Random walks and the arcsine law

Suppose you stand at 0 and flip a fair coin. If the coin comes up heads, you take a step to the right. Otherwise you take a step to the left. How much of the time will you spend to the right of where you started?

As the number of steps N goes to infinity, the probability that the proportion of your time in positive territory is less than x approaches 2 arcsin(√x)/π. The arcsine term gives this rule its name, the arcsine law.

Here’s a little Python script to illustrate the arcsine law.

import random
from numpy import arcsin, pi, sqrt

def step():
    u = random.random()
    return 1 if u < 0.5 else -1

M = 1000 # outer loop    
N = 1000 # inner loop

x = 0.3 # Use any 0 < x < 1 you'd like.

outer_count = 0
for _ in range(M):
    n = 0
    position= 0 
    inner_count = 0
    for __ in range(N):
        position += step()
        if position > 0:
            inner_count += 1
    if inner_count/N < x:
        outer_count += 1

print (outer_count/M)
print (2*arcsin(sqrt(x))/pi)

Playing with continued fractions and Khinchin’s constant

Take a real number x and expand it as a continued fraction. Compute the geometric mean of the first n coefficients.

Aleksandr Khinchin proved that for almost all real numbers x, as n → ∞ the geometric means converge. Not only that, they converge to the same constant, known as Khinchin’s constant, 2.685452001…. (“Almost all” here mean in the sense of measure theory: the set of real numbers that are exceptions to Khinchin’s theorem have measure zero.)

To get an idea how fast this convergence is, let’s start by looking at the continued fraction expansion of π. In Sage, we can type


to get the continued fraction coefficient

[3, 7, 15, 1, 292, 1, 1, 1, 2, 1, 3, 1, 14, 2, 1, 1, 2, 2, 2, 2, 1, 84, 2, 1, 1, 15, 3]

for π to 100 decimal places. The geometric mean of these coefficients is 2.84777288486, which only matches Khinchin’s constant to 1 significant figure.

Let’s try choosing random numbers and working with more decimal places.

There may be a more direct way to find geometric means in Sage, but here’s a function I wrote. It leaves off any leading zeros that would cause the geometric mean to be zero.

from numpy import exp, mean, log
def geometric_mean(x):
    return exp( mean([log(k) for k in x if k > 0]) )

Now let’s find 10 random numbers to 1,000 decimal places.

for _ in range(10):
    r = RealField(1000).random_element(0,1)

This produced


Three of these agree with Khinchin’s constant to two significant figures but the rest agree only to one. Apparently the convergence is very slow.

If we go back to π, this time looking out 10,000 decimal places, we get a little closer:


produces 2.67104567579, which differs from Khinchin’s constant by about 0.5%.

Ellipsoid surface area

How much difference does the earth’s equatorial bulge make in its surface area?

To first approximation, the earth is a sphere. The next step in sophistication is to model the earth as an ellipsoid.

The surface area of an ellipsoid with semi-axes abc is

A = 2\pi \left( c^2 + \frac{ab}{\sin\phi} \left( E(\phi, k) \sin^2\phi + F(\phi, k) \cos^2 \phi\right)\right)




m = k^2 = \frac{a^2(b^2 - c^2)}{b^2(a^2 - c^2)}

The functions E and F are incomplete elliptic integrals

 F(\phi, k) = \int_0^\phi \frac{d\theta}{\sqrt{1 - k^2 \sin^2\theta}}


E(\phi, k) = \int_0^\phi \sqrt{1 - k^2 \sin^2\theta}\,d\theta

implemented in SciPy as ellipeinc and ellipkinc. Note that the SciPy functions take m as their second argument rather its square root k.

For the earth, a = b and so m = 1.

