Imagine the following scenario. The end of civilisation has occurred, zombies have taken over the Earth and all access to modern technology has ended. The few survivors suddenly need to know the value of π and, being a mathematician, they turn to you. What do you do?
If ever you find yourself in this situation, you’ll be glad of the work of Vincent Dumoulin and Félix Thouin at the Université de Montréal in Canada. These guys have worked out how to calculate an approximate value of π using the distribution of pellets from a Mossberg 500 pump-action shotgun, which they assume would be widely available in the event of a zombie apocalypse.
The principle is straightforward. Imagine a square with sides of length 1 and which contains an arc drawn between two opposite corners to form a quarter circle. The area of the square is 1 while the area of the quarter circle is π/4.
Next, sprinkle sand or rice over the square so that it is covered with a random distribution of grains. Then count the number of grains inside the quarter circle and the total number that cover the entire square.
The ratio of these two numbers is an estimate of the ratio between the area of the quarter circle and the square, in other words π/4.
So multiplying this ratio by 4 gives you π, or at least an estimate of it. And that’s it.
This technique is known as a Monte Carlo approximation (after the casino where the uncle of the physicist who developed it used to gamble). And it is hugely useful in all kinds of simulations.
Of course, the accuracy of the technique depends on the distribution of the grains on the square. If they are truly random, then a mere 30,000 grains can give you an estimate of π which is within 0.07 per cent of the actual value.
Dumoulin and Thouin’s idea is to use the distribution of shotgun pellets rather than sand or rice (which would presumably be in short supply in the post-apocalyptic world). So these guys set up an experiment consisting of a 28-inch barrel Mossberg 500 pump-action shotgun aimed at a sheet of aluminium foil some 20 metres away.
They loaded the gun with cartridges composed of 3 dram equivalent of powder and 32 grams of #8 lead pellets. When fired from the gun, these pellets have an average muzzle velocity of around 366 metres per second.
Dumoulin and Thouin then fired 200 shots at the aluminium foil, peppering it with 30,857 holes. Finally, they used the position of these holes in the same way as the grains of sand or rice in the earlier example, to calculate the value of π.
They immediately have a problem, however. The distribution of pellets is influenced by all kinds of factors, such as the height of the gun, the distance to the target, wind direction and so on. So this distribution is not random.
To get around this, they are able to fall back on a technique known as importance sampling. This is a trick that allows mathematicians to estimate the properties of one type of distribution while using samples generated by a different distribution.
Of their 30,000 pellet holes, they chose 10,000 at random to perform this estimation trick. They then use the remaining 20,000 pellet holes to get an estimate of π, safe in the knowledge that importance sampling allows the calculation to proceed as if the distribution of pellets had been random.
The result? Their value of π is 3.131, which is just 0.33 per cent off the true value. “We feel confident that ballistic Monte Carlo methods constitute reliable ways of computing mathematical constants should a tremendous civilization collapse occur,” they conclude.
Quite! Other methods are also available.
Ref: arxiv.org/abs/1404.1499 : A Ballistic Monte Carlo Approximation of π