Gaussian Integers, Rings, Finite Fields, and the Magic Square of Squares

Onno Cain

REVIEW 4 major objections 5 minor 10 references

Gaussian Integers, Rings, Finite Fields, and the Magic Square of Squares

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The 3x3 magic-square-of-squares problem is equivalent to a quartic factorization problem over an abelian extension of the rationals, and the smallest finite field with such a square is F29.

desk verdict A clever reformulation of the magic square problem whose main finite-field classification theorem has a fixable but real counting gap. read the letter →

arxiv 1908.03236 v2 pith:UKZ2H7CA submitted 2019-08-08 math.RA math.NT

Onno Cain This is my paper · ORCID

classification math.RAmath.NT MSC 05B1511R0411R3212E20

keywords magicsquareofsquaresParkerhourglassfinitefieldsGaussianintegersabelianextensionquarticpolynomialringsZ/nZ

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper attacks a long-standing unsolved question—whether nine distinct squared integers can form a 3 by 3 magic square—by translating it into algebra. Its main equivalence, Theorem 3.2, says that a magic hourglass of squares exists exactly when three parameters $\alpha,\beta,\gamma$ in $\mathbb{Z}[i,\sqrt{A},\sqrt{B},\sqrt{C}]$ have equal norms and satisfy $\alpha^4+\beta^4+\gamma^4\in\mathbb{Z}$, with the three fourth powers distinct and non-real. In the special case of the Gaussian integers, unique factorization turns this into a search condition on $\operatorname{Im}[x^4y^4z^4]$. The paper then moves to finite fields, defines a field as Parker when it contains no magic square of nine distinct squares, and proves that $\mathbb{F}_{29}$ is the smallest non-Parker field, with conjectures enumerating all Parker fields and rings of the form $\mathbb{Z}/n\mathbb{Z}$. The finite-field and ring results matter because a proof of infinitely many Parker rings would settle the original integer problem negatively.

What carries the argument

The load-bearing object is the map $\chi(\omega)=(\operatorname{Re}[\omega^2]+\operatorname{Im}[\omega^2],\,\omega\overline{\omega},\,\operatorname{Re}[\omega^2]-\operatorname{Im}[\omega^2])$, which packages one Gaussian-type number into the three entries of a square arithmetic progression $r^2,s^2,t^2$ with $r^2+t^2=2s^2$. Lemma 3.1 shows every integer solution of the congruum equation arises this way, so an hourglass is three such parameters sharing one norm. In the Gaussian case the identity $\operatorname{Im}[x^4y^4z^4]=-4\operatorname{Im}[x^4]\operatorname{Im}[y^4]\operatorname{Im}[z^4]$ is what makes $\alpha^4+\beta^4+\gamma^4$ real. Over finite fields, Lemma 5.3 reduces Parker-ness to the solution count of $x^2+y^2=0$ or $x^2+y^2=2$, Lemma 5.4 parametrizes central-zero squares by three consecutive squares, and Algorithm 6.1 turns these observations into an exhaustive search.

What would settle it

Run Algorithm 6.1 directly on $\mathbb{F}_{23}$: if it returns any nine distinct squares forming a magic square, then $\mathbb{F}_{23}$ is non-Parker and Theorem 5.1 is false. Independently, enumerate all ordered pairs $(x,y)$ in $\mathbb{F}_{23}$ with $x^2+y^2=2$; four or more such pairs with $x^2,y^2\neq 2$ would satisfy Lemma 5.3's condition and invalidate the proof of Corollary 5.3 as written.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery has two parts. The first is the hourglass-to-quartic reduction: a magic hourglass of squares exists if and only if there are $\alpha,\beta,\gamma$ in $K=\mathbb{Z}[i,\sqrt{A},\sqrt{B},\sqrt{C}]$ such that $\alpha\overline{\alpha}=\beta\overline{\beta}=\gamma\overline{\gamma}$, the three fourth powers $\alpha^4,\beta^4,\gamma^4$ are distinct and strictly complex, and $\alpha^4+\beta^4+\gamma^4$ is an integer. The second is the finite-field classification up to order 29: every finite field of order smaller than 29 is Parker, while $\mathbb{F}_{29}$ is not Parker, witnessed by the explicit square given in Section 5; the paper further conjectures that exactly 17 finite fields are Parker and that $\mathbb{Z}/3216\mathbb{Z}$ is the largest Parker ring of the form $\mathbb{Z}/n\mathbb{Z}$.

