E[Y]=e^µ+σ2/2. The mean of L should be close to that number.
Input:

Output:

In case a compound random variable is defined as an expression containing several independent random variables X,Y,… of the same type, you sometimes need to prevent its evaluation when passing it to randvector and similar functions.

Example.
Input:

If you want to generate, for example, the random variable X/Y, you would have to forbid automatic evaluation of the latter expression; otherwise it would reduce to 1 since X and Y are both normald(0,1).
Input:

Output (for example):

To save typing, you can define Z with eval(∗,0) and pass eval(Z,1) to randvector or randmatrix.
Input:

Output (for example):

Parameters of a distribution can be entered as symbols to allow (re)assigning them at any time.
Input:

Now execute the following command line several times in a row. The parameter λ is updated in each iteration.
Input:

Output (by executing the above command line three times):