The following Python code computes the ratio of earth’s surface area as an ellipsoid to its area as a sphere.

from scipy import pi, sin, cos, arccos
from scipy.special import ellipkinc, ellipeinc

# values in meters based on GRS 80
# http://en.wikipedia.org/wiki/GRS_80
equatorial_radius = 6378137
polar_radius = 6356752.314140347

a = b = equatorial_radius
c = polar_radius

phi = arccos(c/a)
# in general, m = (a**2 * (b**2 - c**2)) / (b**2 * (a**2 - c**2))
m = 1

temp = ellipeinc(phi, m)*sin(phi)**2 + ellipkinc(phi, m)*cos(phi)**2
ellipsoid_area = 2*pi*(c**2 + a*b*temp/sin(phi))

# sphere with radius equal to average of polar and equatorial
r = 0.5*(a+c)
sphere_area = 4*pi*r**2


This shows that the ellipsoid model leads to 0.112% more surface area relative to a sphere.

Source: See equation 19.33.2 here.

Update: It was suggested in the comments that it would be better to compare the ellipsoid area to that of a sphere of the same volume. So instead of using the average of the polar and equatorial radii, one would take the geometric mean of the polar radius and two copies of the equatorial radius. Using that radius, the ellipsoid has 0.0002% more area than the sphere.

Number theory determinant and SymPy

Let σ(n) be the sum of the positive divisors of n and let gcd(a, b) be the greatest common divisor of a and b.

Form an n by n matrix M whose (i, j) entry is σ(gcd(i, j)). Then the determinant of M is n!.

The following code shows that the theorem is true for a few values of n and shows how to do some common number theory calculations in SymPy.

from sympy import gcd, divisors, Matrix, factorial

def f(i, j):
    return sum( divisors( gcd(i, j) ) )

def test(n):
    r = range(1, n+1)
    M = Matrix( [ [f(i, j) for j in r] for i in r] )
    return M.det() - factorial(n)

for n in range(1, 11):
    print test(n)

As expected, the test function returns zeros.

If we replace the function σ above by τ where τ(n) is the number of positive divisors of n, the corresponding determinant is 1. To test this, replace sum by len in the definition of f and replace factorial(n) by 1.

In case you’re curious, both results are special cases of the following more general theorem. I don’t know whose theorem it is. I found it here.

For any arithmetic function f(m), let g(m) be defined for all positive integers m by

g(m) = \sum_{d \,\mid \,m} \mu(d) f\left(\frac{m}{d}\right)

Let M be the square matrix of order n with ij element f(gcd(i, j)). Then

\det M = \prod_i^n g(j)

Here μ is the Möbius function. The two special cases above correspond to g(m) = m and g(m) = 1.

A prime-generating formula and SymPy

Mills’ constant is a number θ such that the integer part of θ raised to a power of 3 is always a prime. We’ll see if we can verify this computationally with SymPy.

from sympy import floor, isprime
from sympy.mpmath import mp, mpf

# set working precision to 200 decimal places
mp.dps = 200

# Mills' constant http://oeis.org/A051021
theta = mpf("1.30637788386308069046861449260260571291678458515671364436805375996643405376682659882150140370119739570729")

for i in range(1, 7):
    n = floor(theta**3**i)
    print i, n, isprime(n)

Note that the line of code defining theta wraps Mill’s constant as a string and passes it as an argument to mpf. If we just wrote theta = 1.30637788386308… then theta would be truncated to an ordinary floating point number and most of the digits would be lost. Not that I did that in the first version of the code. :)

This code says that the first five outputs are prime but that the sixth is not. That’s because we are not using Mills’ constant to enough precision. The integer part of θ^3^6 is 85 digits long, and we don’t have enough precision to compute the last digit correctly.

If we use the value of Mills’ constant given here to 640 decimal places, and increase mp.dps to 1000, we can correctly compute the sixth term, but not the seventh.

Mill’s formula is just a curiosity with no practical use that I’m aware of, except possibly as an illustration of impractical formulas.

Relating Airy and Bessel functions

The Airy functions Ai(x) and Bi(x) are independent solutions to the differential equation

y'' - xy = 0

For negative x they act something like sin(x) and cos(x). For positive x they act something like exp(x) and exp(-x). This isn’t surprising if you look at the differential equation. If you replace x with a negative constant, you get sines and cosines, and if you replace it with a positive constant, you get positive and negative exponentials.

