Summary
This lesson constructs the number as the supremum of the sequence , which arises from compound interest: it proves with the binomial theorem that the sequence increases and does not reach the value , and it encloses its supremum between this sequence and the decreasing sequence . It then proves that is also the supremum of the sums of the reciprocals of the factorials, bounds the error of these sums, which makes it possible to compute to any desired precision, and proves that is irrational. Finally, it defines the natural exponential, relates it to the continuous compounding of interest and proves the inequality , which says that the line lies below the graph of the exponential.
Learning objectives
- Prove that the sequence is increasing and bounded, and define the number as its supremum.
- Compute approximations of by means of the sum of the reciprocals of the factorials, with a rigorous bound on the error, and prove that is irrational.
- Relate the natural exponential to compound interest and to its continuous compounding, and compare the effect of compounding at different frequencies.
- Prove and apply the inequality , and interpret it as the position of the line relative to the graph of .
The compound interest sequence
A capital of is deposited for one year at an interest rate of one hundred percent per year. If the interest is paid only once, at the end of the year, the capital doubles and is worth . If it is paid twice, each half-year at half the rate, the interest of the first half-year itself earns interest in the second: the capital is multiplied twice by and is worth . With four quarterly payments it is worth , a little more than 2.44, and with twelve monthly payments, a little more than 2.61. In general, if the year is divided into equal periods and in each one the fraction of the rate is paid, the capital is multiplied times by , and at the end of the year it is worth
As increases, the capital is multiplied more times, but by a factor closer to , and it is not evident which of the two effects prevails: the powers of are all equal to , while the powers of a fixed base greater than grow without bound. This section proves that the sequence increases, but that its terms do not reach the value ; its supremum is the number . The proofs rest on the binomial theorem from the lesson on combinatorics and the binomial theorem, on Bernoulli’s inequality, for , proved in the lesson on the Peano axioms, and on the completeness of the real numbers, stated in the lesson on number systems.
The growth of the sequence
By the binomial theorem, is a sum of terms. The term of index is , and it is written as the product of and positive factors less than ; with the convention that a product with no factors equals , the expression is also valid for and :
In this expression, appears only in the factors , which increase with , and in the number of terms, which also increases. The two effects push in the same direction; that is the idea of the theorem below.
For every natural number we have , that is, the sequence is strictly increasing.
The binomial expansions of and of are compared term by term. For , the figure represents each term as a block in a column that starts at , because the first two terms are equal to in every case.
Proof
By the binomial theorem, with and , the term of index of is . The binomial coefficient is a product of factors divided by , and each factor is divided by one of the factors of the denominator.
The first quotient equals , and each of the ones after it is , for from to . Adding the terms gives the expression for .
The same expression, with in place of , has one more term, the one of index , which is positive because all its factors are.
For , as the denominator increases, decreases, and the factor increases; moreover, all the factors are positive.
By the monotonicity of the product of positive numbers, each term of of index , if there is one, is less than the term of the same index of ; those of indices and are equal to in both sums.
Adding up, does not exceed the sum of the first terms of , which still lacks the positive term of index . Therefore, . In the figure, the terms of the sequence rise.
An upper bound
Growth is not enough for the sequence to have a supremum: the natural numbers also grow. An upper bound is needed, and to obtain it is compared with the sum of the reciprocals of the factorials,
which results from deleting, in the expression for , all the factors . These sums are of interest in their own right, and the section that follows will prove that they approach the same number as , and much faster. Proposed problem 5 of the lesson on combinatorics asked for a proof that ; the second of the lemmas below contains this bound, together with the intermediate inequality , which will be used later. The first compares the factorials with the powers of , and it is what makes it possible to bound by a geometric progression.
For every natural number we have , with equality only if or .
The argument is by induction on , with the principle from the lesson on the Peano axioms: in passing from to , the factorial is multiplied by , which is at least , whereas the power only doubles. For from to , the figure compares the factorial with the power, as two columns.
