COMPUTATION AND MEASUREMENT

A photograph of the planet Jupiter made from images obtained by Voyager 1 on February 5,
1979, showing the Great Red Spot and three of Jupiter’s four largest satellites: lo (in front of
Jupiter), Europa (brightly lit, to the right), and Callisto (barely visible at the bottom left).

Space science is based on a mathematical description of the universe. This mathematical

description is in turn based on defining physical quantities clearly and precisely so that all
observers can agree on any measurement of these quantities. Every measurement has two
parts: a number and a unit. In mathematics, we tend to focus on the numbers and assume that
the units are taken care of; but in scientific work, units receive careful attention through a
procedure known as dimensional analysis, which is illustrated in the first problem.

Among the physical quantities used to describe the universe, some are considered
fundamental quantities whereas others are derived quantities, comparable to the designation
of definitions and undefined terms in a mathematical system. Although it does not really
matter which particular quantities are the ones designated as fundamental, the most common
are length, mass, and time. In scientific work the two major systems of units for these
quantities are the mks (meter-kilogram-second) and the cgs (centimeter-gram-second). Every
measurement is a comparison with the standards that are universally accepted as definitions
of these fundamental units. In astronomy and space science, where large distances are
common, the meter and even the kilometer are too small to be convenient; in Problems 5, 9,
and 10 of this chapter, we show how more suitable units for length are defined.

Dimensional analysis (manipulation of units according to the rules of algebra) is the

procedure used to ensure consistency in the definition and use of units. For example, since
force is, by definition, the product of mass and acceleration, measured respectively in kg and
m/s2 in the mks system, the unit of force in this system must be equivalent to kg · m/s2. A
new term, the newton, was created to describe the unit of force: 1 newton = 1 kg · m/s2.
PROBLEM 1.

Newton’s law of gravitation, one of the most important ideas in space science, states that the
force of gravitational attraction between two bodies of masses M1 and M2 is proportional to
the product of the two masses and inversely proportional to the square of the distance R
separating the two masses. If G is the constant of proportionality, called the universal

gravitational constant, this law can be stated in symbols as . What must be

the unit for G in the mks system?

Solution:

Using dimensional analysis, we equate the known units in accordance with the relationship
above without worrying about the numbers, then solve algebraically to get the unknown unit.
This gives

PROBLEM 2.

We know that in a circle of radius r, if an arc of length s subtends an angle θ and θ is

measured in radians, then s = rθ. Show that the radian is essentially dimensionless (i.e., an
angle of π/4 radians is just the real number π/4).

Solution:

Since r and s are both lengths, in the mks system they will both be measured in meters. From

s = rθ we have . Since the units cancel, θ is dimensionless.

Scientific theories and technological development both require accurate measurements. Since
every measurement is an approximation, an important aspect of scientific and technical work
is the analysis of experimental error and the control of the propagation of error when
computations are made using measured quantities. The use of computers to solve complex
problems by numerical methods has made error analysis even more important because
computers approximate real numbers using finite decimals. Moreover, computers represent
numbers internally using a floating point binary representation. Even though it is not really
necessary to understand the binary numeration system to work with computers, such
knowledge is essential to the analysis and control of error propagation in computational work.
The next problem considers the floating point binary representation of our familiar numbers.

PROBLEM 3.
The binary (i.e., base two) representation of a number uses only two digits, 0 and 1. Whereas
in base ten the actual value of a digit is the product of its nominal value and the appropriate
power of 10 according to the position of the digit with respect to the decimal point, in base
two the value of a digit is the product of its nominal value and the appropriate power of 2. So,
for example, the binary number 10011 has the value that we represent in base ten as 1 × 24 +
0 × 23 + 0 × 22 + 1 × 21 1 × 20, or 16 + 2 + 1 = 19; the binary number 10.011 is the same as
the decimal number 1 × 21 + 0 × 20 0 × 2-1 + 1 × 2-2 + 1 × 2-3, or 2 + 0.25 + 0.125 = 2.375.

a. Determine the binary representations of the decimal numbers 625, 6.25, and 0.0625.

Solution:

625 can be written as the sum

6.25 can be written as the sum

b. Show that it is impossible to represent the decimal fraction 0.2 exactly in a finite binary
code.

Solution:

0.2 = 2/10 = 1/5. To express this in binary notation, we must write 1/5 as a sum of unit
fractions, each having some power of 2 as denominator. Since 1/23 is the largest such fraction
smaller than 1/5, we begin by finding the difference:

Now the largest unit fraction less than 3/(5 · 23) with a power of 2 as denominator is 1/24, so
we next find the difference:
This means that 1/5 = 1/23 + 1/24 + (1/24) (1/5). Since the fraction 1/5 has recurred, multiplied
by 1/24, we see that the first four digits we have found to the right of the binary “decimal
point,” 0.0011, will repeat continuously. In other words, (A quicker
but less intuitive approach to finding this representation is to express 1/5 as the binary
fraction 1/101 and then divide 101 into 1, using binary arithmetic.)