The Airy functions can be related to Bessel functions as follows:

\mathrm{Ai}(x) = \left\{ 	\begin{array}{ll} 		\frac{1}{3}\sqrt{\phantom{-}x} \left(I_{-1/3}(\hat{x}) - I_{1/3}(\hat{x})\right) & \mbox{if } x > 0 \\<br /><br /><br /><br /><br />
                \\<br /><br /><br /><br /><br />
		\frac{1}{3}\sqrt{-x} \left(J_{-1/3}(\hat{x}) + J_{1/3}(\hat{x})\right) & \mbox{if } x < 0 	\end{array} \right.


\mathrm{Bi}(x) = \left\{ 	\begin{array}{ll} 		\sqrt{\phantom{-}x/3} \left(I_{-1/3}(\hat{x}) + I_{1/3}(\hat{x})\right) & \mbox{if } x > 0 \\<br />                 \\<br /> 		\sqrt{-x/3} \left(J_{-1/3}(\hat{x}) - J_{1/3}(\hat{x})\right) & \mbox{if } x < 0 	\end{array} \right.

Here J is a “Bessel function of the first kind” and I is a “modified Bessel function of the first kind.” Also

\hat{x} = \frac{2}{3} \left(\sqrt{|x|}\right)^3

To verify the equations above, and to show how to compute these functions in Python, here’s some code.

The SciPy function airy computes both functions, and their first derivatives, at once. I assume that’s because it doesn’t take much longer to compute all four functions than to compute one. The code for Ai2 and Bi2 below uses np.where instead of if ... else so that it can operate on NumPy vectors all at once. You can plot Ai and Ai2 and see that the two curves lie on top of each other. The same holds for Bi and Bi2.


from scipy.special import airy, jv, iv
from numpy import sqrt, where

def Ai(x):
    (ai, ai_prime, bi, bi_prime) = airy(x)
    return ai

def Bi(x):
    (ai, ai_prime, bi, bi_prime) = airy(x)
    return bi

def Ai2(x):
    third = 1.0/3.0
    hatx = 2*third*(abs(x))**1.5
    return where(x > 0,
        third*sqrt( x)*(iv(-third, hatx) - iv(third, hatx)),
        third*sqrt(-x)*(jv(-third, hatx) + jv(third, hatx)))

def Bi2(x):
    third = 1.0/3.0
    hatx = 2*third*(abs(x))**1.5
    return where(x > 0,
        sqrt( x/3.0)*(iv(-third, hatx) + iv(third, hatx)),
        sqrt(-x/3.0)*(jv(-third, hatx) - jv(third, hatx)))


There is a problem with Ai2 and Bi2: they return nan at 0. A more careful implementation would avoid this problem, but that’s not necessary since these functions are only for illustration. In practice, you’d simply use airy and it does the right thing at 0.

Related links:

SymPy book

There’s a new book on SymPy, a Python library for symbolic math.

The book is Instant SymPy Starter by Ronan Lamy. As far as I know, this is the only book just on SymPy. It’s only about 50 pages, which is nice. It’s not a replacement for the online documentation but just a quick start guide.

The online SymPy documentation is good, but I think it would be easier to start with this book. And although I’ve been using SymPy off and on for a while, I learned a few things from the book.


Need a 12-digit prime?

You may have seen the joke “Enter any 12-digit prime number to continue.” I’ve seen it floating around as the punchline in several contexts.

So what do you do if you need a 12-digit prime? Here’s how to find the smallest one using Python.

>>> from sympy import nextprime
>>> nextprime(10**11)

The function nextprime gives the smallest prime larger than its argument. (Note that the smallest 12-digit number is 1011, not 1012. Great opportunity for an off-by-one mistake.)

Optionally you can provide an addition argument to nextprime to get primes further down the list. For example, this gives the second prime larger than 1011.

>>> nextprime(10**11, 2)

What if you wanted the largest 12-digit prime rather than the smallest?

>>> from sympy import prevprime
>>> prevprime(10**12)

Finally, suppose you want to know how many 12-digit primes there are. SymPy has a function primepi that returns the number of primes less than its argument. Unfortunately, it fails for large arguments. It works for arguments as big as 2**27 but throws a memory error for 2**28.