Load-bearing premise

The proof of Theorem 5.1 rests on Lemma 5.3's claim that a non-Parker field must have four distinct solutions of $x^2+y^2=0$ or $x^2+y^2=2$, together with the hand count of those solutions for $\mathbb{F}_{19}$, $\mathbb{F}_{23}$, and $\mathbb{F}_{27}$; if that count is interpreted differently, the exclusion of $\mathbb{F}_{23}$ falls apart.

Editorial extensions

If this is right

If Theorem 3.2 is right, an integer magic square of squares exists exactly when a quartic reality condition can be met by three equal-norm elements of an abelian extension of the rationals, so the integer search becomes a factorization search.
If Theorem 5.1 is right, the finite-field obstruction begins precisely at order 29: all fifteen smaller finite fields are Parker, and the displayed square modulo 29 is the minimal example.
The Gaussian-integer case yields a concrete search rule: pick a Gaussian integer divisible by $2^7 3^2$ and test its factorizations $xyz$ against the four-imaginary-part identity to look for magic hourglasses.
If the enumeration conjectures are right, exactly 17 finite fields are Parker, so almost every finite field contains a magic square of nine distinct squares.
If Conjecture 9.4 is right, $\mathbb{Z}/3216\mathbb{Z}$ is the largest Parker ring of the form $\mathbb{Z}/n\mathbb{Z}$; since an integer solution would make all sufficiently large $n$ non-Parker, infinitely many Parker rings would rule out an integer solution.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper leaves the converse of Theorem 4.1 open; a natural extension is to search over Gaussian integers satisfying the four-imaginary-part identity and check whether the resulting hourglasses complete to full squares, which would test whether the sufficient condition is also necessary.
Because $\mathbb{F}_p$ contains a square root of $-1$ when $p\equiv 1\pmod 4$, the Gaussian-integer identity can be transplanted to finite fields; running the same factorization search there could yield non-Parker fields that the paper's conjecture list does not include.
The record-count table for rings suggests the count of magic squares up to scaling may be governed by the multiplicative structure of $\mathbb{Z}/n\mathbb{Z}$; checking whether the count is monotone along divisors of $n$ would sharpen Conjecture 9.2 into a structural statement.
A direct recomputation of the solution counts in Corollary 5.3, using ordered rather than unordered pairs, would independently confirm the hand count on which Theorem 5.1 relies; the paper's own Algorithm 6.1 is already set up for exactly this check.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

This paper is worth a read, but the headline result needs work. The genuinely new piece is Theorem 3.2: a clean equivalence between the magic hourglass of squares and a quartic condition over Gaussian-integer-like rings. That's a fresh coordinate system for a classic problem, and the observation that infinitely many Parker rings would rule out the integer solution is correct and useful. The even-characteristic lemma (5.1) is solid, and the computational tables in Sections 7 and 9 are a reasonable addition to the literature, assuming the Sage code checks out.

The soft spot is the proof of Theorem 5.1, and it's a real one. Lemma 5.3 asks for at least four distinct solutions to x^2+y^2=2 with neither square equal to 2, but the proof only shows the four line-pairs of a scaled magic square satisfy the equation; it never shows none of those pairs contains a zero (which forces the other entry to be 2). So the lemma's extra condition is unproved. Then Corollary 5.3 counts solutions ambiguously: on the natural ordered-pair reading, F23 alone has more than four valid solutions (from 3^2+4^2=2, with signs), which contradicts the corollary; on the unordered-pair reading, the lemma's condition still isn't justified. Either way, the published elimination of F19, F23, and F27 doesn't go through. The computation likely supports the theorem, but the written argument doesn't.

Two theorems are left with "Proof: Left undone" (2.2 and 4.2), which is a rigor flag, and the promised Gaussian search method isn't demonstrated past the idea stage. The conjectures in Sections 7 and 9 are explicitly empirical, which I find honest rather than a flaw. Citation-wise, the paper engages the relevant literature and points to Labruna's thesis where appropriate.

Who is it for? Anybody thinking about arithmetic progressions of squares over finite fields or rings, or looking for a new angle on the magic square problem. The reformulation and data are worth having; the proof gap is local. I'd send it to a referee rather than desk-reject, with a clear request to fix the counting lemma and supply or cite the missing proofs. A heavy-revision accept, not a flat reject.

Referee Report

4 major / 5 minor

Summary. The paper studies the 3x3 magic square of squares problem. It first reformulates the 'magic hourglass' subproblem as a quartic-factorization condition over an abelian extension of the rationals (Theorem 3.2), then specializes to the Gaussian integers to obtain a sufficient condition for an hourglass (Theorem 4.1). It then changes setting and asks over which finite fields and rings Z/nZ a 3x3 magic square of distinct squares exists. The main finite-field claim is Theorem 5.1: F29 is the smallest non-Parker finite field. Sections 6-9 present search algorithms and computational observations leading to conjectures about Parker fields and Parker rings, with code made available.

Significance. The paper proposes a genuinely new parametrization of the magic square of squares problem (Theorem 3.2) and gives an explicit finite-field construction over F29. It also provides reproducible Sage code and a clean computational framework for the finite-field and ring questions. If the Section 5 classification proof were made rigorous, the results would be a useful contribution to a well-known open problem. However, the printed proof of Theorem 5.1 depends on several load-bearing counting and parametrization steps that are not valid as written. The conjectures in Sections 7 and 9 are honest and clearly labelled, and the absence of fitted parameters or circular assumptions is a strength.

major comments (4)

[Section 5, Lemma 5.3 and Corollary 5.3] The elimination of F19, F23, and F27 rests on an ambiguous and internally inconsistent count of solutions to x^2 + y^2 = 2. If 'solutions' means ordered pairs (x,y) in F_q^2, then the corollary is false: in F23, 3^2 + 4^2 = 2 alone gives the eight ordered pairs (3,4), (3,19), (20,4), (20,19) and their swaps, and the full solution set has at least 20 ordered pairs, contradicting the claim that no field has 4 distinct solutions. If 'solutions' instead means unordered pairs of square values, then the condition in Lemma 5.3 that x^2,y^2 != 2 is not derived in its proof; moreover, the F23 list in Corollary 5.3 itself contains the pair 0^2 + 5^2 = 2, where 5^2 = 2, so one of the listed solutions violates the lemma's own exclusion condition. The proof of Lemma 5.3 does not rule out a magic square whose center is 1 and whose four lines through the center use the pair {0,2}. Under no stated convention does the printed argument validly establish that F19, F23, and F27 are Parker. This gap is load-bearing for Theorem 5.1.
[Section 5, Lemma 5.4 and Corollary 5.4] The proof that F25 is Parker is factually wrong. Corollary 5.4 states that the squares of F25 'manifestly have no three consecutive squares (excluding 0 and ±1)', but the listed square set includes 0, 1, 2, 3, and 4, so 2, 3, 4 are three consecutive square values. Taking gamma^2 = 2, beta^2 = 3, alpha^2 = 4 satisfies alpha^2 - beta^2 = beta^2 - gamma^2 = 1, and the square roots can be chosen outside {0, 1, -1}. This satisfies the condition of Lemma 5.4 as stated, yet the constructed square has repeated entries in characteristic 5 (for example -4 = 1 and -1 = 4). Thus Lemma 5.4's stated parametrization condition is insufficient to guarantee distinctness, and the corollary's conclusion that F25 is Parker does not follow. The F25 part of the proof of Theorem 5.1 therefore needs a correct argument, not merely the asserted inspection of the square set.
[Section 5, Lemma 5.1] The parametrization of integer 3x3 magic squares used to justify the enumeration of the eight F2 squares is incorrect as printed. The first displayed summand has entries (A, -A, A; -A, 0, A; 0, A, -A), whose row sums are A, 0, and 0; it is not a magic square unless A = 0. The same issue propagates through the full three-summand formula. The list of eight F2 squares may be correct, but the cited parametrization cannot serve as the proof that the list is exhaustive. Since Lemma 5.1 is used to eliminate all finite fields of even order, this proof gap also affects Theorem 5.1.
[Section 2, Theorem 2.2; Section 4, Theorem 4.2] Theorem 2.2 is stated with 'Proof: Left undone', but it is the foundation of Lemma 3.1 and hence of Theorem 3.2, the paper's first central claim. Theorem 4.2 also has 'Proof: Left undone'. The results may be classical or easily provable, but as written the paper does not supply a proof or a precise citation for either. A central equivalence theorem that depends on an unproved stated theorem is not complete; the authors should either prove Theorem 2.2 or cite a standard source with the exact parametrization, and either prove Theorem 4.2 or relegate it to a conjecture/remark.

minor comments (5)

[Section 5, Corollary 5.1] The proof says 'By Lemma 3.1' but the intended reference is Lemma 5.1.
[Section 5, Corollary 5.4] The set of squares of F25 is written with duplicate entries in a set, and the element x is not defined; the field F25 should be given by an explicit irreducible polynomial, e.g. F5[t]/(t^2 - 2), and the nine square values should be listed as a genuine set of distinct elements.
[Section 3, Observation 3.1] The parenthetical '(we think?)' is not appropriate for a formal proof; Observation 3.1 should be proved, removed, or clearly labelled as a question.
[Section 3, Theorem 3.2] The ring K is defined first as Z[i, sqrt(A), sqrt(B), sqrt(C)] but Observation 3.1 immediately switches to Q(i, sqrt(A), sqrt(B), sqrt(C)); the relationship between the ring and the field should be clarified.
[Section 7, Observation 7.2] The table header says 'Field' but the entries are finite fields F_p; writing 'F_p' in the table would avoid ambiguity with the word 'field'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivations are forward proofs from definitions, external classical results, and explicit computations, with no fitted parameter renamed as prediction.

full rationale

All central claims are proved from definitions and external facts rather than from the conclusions they establish. Theorem 3.2 rewrites the magic-hourglass condition into a Gaussian-integer factorization condition, but the proof constructs the auxiliary elements from the square progressions and verifies the equivalence in both directions; this is a genuine mathematical equivalence, not a self-definitional loop. Theorem 4.1 is a conditional existence statement proved by an explicit identity and does not presuppose the existence of a magic hourglass. The finite-field classification in Section 5 is supported by explicit enumerations and computations; Lemma 5.3 is a necessary condition, and Corollaries 5.3 and 5.4 apply it by counting solutions, which is empirical verification rather than circular reasoning. The conjectures in Sections 7 and 9 are presented as computational observations and are not disguised predictions of fitted inputs. The note that infinitely many Parker rings would imply no integer magic square of squares is a logical implication, not a circular step. No self-citation is load-bearing in the argument, and no uniqueness theorem is imported from the author's prior work. Any weaknesses in the proof of Theorem 5.1 concern the rigor of the solution-counting argument, which is a correctness issue rather than circularity.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or mathematical entities; 'Parker' is a definitional label for a field or ring with no magic square of nine distinct squares. The central results rest on standard algebraic number theory and finite field facts, plus two unproven structural assumptions about scaling and solution counts.

assumptions (4)

ad hoc to paper A non-Parker finite field must contain at least 4 distinct solutions to x^2+y^2=0 or x^2+y^2=2 with the stated exclusions, and fewer than 4 such solutions implies Parker.
This is the load-bearing step in Corollary 5.3. The paper's counting of solutions is inconsistent with the lemma's statement, so this is an unproven assumption rather than a derived fact.
domain assumption Any magic square with central entry 0 can be scaled so that a corner is 1 and the opposite corner is -1 (Lemma 5.4).
Relies on corners having nonzero squares in a distinct-squares magic square with central entry 0; the paper asserts it without proof. If a corner square were 0, scaling would be impossible.
standard math The 8 matrices listed in Lemma 5.1 are all 3x3 magic squares over F2, and every magic square over F_{2^k} decomposes as a sum of these.
Follows from linearity and the standard 3-parameter parametrization of magic squares cited to [8]; checkable but not proven in detail here.
standard math Unique factorization in Z[i] (cited to [6]) and cyclic multiplicative groups of finite fields (cited to [7]).
Standard background invoked in Sections 4 and 5 for the Gaussian integer parametrization and for counting squares in finite fields.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Gaussian Integers, Rings, Finite Fields, and the Magic Square of Squares." pith.science (2026). https://pith.science/paper/UKZ2H7CA

@misc{pith2026190803236,
  author       = {Pith},
  title        = {Pith review of: Gaussian Integers, Rings, Finite Fields, and the Magic Square of Squares},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UKZ2H7CA}},
  note         = {Machine review of arXiv:1908.03236}
}

read the original abstract

We show the 3 by 3 magic square of squares problem equivalent to solving quartic polynomials with certain factorization constraints over an abelian extension of the rationals. We analyze a particular case in which said extension is assumed to be the Gaussian integers resulting a new search method. Additionally, the magic square of squares is analyzed over finite fields and rings of the form Z/nZ resulting in some conjectures enumerating the rings and finite fields in which a magic square of squares can be constructed. Code is made available.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

10 extracted references · 10 canonical work pages

[1]

Christian Boyer, Magic Square of Squares , http://www.multimagie.com/English/SquaresOfSquares.htm

work page
[2]

Matt Parker, The Parker Square , Numberphile, [interview by Brady Haran] , https://www.youtube.com/watch?v=aOT bG-vWyg

work page
[3]

Robertson, Magic Squares of Squares , Mathematics Magazine, vol

John P. Robertson, Magic Squares of Squares , Mathematics Magazine, vol. 69, no. 4, 1996, pp. 289–293. JSTOR, www.jstor.org/stable/2690537

work page arXiv 1996
[4]

MathWorld, Congruum Problem , http://mathworld.wolfram.com/CongruumProblem.html

work page
[5]

Eknath Ghate, The Kronecker-Weber Theorem, Summer School on Cyclotomic ﬁelds, Pune, June 7-30, 1999

work page 1999
[6]

Keith Conrad, The Gaussian Integers , https://kconrad.math.uconn.edu/blurbs/

work page
[8]

Math Pages, Magic Square of Squares , https://www.mathpages.com/home/kmath417/kmath417.htm

work page
[9]

SageMath, the Sage Mathematics Software System (Version 8.4 ), The Sage Developers, 2015, http://www.sagemath.org

work page 2015

Show all 10 references

[10]

Giancarlo Labruna, Magic Squares of Squares of Order Three Over Finite Fields , https://digitalcommons.montclair.edu/etd/138/
[11]

Code repository: https://github.com/onnomc/parker-ring- search 15

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Christian Boyer, Magic Square of Squares , http://www.multimagie.com/English/SquaresOfSquares.htm

work page

[2] [2]

Matt Parker, The Parker Square , Numberphile, [interview by Brady Haran] , https://www.youtube.com/watch?v=aOT bG-vWyg

work page

[3] [3]

Robertson, Magic Squares of Squares , Mathematics Magazine, vol

John P. Robertson, Magic Squares of Squares , Mathematics Magazine, vol. 69, no. 4, 1996, pp. 289–293. JSTOR, www.jstor.org/stable/2690537

work page arXiv 1996

[4] [4]

MathWorld, Congruum Problem , http://mathworld.wolfram.com/CongruumProblem.html

work page

[5] [5]

Eknath Ghate, The Kronecker-Weber Theorem, Summer School on Cyclotomic ﬁelds, Pune, June 7-30, 1999

work page 1999

[6] [6]

Keith Conrad, The Gaussian Integers , https://kconrad.math.uconn.edu/blurbs/

work page

[7] [8]

Math Pages, Magic Square of Squares , https://www.mathpages.com/home/kmath417/kmath417.htm

work page

[8] [9]

SageMath, the Sage Mathematics Software System (Version 8.4 ), The Sage Developers, 2015, http://www.sagemath.org

work page 2015