Proof
Base case: for both sides equal .
Induction hypothesis: suppose that the inequality holds for a natural number .
The factorial of is that of multiplied by , which is at least ; since all the factors are positive, the induction hypothesis gives the inequality for .
By the principle of induction, the inequality holds for every natural number . If , the factor is at least , and the inequality for is strict: equality occurs at and at , and only there.
For every natural number we have ; the first inequality is strict for .
Each term of the expansion of is bounded by the corresponding term of , and each term of by one of a geometric progression with common ratio . For , the figure compares the three columns on the common base .
Proof
Each factor lies between and ; therefore their product does not exceed , and each term of does not exceed the term of .
Adding for from to gives the first inequality. If , the term of index has the factor , and the inequality is strict.
By the lemma above, for every natural number ; on taking reciprocals of two positive numbers, the inequality reverses.
We separate the term of index and bound the others.
The highlighted expression is the sum of terms of a geometric progression with first term and common ratio , computed in the lesson on summations and progressions.
Combining both inequalities gives the lemma. In the figure, with , the geometric column stops short of , and the other two lie below it.
A decreasing sequence that accompanies it
If the base is raised to an exponent one greater, a second sequence is obtained, whose terms are greater than those of the first, because they are multiplied by a factor greater than :
Its first terms are , and : they go down. The theorem below proves this in general; the idea is to study the quotient of two consecutive terms, and the tool, Bernoulli’s inequality.
For every natural number we have , that is, the sequence is strictly decreasing.
It suffices to prove that the quotient is greater than , because the terms are positive. The figure shows the first terms of both sequences.
Proof
We write each base as a quotient, split off a factor from the power in the denominator and combine the two powers with exponent into a single one.
The numerator of the base exceeds the denominator by , because .
By Bernoulli’s inequality, with and exponent , the power of this base is at least .
We substitute into the quotient and reduce to a common denominator; the sum equals .
We expand the numerator and the denominator: the numerator exceeds the denominator by .
Therefore, . In the figure, the terms go down while the terms go up, and each lies above each .
The number e
The two sequences approach each other, but they do not cross: every term of the first is less than every term of the second, even if their indices are different.
For all natural numbers and we have .
The argument passes through the larger of the two indices: the first sequence increases, the second decreases, and at each index the second lies above the first. The figure follows the case and .
Proof
At the same index , the term is multiplied by , which is greater than , and is positive.
Let be the larger of the natural numbers and . Since and the sequence is increasing, its term of index does not exceed the one of index .
Since and the sequence is decreasing, its term of index does not exceed the one of index .
Combining the three inequalities, the middle one strict, gives the lemma. In the figure, with and , the larger index is .
It follows that the set of the terms is nonempty and bounded above, by or by any . By the completeness of the real numbers, it has a supremum, which is denoted by the letter .
Definition. The number is the supremum of the set of the terms of the sequence , for a natural number.
For every natural number we have .
We use the definition of the supremum as the least upper bound and the monotonicity of both sequences. The figure shows the two sequences, which approach from opposite sides.
Proof
By the lemma above, every term of the first sequence is less than every term of the second.
Each is therefore an upper bound of the set of the , and the supremum, which is the least of the upper bounds, exceeds none of them.
Since is an upper bound and the sequence is strictly increasing, each term is less than the one after it, which does not exceed .
Likewise, since the sequence is strictly decreasing, does not exceed the term after it, which is less than .
Combining both inequalities gives the theorem: is the only number that lies between all the and all the .
With a calculator one finds that exceeds 2.7 and that is less than 2.8; therefore, lies between 2.7 and 2.8. The bracketing is slow: the distance between the two sequences is , which decreases like the reciprocal of .
The number is approximately 2.718281828. Jakob Bernoulli found it in 1683 while studying compound interest, and Leonhard Euler, who discovered a good many of its properties, denoted it by the letter . Together with , it is one of the two fundamental constants of analysis.
Loading the content…
Completed
Sign in to save your progress