The number of primes less than n is approximately n / log n, so there are about 32 billion primes between 1011 and 1012. According to Wolfram Alpha, the exact number of 12-digit primes is 33,489,857,205. So if you try 12-digit numbers at random, your chances are about 1 in 30 of getting a prime. If you’re clever enough to just pick odd numbers, your chances go up to 1 in 15.

Synchronizing cicadas with Python

Suppose you want to know when your great-grandmother was born. You can’t find the year recorded anywhere. But you did discover an undated letter from her father that mentions her birth and one curious detail:  the 13-year and 17-year cicadas were swarming.

You do a little research and find that the 13-year cicadas are supposed to come out next year, and that the 17-year cicadas came out last year. When was your great-grandmother born?

Since 13 and 17 are relatively prime, the 13-year and 17-year cicadas will synchronize their schedules every 13 × 17 = 221 years. Suppose your great-grandmother was born n years ago. The remainder when n is divided by 13 must be 12, and the remainder when n is divided by 17 must be 1. We have to solve the pair of congruences n = 12 mod 13 and n = 1 mod 17. The Chinese Remainder Theorem says that this pair of congruences has a solution, and the proof of the theorem suggests an algorithm for computing the solution.

The Python library SymPy has a function for solving linear congruences.

>>> from sympy.ntheory.modular import solve_congruence
>>> solve_congruence( (12, 13), (1, 17) )
(103, 221)

This says we’re 103 years into the joint cycle of the 13-year and 17-year cicadas. So your great-grandmother was born 103 years ago. (The solution 324 = 103 + 221 is also mathematically possible, but not biologically possible.)

You can use the same SymPy function to solve systems of more congruences. For example, when is the next year in which there will be summer Olympic games and the 13-year and and 17-year cicadas will swarm? Here are a couple ways to approach this. First, you could find the last time this happened, then find when it will happen next. You’d need to solve n = 1 mod 4 (since we had summer Olympics last year) and  n = 12 mod 13 and n = 1 mod 17.

>>> solve_congruence( (1, 4), (12, 13), (1, 17) )
(545, 884)

So the cicadas and the summer Olympics were in sync 545 years ago. (Well, they would have been if the modern Olympics had existed in the middle ages.) This says they’ll be in sync again in 885 – 545 = 339 years.

Here’s a more direct approach. We want to know when the summer Olympics will be 3 years ahead of where they are now in the cycle, when the 13-year cicadas will be 1 year ahead, and the 17-year cicadas will be 16 years ahead.

>>> solve_congruence( (3, 4), (1, 13), (16, 17) )
(339, 884)

By the way, you can use negative integers with the congruences, so you could have used (-1, 17) to say the 17-year cicadas will be 1 year back instead of 16 years ahead in their cycle.

Recognizing numbers

I was playing around with SymPy, a symbolic math package for Python, and ran across nsimplify. It takes a floating point number and tries to simplify it: as a fraction with a small denominator, square root of a small integer, an expression involving famous constants, etc.

For example, suppose some calculation returned 4.242640687119286 and you suspect there’s something special about that number. Here’s how you might test where it came from.

>>> from sympy import *
>>> nsimplify(4.242640687119286)

Maybe you do a calculation numerically, find a simple expression for the result, and that suggests an analytical solution.

I think a more common application of nsimplify might be to help you remember half-forgotten formulas. For example, maybe you’re rusty on your trig identities, but you remember that cos(π/6) is something special.

>>> nsimplify(cos(pi/6))

Or to take a more advanced example, suppose that you vaguely remember that the gamma function takes on recognizable values at half integer values, but you don’t quite remember how. Maybe something involving π or e. You can suggest that nsimplify include expressions with π and e in its search.

>>> nsimplify(gamma(3.5), constants=[pi, E])

You can also give nsimplify a tolerance, asking it to find a simple representation within a neighborhood of the number. For example, here’s a way to find approximations to π.

>>> nsimplify(pi, tolerance=1e-5)

With a wider tolerance, it will return a simpler approximation.

>>> nsimplify(pi, tolerance=1e-2)

Finally, here’s higher precision approximation to π that isn’t exactly simple:

>>> nsimplify(pi, tolerance=1e-7)
exp(141/895 + sqrt(780631)/895)

Rolling dice for normal samples: Python version

A handful of dice can make a decent normal random number generator, good enough for classroom demonstrations. I wrote about this a while ago.

