# A note on the paper arXiv:2112.14547

We give historical remarks related to arXiv:2112.14547 ("A New Method of Construction of Permutation Trinomials with Coefficients 1", by Guo et al.). In particular, we show that the "new" permutation polynomials in that paper are actually well known, and that the three conjectures resolved in that paper had all been resolved previously. In addition we give a simpler derivation of these permutation polynomials than had been given previously, which demonstrates the general method of producing permutation polynomials that was introduced in arXiv:1310.0776.

## Authors

• 1 publication
06/15/2018

### Permutation polynomials and complete permutation polynomials over F_q^3

Motivated by many recent constructions of permutation polynomials over F...
01/17/2019

### New Results about the Boomerang Uniformity of Permutation Polynomials

In EUROCRYPT 2018, Cid et al. BCT2018 introduced a new concept on the cr...
12/29/2021

### A New Method of Construction of Permutation Trinomials with Coefficients 1

Permutation polynomials over finite fields are an interesting and consta...
12/21/2017

### A Recursive Construction of Permutation Polynomials over F_q^2 with Odd Characteristic from Rédei Functions

In this paper, we construct two classes of permutation polynomials over ...
12/07/2017

### A general cipher for individual data anonymization

Over the years, the literature on individual data anonymization has burg...
11/18/2021

### C-OPH: Improving the Accuracy of One Permutation Hashing (OPH) with Circulant Permutations

Minwise hashing (MinHash) is a classical method for efficiently estimati...
09/10/2021

### C-MinHash: Practically Reducing Two Permutations to Just One

Traditional minwise hashing (MinHash) requires applying K independent pe...
##### This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

## 1. Introduction

A polynomial is called a permutation polynomial if the function permutes . The recent paper [2] purports to provide new classes of permutation polynomials, and to resolve three conjectures from the literature. Here we show that the permutation polynomials in that paper are in fact well-known, and that the conjectures had been resolved previously.111Shortly after [2] appeared on the arXiv, I emailed the content of this note to the third and fifth authors of that paper (I was not able to find email addresses for the other authors), and suggested that [2] should be revised in light of the content of this note. Since I did not receive a reply, and no new version of [2] has been posted, I am posting this note now, in order to help members of the permutation polynomials community avoid spending further time and effort on rediscovering known results. We also give a new proof of the main result of [2], which is significantly simpler and more direct than all previous proofs, and which demonstrates the general method of producing permutation polynomials that was introduced in [11].

The main result of [2] is as follows.

###### Theorem 1.1.

Let be positive integers with , and write , , and . Let be an integer, and let be positive integers such that

 d1 ≡Q−R+u(q+1)(modq2−1) d2 ≡Q+R+(u−R)(q+1)(modq2−1) d3 ≡−(Q+R)+(u+Q)(q+1)(modq2−1).

If then permutes .

###### Remark.

The above statement includes all permutation polynomials that can be inferred from any interpretation of [2, Thm. 3.1]. The statement of the latter result does not require the to be positive, and does not say that its expressions for the ’s should be interpreted as congruences mod (). However, we assume that the authors of [2] intended to state their result as above. The positivity condition is needed in order to make their result be true (and indeed, negative ’s would not yield polynomials), and after imposing positivity then the congruences becomes natural, since such congruences do not affect whether permutes .

We will use the following definition.

###### Definition 1.2.

We say that polynomials are multiplicatively equivalent if for some and some positive integer such that .

The following properties of multiplicative equivalence are immediate:

1. Multiplicative equivalence is an equivalence relation on .

2. If are multiplicatively equivalent then permutes if and only if permutes .

3. If and are multiplicatively equivalent and then has at most as many terms as does .

In light of the above properties, multiplicative equivalence is a natural equivalence relation to use when deciding whether one permutation polynomial with few terms is essentially the same as another.