The reader can use the method of part (b) to show that the decimal fractions 0.1, 0.3, 0.4, 0.6,
0.7, 0.8, 0.9 also have infinitely repeating binary representations.

c. Almost all computers use a floating point binary representation for numbers. In this
system, every number is expressed in the form 0.d1d2 ... dn × 2m, where d1 = 1, di = 0 or 1 for i
= 2, 3, ... , n, and m is an integer. For example, the floating point representations of the
numbers in part (a) would be

Different computers have differing capabilities both with respect to the length (n) of the
string of 0’s and l’s that can be stored for any single number and with respect to the exponent
m that can be stored. The limits available for n and m determine the largest and smallest
number a computer can represent and also the size of the errors that must result when a
number with an infinitely repeating representation must be stored with only a finite string
length available.

If a certain computer can store only an eight-digit string (n = 8), then the representation for
the decimal fraction 0.2 will be stored as 0.11001100 × 2-2. What number is this, and what is
the difference between this number and 0.2?

Solution:
We now state two definitions used in error analysis. These definitions can be applied to both
measurement errors and the errors that arise because of the way in which numbers are
represented in computers. It is probably worth noting in this context that the term
measurement error as used here does not imply that the measurement has been carelessly
made but rather refers to the fact that every measuring instrument is limited in accuracy and
can never provide more than an estimate of a true value.

Let ΧT be the true value of a specified quantity, and let X be the value of this quantity as
measured or as represented in the computer. Then:

Observe that absolute error has the same units as the quantity under consideration, whereas
relative error (usually reported as a percent) is dimensionless.

The relative error is considered to be the indicator of how good a measurement or any other
approximation is. For example, a measurement of 2.5 mm with a possible absolute error of

0.05 mm has a relative error of , or 2 percent, whereas a measurement of 1250 km, with

a (much larger) possible absolute error of 5 km has a much smaller relative error of , or
0.4 percent. Awareness of the appropriate tolerance for relative error is a vital ingredient of
scientific work.

PROBLEM 4.

What are the absolute and relative errors if a computer that has an eight-bit binary digit string
represents 0.2 as 0.11001100 × 2-2?
Solution:

From Problem 3c, the absolute error is 1/1280, or about 0.0008. The relative error
0.0008/0.2 = 0.004, or 0.4 percent.

The use of significant figures is helpful in error analysis. The number of significant figures is
defined as the number of digits that can be assumed to be correct, starting at the left with the
first nonzero digit, and proceeding to the right. By this definition, 10.62, 0.05713, and 4.600
all have four significant figures. A number such as 4300 is ambiguous. This ambiguity may
be resolved by using scientific notation, since we may write the number as 4.3 × 103, 4.30 ×
103, or 4.300 × 103 according to whether the number has two, three, or four significant
figures, respectively.

When approximate numbers are added or subtracted, it can be shown that the absolute error
in the sum or difference could be as large as the sum of the absolute errors of the individual
numbers. When approximate numbers are multiplied or divided, it can be shown that the
relative error of the result could be as large as the sum of the relative errors of the individual
numbers. This means that for sums and differences of approximate numbers, the number of
decimal places considered significant can never be greater than the number of decimal places
in the least precise addend. For products and quotients, the number of significant figures can
never be more than the smallest number of significant figures in the individual factors.
Wherever appropriate, numerical results will be given in accordance with these guidelines.

PROBLEM 5.

Earth’s orbit around the Sun is elliptical, but in many cases it is sufficiently accurate to
approximate the orbit with a circle of radius equal to the mean Earth-Sun distance of 1.49598
× 108 km. This distance is called the Astronomical Unit (AU). Listed in the chart that follows
are actual Earth-Sun distances, given to five significant digits, on the first day of each month
of a representative year. (The American Ephemeris lists daily distances and the actual times
for these distances to seven significant digits.)

Date Distance (× l08 km)

1 January 1.4710
1 February 1.4741
1 March 1.4823
1 April 1.4949
1 May 1.5073
1 June 1.5169
1 July 1.5208
1 August 1.5183
1 September 1.5097
1 October 1.4977
1 November 1.4848
1 December 1.4751
a. To how many significant digits is it reasonable to approximate the Earth-Sun distance as
though the orbit were circular?

Solution:

To two significant digits, each of the distances in the table can be given as 1.5 × 108 km.

b. What are the largest possible absolute and relative errors in using the Astronomical Unit as
the Earth-Sun distance in a computation instead of one of the distances from the table?

Solution:

The procedure of dimensional analysis, described earlier, is easily adapted and commonly
used in science and technology for the task of unit conversion. Recall that in dimensional
analysis the units are manipulated in accordance with the rules of algebra.

Suppose we wish to change a length of 623 cm to meters. The adaptation of dimensional

analysis for unit coversion involves multiplication by a factor unit chosen according to the
following simple principles: the factor unit is a fraction with a value of 1, whose numerator is
expressed in terms of the unit we wish to have and whose denominator is expressed in terms
of the unit we wish to change. Since 100 cm = 1 m, in order to change 623 cm to m, we
perform the multiplication

“canceling the cm in numerator and denominator to get 623/100 m, or

6.23 m.