My original post included Mathematica code for calculating how close to normal the distribution of the sum of the dice is. Here I’d like to redo the code in Python to show how to do the same calculations using SymPy. [Update: I’ll also give a solution that does not use SymPy and that scales much better.]

If you roll five dice and add up the spots, the probability of getting a sum of k is the coefficient of xk in the expansion of

(x + x2 + x3 + x4 + x5 + x6)5 / 65.

Here’s code to find the probabilities by expanding the polynomial and taking coefficients.

from sympy import Symbol

sides = 6
dice = 5
rolls = range( dice*sides + 1 )

# Tell SymPy that we want to use x as a symbol, not a number
x = Symbol('x')

# p(x) = (x + x^2 + ... + x^m)^n
# where m = number of sides per die
# and n = number of dice
p = sum([x**i for i in range(1, sides + 1)])**dice

# Extract the coefficients of p(x) and divide by sides**dice
pmf = [sides**(-dice) * p.expand().coeff(x, i) for i in rolls]

If you’d like to compare the CDF of the dice sum to a normal CDF you could add this.

from scipy import array, sqrt
from scipy.stats import norm

cdf = array(pmf).cumsum()

# Normal CDF for comparison
mean = 0.5*(sides + 1)*dice
variance = dice*(sides**2 -1)/12.0
temp = [norm.cdf(i, mean, sqrt(variance)) for i in roles]
norm_cdf = array(temp)

diff = abs(cdf - norm_cdf)
# Print the maximum error and where it occurs
print diff.max(), diff.argmax()

Question: Now suppose you want a better approximation to a normal distribution. Would it be better to increase the number of dice or the number of sides per dice? For example, would you be better off with 10 six-sided dice or 5 twelve-sided dice? Think about it before reading the solution.

Update: The SymPy code does not scale well. When I tried the code with 50 six-sided dice, it ran out of memory. Based on Andre’s comment, I rewrote the code using polypow. SymPy offers much more symbolic calculation functionality than NumPy, but in this case NumPy contains all we need. It is much faster and it doesn’t run out of memory.

from numpy.polynomial.polynomial import polypow
from numpy import ones

sides = 6
dice = 100

# Create an array of polynomial coefficients for
# x + x^2 + ... + x^sides
p = ones(sides + 1)
p[0] = 0

# Extract the coefficients of p(x)**dice and divide by sides**dice
pmf = sides**(-dice) * polypow(p, dice)
cdf = pmf.cumsum()

That solution works for up to 398 dice. What’s up with that? With 399 dice, the largest polynomial coefficient overflows. If we divide by the number of dice before raising the polynomial to the power dice, the code becomes a little simpler and scales further.

p = ones(sides + 1)
p[0] = 0
p /= sides
pmf = polypow(p, dice)
cdf = pmf.cumsum()

I tried this last approach on 10,000 dice with no problem.

Moments of mixtures

I needed to compute the higher moments of a mixture distribution for a project I’m working on. I’m writing up the code here in case anyone else finds this useful. (And in case I’ll find it useful in the future.) I’ll include the central moments first. From there it’s easy to compute skewness and kurtosis.

Suppose X is a mixture of n random variables Xi with weights wi, non-negative numbers adding to 1. Then the jth central moment of X is given by

E[(X - \mu)^j] = \sum_{i=1}^n \sum_{k=0}^j {j \choose k} (\mu_i - \mu)^{j-k} w_i E[(X_i- \mu_i)^k]

where μi is the mean of Xi.