###### Remark.

What we call multiplicative equivalence has been called “quasi-multiplicative equivalence” in previous papers. The term “multiplicative equivalence” has been defined previously to mean two different things, neither of which is equivalent to the above definition. However, we suggest that the above definition should be used in the future, for the betterment of the subject – for instance, the use of the previous definitions has led authors to spend time and effort producing permutation polynomials that could have been obtained immediately from previously known permutation polynomials by composing with .

We will show that the permutation polynomials in Theorem 1.1 have appeared in the following previous results (listed according to the order in which the relevant papers were posted in the public domain):

1. Some instances of the permutation polynomials in Theorem 1.1 are special cases of the much more general classes of permutation polynomials in [11, Thm. 1.1 and 1.2]; however, the paper [11] does not draw attention to the relevant special cases.

2. If

is odd and

is even then every permutation polynomial in Theorem 1.1 appears in one of [7, Cor. 3.8, 3.9, 3.12, and 3.13]. If is even and is odd then every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in one of [7, Cor. 3.8, 3.9, 3.12, and 3.13].

3. Every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in [4, Thm. 1].

4. Every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in [1, Thm. 4.2].

5. Every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in [8, Thm. 3.2].

6. If is odd and then every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in [5, Thm. 1].

7. Every permutation polynomial in Theorem 1.1 is multiplicatively equivalent to a permutation polynomial in [9, Thm. 1.1].

In addition, [2] purports to be the first paper to resolve three conjectures. We note that those conjectures were proved previously. Specifically, the two conjectures from [6] were first resolved as parts (a) and (b) of [11, Cor. 1.4], and Conjecture 2 of [3] was first resolved in [7, Thm. 4.2].

In the next section we give a very short and simple proof of Theorem 1.1, based on the arguments in [11], which pinpoints the key reason why permutes . This proof avoids the non-conceptual computations occurring in all previous proofs of Theorem 1.1. It turns out that the same approach can be used to deduce all the results mentioned above, in addition to dozens of other results from the literature and arbitrarily many as-yet unpublished results. We encourage readers interested in permutation polynomials to look into [11], so that they can focus their attention and time on producing results which do not follow immediately from the arguments in that paper. We conclude this note in Section 3 by explaining how Theorem 1.1 connects with previous results.

## 2. Proof of Theorem 1.1

In this section we give a new proof of Theorem 1.1. We write for the set of -th roots of unity in , and if then we write for the polynomial obtained from by raising all coefficients to the -th power. We first restate the condition for certain polynomials to permute in terms of whether an associated polynomial permutes , via the following special case of an easy and much-used lemma from [10].

###### Lemma 2.1.

Write where is a positive integer, is a prime power, and . Then permutes if and only if and permutes .

We next translate the condition that permutes into the condition that an associated rational function permutes , where typically has much lower degree than does . We do this in the following trivial lemma, which encodes a procedure introduced in [11].

###### Lemma 2.2.

Write where is an integer, is a prime power, and . Then maps into , and if has no roots in then induces the same function on as does for any integer with . In particular, permutes if and only if permutes and has no roots in .

A key ingredient in [11] is degree-one rational functions which map to either or . In this note we use

 ρ(X):=X+ωωX+1

where is a prescribed order- element of , with being a power of . The following result is a special case of [11, Lemmas 2.1 and 3.1], and also is easy to verify directly.

###### Lemma 2.3.

Let where . If is even then permutes , and if is odd then interchanges and .

Pick any nonconstant . In light of Lemma 2.3, if is even then permutes if and only if permutes , and if is odd then permutes if and only if permutes . We will show that the permutation polynomials in Theorem 1.1 correspond to rational functions permuting (via Lemmas 2.1 and 2.2) where with . The following result presents these compositions in the cases we need; it can be verified by a routine computation.

###### Lemma 2.4.

Let be positive integers with , and write , , and . Then

 X(−1)m∘XQ+R+XQ+1XQ+R+XR+1={ρ∘XR−Q∘ρ if ℓ≡m(mod2)ρ∘XR+Q∘ρ if ℓ≢m(mod2).

Now we prove Theorem 1.1.

###### Proof of Theorem 1.1.

Note that and . Thus , so the hypothesis implies that . By Lemma 2.1 and Lemma 2.2, it suffices to show that has no roots in and permutes , where

 g(X):=XQ+RA(q)(1/X)A(X)=XQ+R+XQ+1XQ+R+XR+1.

We first show that has no roots in . Suppose to the contrary that satisfies . Then also

 0=αQ+RA(α)q=αQ+RA(αq)=αQ+RA(1α)=αQ+αQ+R+1.

Thus , so . Since , it follows that ; but plainly , contradiction.

It remains to show that permutes . First suppose . Then the hypothesis implies that is even, so that permutes and also permutes by Lemma 2.3. Thus Lemma 2.4 implies that permutes

Now suppose . If is odd then interchanges and , and we have so that permutes , whence permutes by Lemma 2.4. Finally, if is even then permutes , and since we have , so that permutes , whence again permutes . ∎

###### Remark.

The method used in the above proof can be used to produce enormous collections of permutation polynomials over , for any prime power . One can start with any rational function which permutes , and any degree-one such that and , in order to obtain a rational function which permutes . It turns out that can always be written in infinitely many ways as where and has no roots in . If either is even or is odd then there exist positive integers such that and , so that permutes . By applying this procedure to the most well-known permutation rational functions over , and using certain choices of , , and , one obtains huge classes of permutation polynomials over which include as very special cases essentially all known permutation polynomials of the form . We will elaborate on this remark in forthcoming joint papers with Zhiguo Ding.

## 3. Connection with previous results

In this section we explain the connection between Theorem 1.1 and previous results. The combination of [7, Cor. 3.8, 3.9, 3.12, and 3.13] is as follows.

###### Proposition 3.1.

Let be positive integers with odd and even, and write , , and . If then

 g(X):=Xr(X(S+T)(q−1)+XT(q−1)+1)

permutes if and only if .

This implies Theorem 1.1 in case is odd and is even, since if we put , , and then the polynomial in Proposition 3.1 is congruent mod to the polynomial in Theorem 1.1. Next suppose that is even and is odd, and put , , and . Then one can check that the polynomials from Proposition 3.1 and from Theorem 1.1 satisfy

 f(X)≡g(Xq2−2)(modXq2−X),

so that and are multiplicatively equivalent.

Theorem 1 of [4] is as follows:

###### Proposition 3.2.

Let and be positive integers, and write and . Suppose that , and let and be positive integers such that and . Then permutes .

###### Remark.

The statement of [4, Thm. 1] has the additional hypothesis , but that hypothesis is not used in the proof of that result.

We now show that all the permutation polynomials in Theorem 1.1 are multiplicatively equivalent to permutation polynomials in Proposition 3.2. Assume the hypotheses of Theorem 1.1. Replace by for a positive integer which is large enough so that ; note that this replacement does not change the congruence class of mod , and hence does not affect the truth of the hypotheses or conclusion of Theorem 1.1, while also not affecting the multiplicative equivalence class of the permutation polynomial in Theorem 1.1. Write and , so that the hypotheses of Theorem 1.1 imply that . Writing , we see that the polynomial in Proposition 3.2 satisfies

 g(Xv)=Xv+Xv+rv(q−1)+Xv+sv(q−1).

Since , we have

 rv(q−1) ≡r(Q−R)(q−1)(modq2−1) =r(T−1)R(q−1) ≡TR(q−1)(modq2−1) =Q(q−1),

and likewise

 sv(q−1)≡−R(q−1)(modq2−1).

It follows that the polynomial in Theorem 1.1 satisfies

 g(Xv)≡f(X)(modXq2−X),

so that and are multiplicatively equivalent.

Each of the results [1, Thm. 4.2], [8, Thm. 3.2], and [9, Thm. 1.1] generalizes [4, Thm. 1], and hence includes special cases that are multiplicatively equivalent to each of the permutation polynomials in Theorem 1.1. If is odd and then the same is true of [5, Thm. 1].