More complex conversions can be done using multiplication by several factor units and those
readers wishing to convert between British and metric units can also use this method. For
example, the speed of light, 3.00 × 105 km/sec, can be found in miles per hour:

PROBLEM 6.
The deep space probe Pioneer 10 took 21 months to get from Mars to Jupiter, a distance of
998 million kilometers. Use the factor unit technique to find its average speed in kilometers
per hour during that period.

Solution:

PROBLEM 7.

a. Recall that the Astronomical Unit (Earth-Sun distance) discussed in Problem 5 is 1.496 ×
108 km, to four significant figures. Find the Earth-Sun distance in miles to three significant
figures.

Solution:

b. The chart that follows gives the mean distance in kilometers of each planet in the solar
system from the Sun. Express these distances in AU, using a suitable number of significant
digits.

Planet Distance (km × 108)

Mercury 0.579
Venus 1.08
Mars 2.27
Jupiter 7.78
Saturn 14.3
Uranus 28.7
Neptune 45.0
Pluto 59.1

Solution:
Since each distance in the table has three significant digits, and the factor unit

has an exact number in the numerator and six significant digits in the
denominator, the distances in AU can be given to three significant digits. Multiplying by the
factor unit shown gives the following distances in AU:

PROBLEM 8.

The Solar Maximum Mission (SMM) satellite orbits Earth at a height of 560 km. In many
computations, the Earth-Sun distance of 1.5 × 108 km is used to approximate the distance of
SMM from the Sun. What is the maximum relative error of this approximation?

Solution:

The distance of SMM from the Sun is contained within the range (Earth-Sun distance ±
(Earth diameter + 560 km)), or (Earth-Sun distance ± 6930 km). From Problem 5, if 1.5 × 108
km is used as the Earth-Sun distance, the absolute error ≤2.48 × 106 km. 6930 km is much
smaller than this error, so that the absolute and relative errors incurred in using 1.5 × 108 km
as the SMM-Sun distance are the same as those of part (b) of Problem 5. If greater accuracy
is required, it

will be necessary to use daily Ephemeris values such as those in the listing given in Problem
7. In this event, it is still true that since 6930 km < 7 × 103 km, the relative error in
approximating the SMM-Sun distance with the Earth-Sun

distance or about 0.005 percent.

The Astronomical Unit (AU), although useful for measuring distances within the solar
system, is too small to be convenient for distances to stars. We shall therefore consider two
other units of length used by astronomers. The first is called the light-year.

PROBLEM 9.

The light-year is the distance traveled by light during one Earth year. To three significant
digits, the speed of light is 3.00 × 105 km/s. Find the length of the light-year in km and in
AU.

Solution:
1 Earth year = 365.25 days = 365.25 × 24 × 60 × 60 seconds. In one year, light travels 3.00 ×
105 × 365.25 × 24 × 60 × 60 km = 9.47 × 1012 km. To express this distance in AU, 1 light-

year = 9.47 × 1012 km × = 6.31 × 104 AU.

The parsec is the astronomical unit of distance that relates to observational measurements. In
order to define this unit, we must consider the fact that when we observe the heavens, we
have no direct perception of depth or distance. A useful model developed to portray the
heavens is the celestial sphere. In this model, Earth is surrounded by an imaginary sphere
with infinite radius. A coordinate system, similar to latitude and longitude, is imposed on the
celestial sphere by projecting Earth’s rotation axis on the sphere to identify the celestial north
pole (CNP) and celestial south pole (CSP) as shown in Fig. 2.1. Since the radius of the
celestial sphere is infinite, all parallel lines point to the same spot on the sphere, and so every
line parallel to Earth’s rotation axis also points to the celestial north and south poles.

The extension of Earth’s equatorial plane intersects the celestial sphere in a great circle called
the celestial equator. Now a system of small circles of declination (δ), comparable to latitude
circles on Earth, is imagined on the celestial sphere, and a system of great circles called right
ascension (a) circles, comparable to longitude, passing through the two poles, completes the
coordinate system (Fig. 2.2).

Fig. 2.1
Fig. 2.2

Every star or celestial object can now have its position identified by the ordered pair (α, δ).
Because Earth rotates with respect to the celestial sphere, the time of observation must also
be known in order to use the coordinate system. Differences in the positions of two objects on
the celestial sphere are expressed in terms of the angle subtended at Earth by the arc joining
these points.

As Earth revolves around the Sun, very distant stars show no discernible changes in position,
but closer stars will show apparent motion with respect to the celestial sphere when viewed
from different points in Earth’s orbit, as shown in Fig. 2.3. This apparent motion is called
parallactic motion, and the change in position is called the parallax angle. In this context, 1
parsec is defined as the distance at which the radius of Earth’s orbit subtends an angle
measuring 1 arc-second (see Fig. 2.4).
Fig. 2.3

Fig. 2.4