In my particular application, I’m interested in a mixture of normals and so the code below computes the moments for a mixture of normals. It could easily be modified for other distributions.

from scipy.misc import factorialk, comb

def mixture_central_moment(mixture, moment):

    '''Compute the higher moments of a mixture of normal rvs.
    mixture is a list of (mu, sigma, weight) triples.
    moment is the central moment to compute.'''

    mix_mean = sum( [w*m for (m, s, w) in mixture] )

    mixture_moment = 0.0
    for triple in mixture:
        mu, sigma, weight = triple
        for k in range(moment+1):
            prod = comb(moment, k) * (mu-mix_mean)**(moment-k)
            prod *= weight*normal_central_moment(sigma, k)
            mixture_moment += prod

    return mixture_moment

def normal_central_moment(sigma, moment):

    '''Central moments of a normal distribution'''

    if moment % 2 == 1:
        return 0.0
        # If Z is a std normal and n is even, E(Z^n) == (n-1)!!
        # So E (sigma Z)^n = sigma^n (n-1)!!
        return sigma**moment * factorialk(moment-1, 2)

Once we have code for central moments, it’s simple to add code for computing skewness and kurtosis.

def mixture_skew(mixture):

    variance = mixture_central_moment(mixture, 2)
    third = mixture_central_moment(mixture, 3)
    return third / variance**(1.5)

def mixture_kurtosis(mixture):

    variance = mixture_central_moment(mixture, 2)
    fourth = mixture_central_moment(mixture, 4)
    return fourth / variance**2 - 3.0

Here’s an example of how the code might be used.

# Test on a mixture of 30% Normal(-2, 1) and 70% Normal(1, 3)
mixture = [(-2, 1, 0.3), (1, 3, 0.7)]

print "Skewness = ", mixture_skew(mixture)
print "Kurtosis = ", mixture_kurtosis(mixture)

Related post: General formula for normal moments

New introduction to SciPy

The Python stack for scientific computing is more modular than say R or Mathematica. Python is a general-purpose programming language that has libraries for scientific computing. R and Mathematica are statistical and mathematical programming languages that have general-purpose features. The Python approach has its advantages — I’d rather do math in a general language than do general programming in a mathematical language — but it takes longer to learn. The components of the Python stack work well together, but someone new to Python has to discover what components they’ll need.

Several books have come out recently to help someone learn Python and the components for numerical computing. The latest is Learning SciPy for Numerical and Scientific Computing by Francisco J. Blanco-Silva.

This book covers the things you’d expect, including SciPy, NumPy, and Matplotlib. The only exception may be IPython. But no book can cover everything. And since IPython is an environment more than a library, it makes some sense to leave it out.

In addition to the usual topics, the book includes several important topics that are not as commonly covered. For example, it devotes a good amount of space to special functions and integration in a chapter on numerical analysis. I share the author’s assessment that this is “one of the most interesting chapters in the book.”

There are three chapters on more specific applications: signal processing, data mining, and computational geometry. These chapters give an introduction to their topics as well as how to carry out computations in SciPy.

The final chapter of the book is on integration with other programming languages.

Learning SciPy for Numerical and Scientific Computing covers the material you need to get going. It’s easy to read, and still fairly small: 150 pages total, about 130 pages of content. This is the right size for such a book in my opinion. There’s plenty of online documentation for more details, but it helps to have a good overview such as this book before diving into reference material.

Suffix primes

MathUpdate tweeted this afternoon that

Any number made by removing the first n digits of 646216567629137 is still prime.

and links to sequence A012885 in the Online Encyclopedia of Integer Sequences (OEIS). The OEIS heading for the sequence is

Suffixes of 357686312646216567629137 (all primes)

which implies you can start with an even larger number, cutting off the first digit each time and producing a sequence of primes.

The following Python code verifies that this is indeed the case.

    from sympy.ntheory import isprime

    x = "357686312646216567629137"

    while x:
        print isprime(int(x))
        x = x[1:]

Update: lucio wrote a program to show that the prime given here is the longest one with the suffix property.

    def extend_prime(n, result):
        for i in range(10):
            nn = int(str(i) + str(n))
            if nn == n: continue
            if isprime(nn):
                extend_prime(nn, result)
        return result        

    print "Max Prefix Prime:", max(extend_prime("", []))

One minor suggestion: by using range(1, 10) rather than range(10) above, i.e. eliminating 0, the line if nn == n: continue could be eliminated.

Instead of calling max, you could call len to find that there are 4260 suffix primes.

Here’s a list of all suffix primes created by the code above and sorting the output.

Other novelty primes: