Refacte monogramRefacte

Mathematics / Algebraic number theory

IBM Ponder This July 2010

Algebraic numbers, polynomial roots, and modular arithmetic.

In this exposition

The mechanism

Turn a rounding condition into finite arithmetic

The useful change of viewpoint is to collect algebraic conjugates into a trace. An integer recurrence then records the quantities relevant to rounding, and a companion matrix organises their residues. The numerical-looking question becomes a finite arithmetic calculation.

What becomes elementary: Trace identities connect the original approximation problem to an exact recurrence and modular computation.

Follow the bridge from conjugates to the recurrence before using the modular calculation. The original challenge and the supplied verification materials remain linked with the study.

Explore further constructions. Edition dates, sources and reference chapters identify the scope of each study.

Abstract

We solve the July 2010 IBM Ponder This challenge asking for a natural number such that

The central observation is that

is a cubic Pisot unit: it is an algebraic integer greater than , while its other two conjugates have absolute value less than . Their powers complete to the integer trace , with an error that decays exponentially. A companion matrix then turns the trace recurrence into finite modular arithmetic. An order calculation modulo and , followed by lifting to and , gives an explicit solution.

Challenge (IBM Ponder This, July 2010). The July 2010 IBM Ponder This challenge asks for a natural number such that

Here denotes the nearest integer to . In the argument below, the number being rounded will always be strictly within of a unique integer, so no convention about half-integers is needed.

1 Introduction and strategy

The operation to simplify is rounding a very large real number. Seek an exact integer sequence that differs from only by terms tending to zero. Adding the powers of the other conjugates supplies such a sequence; a recurrence will then let us calculate it modulo .

Set

The key point is that is a cubic algebraic integer whose other two conjugates lie strictly inside the unit circle. Thus, if and are the other conjugates, then

is an integer, while rapidly tends to . Therefore, for all sufficiently large ,

The rounding in the problem is therefore naturally explained by the trace map.

The remaining task is to force this trace to be modulo . The trace sequence satisfies a linear recurrence coming from the minimal polynomial

Equivalently, it is encoded by the companion matrix

Since , this matrix is invertible modulo every positive integer. The congruence problem becomes a finite group problem: find a power of that is congruent to the identity modulo , then use the fact that the trace at index is .

We will prove that

works. This value is not claimed to be minimal; the point is to give a clean Galois-theoretic and group-theoretic certificate.

2 The cubic Pisot unit

Let

We first determine its minimal polynomial and its conjugates.

Lemma 1. The number has minimal polynomial

Its three conjugates are

Moreover,

Proof. We use the identity

Taking , we have , hence

Since , this gives

Expanding,

Thus is a root of

The rational root test shows that has no rational root, since neither nor is a root. Therefore is irreducible over , and so is the minimal polynomial of .

The same trigonometric identity also shows that

are roots of , because

and both and equal . These three roots are distinct, so they are precisely the three conjugates of .

It remains to bound the two smaller conjugates. We compute

so one root lies in . Also

so one root lies in . Finally,

so the remaining root lies in . Since

the other two conjugates must lie in and . Hence

From the cyclotomic viewpoint, if

then

Thus lies in the real cubic subfield of the -th cyclotomic field. The three conjugates above correspond to the three embeddings of this real cubic field. This is the Galois-theoretic source of the trace that appears next.

3 The trace sequence and the rounding step

Let

For , define

Lemma 2. For every , one has

Moreover, for every ,

Proof. Since is monic, is an algebraic integer. Also,

implies

Thus is a unit. Hence is an algebraic integer for every , and therefore its trace

is an integer.

For , we have

Since is an integer, this proves

for all .

Thus the original problem has been reduced to the following congruence problem:

Find such that .

4 The companion matrix

The trace sequence satisfies a recurrence coming from the polynomial

Indeed, every root satisfies

Multiplying by and summing over the three conjugates gives

It is convenient to encode this recurrence using the companion matrix

Then

Therefore the eigenvalues of are , and for every ,

We will also need the trace at index . Since

the matrix is invertible over , and a direct computation gives

Hence

This agrees with

which also follows from the elementary symmetric relation

So our goal is to find a large positive such that

Then

and therefore

5 Orders modulo 2 and 5

We now compute suitable orders of modulo and .

Proposition 1. Modulo , the matrix has order . Modulo , the matrix has order . In particular,

Proof. Let

For a prime , let denote the reduction of modulo , and let denote the reduction of modulo .

First consider . We have

This polynomial has no root in , since

As a cubic with no root, it is irreducible over .

Since is the characteristic polynomial of , and is irreducible of degree , it is also the minimal polynomial of . Thus

Under this isomorphism, corresponds to the class . Since , and has order , we get

and has order . Therefore

Now consider . We have

Checking the five elements of ,

so has no root in . Hence it is irreducible over .

Again,

Let . The norm of from to is

On the other hand, since is a root of the monic cubic , the product of its three conjugates is . Therefore

It follows that

Moreover, , and , since is not a root of . The only divisors of are , so has order . Therefore

6 Lifting to 2⁹ and 5⁹

We use a standard lifting lemma.

Lemma 3. Let be a prime, and let be an integer matrix satisfying

Then, for every ,

Proof. We argue by induction on . The case is the hypothesis.

Assume

for some integer matrix . Then

Expanding,

Every term after is divisible by . Hence

This completes the induction.

Applying the lemma to with , we get

Applying it to with , we get

Now set

Since , this least common multiple is

Thus

Because and are coprime, the Chinese remainder theorem gives

We can now finish the proof.

Theorem 1. The integer

satisfies

Proof. Let

We have just shown that

Therefore

Taking traces,

But

and

Hence

Since , the rounding lemma gives

Thus

Substituting

and

gives

Therefore

is a valid answer.

7 A coarser but very uniform variant

The exact order computation modulo and gives a reasonably small exponent. If one wants an even more uniform group-theoretic certificate, one can avoid computing the exact orders of modulo and .

Since

the reduction of modulo any prime lies in

Therefore the order of modulo divides

For ,

For ,

Thus

The same lifting lemma gives

Hence

satisfies

Exactly as before,

so

Since , we again have

Therefore the coarser group-order argument gives the valid exponent

This version uses only the general finite group bound

The sharper proof above replaces those large group orders by the actual orders and of this particular companion matrix modulo and .

8 Summary of the mechanism

The proof rests on three linked ideas.

First, the number

is not just a trigonometric curiosity. It is a cubic algebraic integer with minimal polynomial

Its other two conjugates have absolute value less than , so powers of are very close to the integer traces

This explains the appearance of the rounding function.

Second, the trace sequence

satisfies the same recurrence dictated by the minimal polynomial:

The companion matrix packages this recurrence into the identity

Third, because the companion matrix is invertible modulo , its powers are periodic modulo . Choosing so that

forces

The trace of is , so

Since is exactly the rounded value of , the desired divisibility follows.