Why compare a number with its divisors?
The proper positive divisors of are , and their sum is . The same happens for : . Such a coincidence suggests a classification problem. Instead of listing proper divisors afresh, include the number itself and use the divisor-sum function . Then the defining equation becomes , a form that works well with prime factorization. The goal is to turn this additive condition into restrictions on the prime factors, and then to prove those restrictions are sufficient.
We assume unique prime factorization and revisit the multiplicativity of below.
Divisor sums and a construction
Definition 4.2.1
(Proper divisors and the three classes)
For a positive integer , a proper divisor is a positive divisor other than . Call perfect, deficient, or abundant according as the sum of its proper divisors equals, is less than, or exceeds . Equivalently, compare with . In particular is deficient: its proper-divisor sum is the empty sum .
Concept lensAlgebraic
Perfect means the reciprocals of the divisors sum to two
Pairing each divisor of with gives
Thus is perfect exactly when ; for , . In this form a proper multiple of keeps every reciprocal with and adds , so its sum is strictly larger. Every proper multiple of a perfect or abundant number is abundant; for example .
If , each divisor is uniquely obtained by choosing one exponent from to for every prime. Expanding the product of the corresponding sums gives
This also proves when : the prime-power choices are disjoint. The coprimality condition matters; for example but .
Worked example 4.2.2
(Classifying with divisor sums)
Consider . Since the prime-power factors are coprime,
The proper-divisor sum is , so is abundant. Indeed its proper divisors are , whose sum is . Compare : here , so is deficient. Subtracting the number itself explains why comparison with , rather than , gives the correct classification.
How the candidate formula arises
A power of two almost works: for ,
Its proper-divisor sum falls short by exactly . Multiplying the number by an odd prime introduces a second choice, or , into every divisor. Because is coprime to , the ratio of divisor sum to number becomes
For this ratio to equal , we need
Thus the familiar formula is forced once we look among numbers with a power of two and one odd prime to the first power. It remains to show that these candidates work when the odd factor is prime, and that an even perfect number cannot have a more complicated odd factor. Those are the two directions of the theorem below. This calculation motivates the shape; it has not yet justified restricting the search to that shape.
Theorem 4.2.3
(Euclid's construction)
Let be an integer. If is prime, then is perfect.
Proof
(Divisor sum over two coprime factors)
Indeed is odd, so the two factors of are coprime. Therefore
Both the primeness of and the coprimality are used: is not valid for a composite .
Counterexample mode
The construction needs a prime odd factor
It is tempting to read the theorem as a formula: every with is perfect. The first composite case already refutes this. For , the candidate is , and
The divisor sum is three times the number, not twice, so is abundant. The extra divisors and of are exactly what the step cannot see. The repaired statement keeps the hypothesis that is prime; the next subsection shows that a composite value always makes the candidate abundant.
Worked example 4.2.4
(From a Mersenne prime to a perfect number)
Take . Then is prime, giving . Rather than enumerate ten divisors, calculate . Thus the proper divisors sum to .
To see exactly which divisors this product counts, arrange them in two rows:
The first row has odd part , and the second has odd part . Since is prime, there is no third possible odd part. To get the proper divisors, remove only the final entry . The remaining first row sums to and the remaining second row sums to , giving altogether.
The same description works for every prime . The first row is for , and the second is for the same range. All first-row entries are proper divisors because . Only the last second-row entry is the number itself. Consequently the proper-divisor sum is
This is the construction proof read directly from its divisors. It also shows where the prime hypothesis enters: if the odd factor were composite, its nontrivial proper divisors would create additional rows that this two-row description misses.
Definition 4.2.5
(Mersenne numbers)
A number of the form is called a Mersenne number; when it is prime it is a Mersenne prime.
If is prime then must be prime. To see this, if with , then
Both factors exceed , so is a proper factor. This gives a factorization whenever the exponent is composite; it is not a primality test for the remaining exponents. The converse is false: is prime, but . Testing the exponent alone is therefore insufficient.
What happens when the odd factor is composite?
In fact a composite Mersenne number makes the candidate abundant. Put and , still with . The geometric sum is , regardless of whether is prime. If is composite, it has a divisor strictly between and , so . Hence
The failure has a precise reason: the extra divisors of contribute more than the construction allows. We do not need to know their complete list to obtain this strict inequality.
Worked example 4.2.6
(A candidate with too many divisors)
For , the candidate is . Since ,
The target would be . Equivalently, the actual proper-divisor sum is . The number contributes , whereas the prime-factor calculation would require only . This is why substituting a composite value into the formula does not produce a perfect number.
Why every even perfect number has this form
Theorem 4.2.7
(Euclid–Euler theorem)
An even positive integer is perfect if and only if it is for some integer for which is prime.
The construction proves one direction. The converse needs no prior assumption about the number or exponents of odd prime factors. We use the full divisor sum to show that there is room for only one odd prime.
Proof
(The odd part has room for one prime only)
Separate the power of two. Write an even perfect number as , where and is odd. Here is the exact exponent of : divide by until the remaining factor is odd. This makes and coprime, so multiplicativity is valid for these two factors. Put . Then , and gives
Force a divisor into the odd part. The integers and are coprime, so Euclid's lemma gives . Notice that this use of the lemma does not require itself to be prime. We can also see the divisibility directly by rearranging the equation as . Write with . Substitution now gives
Use the remaining divisor-sum budget. The equation says that all proper divisors of must sum to . But , and ensures , so is already a proper divisor. If , the separate proper divisor makes their sum at least , too large. Equivalently, would be three distinct positive divisors of with sum . All divisors are positive, so including any others cannot repair this contradiction.
Thus and . The displayed equation becomes . Since , its divisors include and ; a composite would have another positive divisor, making the sum larger. Hence is prime. Setting completes the converse. In particular, the argument also rules out the possibility : it has already forced to be a multiple of .
Proof X-ray
Which step uses which hypothesis
The converse uses three facts, each at one place. First, the exact exponent is what makes and coprime, so that is legitimate. With a non-exact split the equation is false: writing would give , whereas . Second, the divisibility comes from alone; at this stage nothing is known about the primality of .
The decisive move is the budget argument. After , the equation fixes the total of all proper divisors of at , and is itself a proper divisor because . Positivity of divisors then turns the presence of the extra divisor into a strict inequality whenever . Once , the same budget, now , forces to be prime. Primality is therefore an output of the proof, not an assumption, and the argument closes without any case analysis on the odd prime factors of .
This proof does not assume that the odd factor is prime. It derives first that the geometric sum divides that factor, and only then shows no extra divisor can occur. The theorem classifies even perfect numbers; it neither proves that an odd perfect number exists nor that none exists. Those are separate questions.
Worked example 4.2.8
(Recognizing the reverse direction)
Suppose the even perfect number under investigation has : it is divisible by but not by . The proof forces its odd factor to be , and also forces to be prime. As , this is impossible. Thus no even perfect number has exactly three factors of . This does not rule out arbitrary numbers divisible by ; the exact exponent matters.
Why the correspondence is one-to-one
For a Mersenne prime , the construction assigns the even perfect number
The inverse is concrete: take the odd part of , that is, divide by its largest power of . Because is odd, this recovers exactly . Alternatively, the exact exponent recovers . Thus two different Mersenne primes cannot give the same . The converse theorem shows that every even perfect number has such a preimage, so the assignment is both injective and surjective.
For instance, has odd part . It therefore comes from and , with no second choice of exponent. This recovery applies to an even number only after it is known to be perfect: for example, has odd part but is not the perfect number associated with , which is .
Common pitfalls
A prime exponent is necessary but not sufficient for a Mersenne number to be prime. Conversely, finding a composite Mersenne number excludes only the even perfect number associated with that exponent. It says nothing by itself about other exponents or about odd perfect numbers. Also, even when is multiplicative, multiplying its values at noncoprime arguments can overcount divisors: use the prime-power formula instead. The theorem is a bijection between Mersenne primes and even perfect numbers, but it does not establish that either collection is infinite.
Common mistake
(Factoring the odd part too early)
After writing , one might try to replace by . Nothing has shown that and are coprime. For example, taking gives , whereas . These values are an illustration of the invalid multiplication, not a proposed perfect number. The converse proof avoids the issue: it uses only that divides and that sums of positive divisors give lower bounds. A useful proof habit is to check the hypotheses again whenever a factorization changes.
Practice
Quick check
(Classification)
Classify and as deficient, perfect, or abundant, showing the divisor-sum calculation.
Solution · Answer
, so is deficient. Also , so is abundant.
Quick check
(Construct a perfect number)
Use the theorem to find the even perfect number with , verifying the needed primality.
Solution · Answer
. A composite positive integer has a prime divisor at most its square root: in a product of two factors greater than , the smaller factor is at most the square root and has a prime divisor. Here , so the primes to check are . The first three are ruled out by parity, digit sum and the final digit; also . Hence is prime and the answer is .
Quick check
(An excluded exponent)
Show that no even perfect number has exactly factors of .
Solution · Answer
Here and the converse would require to be prime. But , so no such even perfect number exists.
Quick check
(A false converse)
A student argues that is prime, so is prime. Identify the error and give a factorization.
Solution · Answer
The implication goes in the other direction: primality of forces primality of . The converse fails because .
Quick check
(A complete bounded classification)
Find all even perfect numbers less than . Explain why checking only a few exponents is enough; a list of examples alone does not establish completeness.
Solution · Answer
Every even perfect number has with . Since , we have . If , this is at least , already too large. Thus only need checking. They give odd factors . The first, second and fourth are prime: for , checking the primes suffices because its square root is less than ; the smaller cases are immediate. The factor is composite. The complete list is consequently . Completeness comes from the converse together with the bound, while the construction verifies that all three listed numbers work.
Quick check
(Divisibility versus an exact exponent)
A student says that no even perfect number is divisible by , because is composite. Explain the mistake and give a counterexample.
Solution · Answer
The composite value excludes exactly , not every . The perfect number is divisible by and has . Its odd factor is , a different Mersenne number. Saying does not say that is the largest power of two dividing .