Seeing Further (39 page)

Reed–Solomon codes are a typical example. These are the codes that NASA used to detect and correct potential errors in the Rovers’ images of the Martian surface as they were beamed across the vastness of the solar system to planet Earth. More familiar devices, such as CD players, also would not work without Reed–Solomon codes. These codes hinge on, and were motivated by, the algebraic legacy of Boole and Galois. They transform the digital data that represents music in a way that makes it easy to spot, and put right, any errors that occur when the CD is being played. Virtually all of today’s digital communications are wholly reliant on sophisticated and very modern mathematical coding methods. None of it would work without them. And that turns out to be just the tip of a very large iceberg.

A few weeks ago I looked through a randomly chosen issue of
New Scientist
magazine. Of the fifty or so stories reported, there were a dozen that – to my sensitive eye – involved a significant amount of mathematics. Not one story mentioned this, though a few hinted about ‘models’ of the process under study. When the contribution of mathematics is hidden that far behind the scenes, it is hardly surprising that the media and the public have little idea of what mathematics is, or what it is good for.

Sometimes mathematics should be kept behind the scenes. When I listen to music in my car, I don’t want to have to think about the intricacies of Galois fields. When NASA engineers are firing a space probe’s rockets to nudge it into the right entry trajectory to prevent it burning up in the Martian atmosphere, they don’t want to be worrying about differential
equations. But
someone
has to do the sums, write the program, design the algorithm, invent the concept, or prove the theorem. Someone has to provide the tools for the job and make sure they are reliable. If neither the media, nor the public, nor even practising scientists realise that this hidden mathematics exists, we will stop training mathematicians, and the necessary people will cease to exist too.

To most of us, ‘mathematics’ is something we did at school, and promptly forgot. Curiously, many of us also think that what we did at school was
the whole of mathematics:
all done and dusted. And pointless, now that we’ve got computers to do the sums for us. Some of us discover there is more to it than that. Some go on to university, take a science degree, and come to grips with statistics (in biology or medicine), differential equations (physics and engineering), or mathematical logic (computer science). And the mental picture that we get is that there’s a certain amount of genuinely
useful
stuff (statistics, differential equations, mathematical logic …) plus a lot of highbrow intellectual fun and games that never has been and never will be useful to anyone living in the ‘real world’.

Both of these views of mathematics are caricatures; real mathematics is quite different. Today’s mathematics is intimately bound up with two key areas of human knowledge and activity: the natural world, and the society in which we live. Human understanding of our planet, and our universe, rests heavily on the shoulders of mathematics. So does the day-to-day working of our world. Take the hidden mathematics away, and today’s world would fall to pieces. That statement applies to a lot of the apparently esoteric parts of the subject, as well as the more obviously applicable ones – partly because mathematics is an interconnected whole, but also because the esoteric concepts are often very general and very powerful. New and unexpected applications are common.

The ‘classical’ areas of mathematics are mainly those that led up to, or developed from, calculus – continuous mathematics, where everything can be subdivided into pieces that are as tiny as you wish. Most core mathematical
physics and classical applied mathematics, such as acoustics or aerodynamics or elasticity theory, are of this kind. An important newcomer is discrete mathematics, which is suited to the digital age. Here the basic ingredients come in indivisible packets; essentially, anything whose natural description uses whole numbers or finite lists of symbols. Straddling both areas is the theory of probability, a mathematical description of uncertainty.

Geometry is also crucial. Despite appearances to the contrary, mathematics is primarily visual, and the formal symbolism tends to be closely related to some kind of mental image. Today’s geometric thinking, however, takes a variety of forms, few of them resembling the traditional geometry of Euclid. Modern mathematics rightly places value on generality, when appropriate. That naturally leads to a degree of abstraction, because the focus of attention has to shift from ‘what objects are we looking at?’ to ‘what properties are we assuming?’ Logical proof remains central to the enterprise; it’s how mathematicians keep themselves and their subject honest. Computers now play an increasingly central role. They seldom solve problems without further thought, but they can create a huge improvement in our understanding when they are used intelligently.

Mathematics, embodied in digital devices, has made technologies possible that seem to verge on magic. In February 2008 my wife and I spent two weeks exploring the private tombs of the Egyptian nobility, from Cairo down to Aswan. We took more than 1,400 photographs with two digital cameras; the whole lot were recorded on three 1-gigabyte memory cards, each the size of a postage stamp. The engineering feats involved are amazing, and they rest on all sorts of advances in materials science, photolithography, even quantum mechanics. Those advances required a lot of mathematics, as it happens, but I want to focus on just one aspect of digital cameras: data compression. The quantity of raw information required to specify 1,400 high-resolution colour pictures is far larger than those three cards can hold. Despite huge advances in miniaturisation, you simply cannot get that amount of data into such a small space.

Yet the pictures exist. I can print them out, or put them on the computer screen. How do the camera manufacturers cram so much information into so little memory? It may seem like magic, but the magic is mostly invisible mathematics. The clue lies in the names of the image files, which on my camera look something like P1000565.JPG. This tells the computer that the file is formatted using the JPEG standard, issued by the Joint Photographic Experts Group in 1992. This format uses various features of human vision, and typical images, to ‘compress’ the image data substantially.

In general terms, a computer represents a picture as a list of numbers. The list represents a rectangular array of tiny picture elements, called pixels, and the numbers describe the colour and the brightness of each pixel. If you do the sums, however, you find that there’s nowhere near enough space in a memory card to hold all the pictures that undeniably are in there. It’s not just like trying to get a quart into a pint pot: more like getting a tanker-load of milk into a pint pot.

This problem is a common one in the digital world, and it is usually tackled by compressing the data – reducing the quantity of information while retaining enough of it to do the job. Just as you can get more luggage into the car if you load it in the right way, so you can get more of the important data into a computer file if you leave out stuff that’s not really relevant, or take advantage of certain inbuilt redundancies. For instance, many photographs have a large area of blue sky. Instead of repeating the code for ‘blue’ thousands of times, once for each pixel, we could tell the computer ‘colour everything in this rectangle blue’, and specify the rectangle by listing its corners. Suddenly thousands of numbers collapse to a few dozen. That’s not how JPEG works, but it shows how redundancies in a list of numbers may make the list compressible. The actual procedure is carefully tailored to what can be done efficiently inside a small camera. The details don’t really matter for my main point, but I want you to appreciate that there
are
details, which use several different mathematical ideas. So please indulge me while I tell you just how cunning the process is.

JPEG starts by splitting the data into three separate arrays. One lists how bright each pixel is. The other two take advantage of the fact that the colours perceived by the eye can be specified as points in a plane, the ‘colour triangle’. A plane is two-dimensional, so each point can be defined using just two numbers, its horizontal and vertical coordinates. These ‘colour components’ form the other two lists. The human eye is more sensitive to variations in brightness than in colour, so the two lists of colour components can be shortened – usually they are reduced to one quarter of their original size – by using a coarser list of colours.

The next step uses a trick introduced by the French mathematician Joseph Fourier in 1824 – a year after his election to the Royal Society, as it happens – who at the time was working on the flow of heat. In general terms, Fourier’s idea was to represent a pattern of numbers by combining specific patterns with different frequencies – much as the note played by a clarinet is made up from a fundamental ‘pure’ note and various higher-pitched ‘harmonics’, all added together in suitable proportions. JPEG uses a similar trick for spatial patterns of numbers, treating each of its three arrays in the same way. First, the array is broken up into 8x8 blocks of pixels. Then each block is transformed into a list of its spatial frequencies in the horizontal and vertical directions. Roughly, this splits the pattern into black-and-white stripes of various thicknesses, and works out how much of each stripe you need to reconstruct the actual image. This step employs a fast Fourier transform, exploiting number-theoretic features of binary numerals to speed up a difficult computation; this is why 8x8 blocks are used, eight being a power of two. The Fourier transform does not compress the data, but rewrites it in a compressible form. The eye is fairly insensitive to high-frequency stripes, so these can be ignored. Medium-frequency stripes can be specified using smaller numbers, which occupy less space on the memory card.

This is not the end: two more tricks are used to squash even more pictures into the same space. If you run through the resulting array of
numbers in a zigzag order, from low frequency components to high ones, you typically find runs of repeated numbers, such a
s 7 7 7 7 7 7 7 7 7.
Coding this as ‘9 consecutive 7’s’ converts it to 9 7, which is shorter. Finally, another coding method called Huffman coding is used on the resulting file, which compresses it even further.

So JPEG coding is quite complex, with sophisticated mathematical features. You don’t need to know how it works to use your digital camera, but without the underlying ideas, that camera could never have been made. Now think of future developments, video cameras, cramming a camera into a mobile phone along with dozens of other applications … We desperately need people who can understand that sort of mathematics.

At any rate, my wife and I were able to take lots of pictures without carrying sacks full of film because a lot of mathematically sophisticated engineers noticed that something that a nineteenth-century Frenchman invented for a completely different reason happened to have an unexpected use. But the hidden mathematics behind our holiday didn’t stop there. Without a lot of other mathematics, often with similarly impractical or outmoded origins, we could never have got to Egypt to take the pictures.

Our flight was booked over the Internet and all Internet communications rely on error-correcting codes to ensure that messages are not garbled along the way by electrical interference. Like the codes used by the Mars Rovers, these techniques rely heavily on abstract algebra. The airline’s schedules were designed using mathematical methods to improve efficiency – graph theory and linear algebra. Then there was radar, weather-forecasting, even the statistical analysis of different breeds of vegetables that governed the crops from which the airline food was made.

None of this is much use if the aircraft never gets to its intended destination. In the early days of navigation, when the great European explorers were mapping the globe in small wooden sailing ships, navigation was a major consumer of mathematics. Even finding the size and shape of the Earth involved mathematical calculations, as well as experimental observations.
Today we have GPS, the Global Positioning System, which comprises about fifteen satellites orbiting the Earth, sending out signals. A triumph of electronics and engineering, obviously. But mathematics?

Leaving aside the heavy use of mathematics in designing and building launch vehicles and satellites, and in calculating orbital dynamics, let me focus solely on the signalling system that GPS uses. Each satellite transmits a signal, which can be used to work out how far away the satellite is from the GPS receiver (on board the aeroplane, ship, car, yacht, or inside someone’s mobile phone). These distances, coupled with knowledge of the positions of the satellites, make it possible to calculate the location of the receiver on the surface of the Earth. That’s another highly mathematical step, which I will also ignore.

How do the signals convey distances?

Imagine that the satellite is playing a tune, and that you have access to a second ‘copy’ of that tune, being sent out from a known source that is in synchrony with the satellite. Because the satellite is further away than the reference source, the signal from the satellite is slightly delayed, by a time equal to the difference in distances divided by the speed of light. The time delay can be measured, very accurately, and the distance is obtained by multiplying that by the speed of light.

Instead of tunes, the signals are sequences of pseudo-random numbers – apparently patternless sequences generated by a fixed mathematical recipe. Both the satellite and the reference source know this recipe, so they can generate and recognise the same signals. So here we find a very practical application of the mathematics of pseudo-random numbers. If you use SatNav in your car, you are a major consumer of the hidden mathematics that runs our world.

Still pursuing the hidden mathematics that made my holiday possible, there is the small matter of designing an aircraft that stays up, one of the heaviest uses of mathematics in the whole enterprise. Nearly all of the analysis of airflow past an aircraft nowadays is done using ‘numerical wind-tunnels’, which are mathematical simulations. They are much easier to use than physical wind-tunnels, and if anything, more accurate. They have innumerable other applications. They are essential to the design of Formula 1 and NASCAR racing cars, where effective aerodynamics is needed to keep the car on the track and reduce air resistance. If that’s not green enough for you, the same techniques improve the fuel efficiency of ordinary road vehicles. Even the dynamics of a football has been analysed mathematically, with useful practical implications about how to make the ball behave unpredictably, which can help it get past the keeper into the goal. Computational Fluid Dynamics also has medical applications to blood flow and heart disease.