Path 5: Cross-Field Analogies (Function Fields & Other Domains) Path

Solve “Goldbach-like” problems in easier settings and port the insights.

Rationale

Often, number theory problems have analogs in algebraic settings where more tools (like geometry or exact formulas) are available. For Goldbach, one prominent analog is in the ring of polynomials over a finite field, $\mathbb{F}_q[T]$. Primes correspond to irreducible polynomials. The Goldbach analog asks: is every polynomial of even degree the sum of two irreducible polynomials? This was proved in function fields (for large $q$) by Hayes (1965) and recently strengthened by Sawin & Shusterman (2022), who solved the twin primes and Goldbach analogs over $\mathbb{F}_q[T]$ using geometric techniques. The function field results suggest that no combinatorial obstruction exists – in those cases the only thing preventing a Goldbach decomposition would be a global symmetry or parity issue, which can be overcome. By studying these analog proofs, one may identify which properties of the function field setting make the conjecture easier (for example, the Riemann Hypothesis for function field zeta functions is true (the Weil conjectures), giving very strong distribution results for irreducible polynomials). Another analogy is replacing “primes” with other subsets of integers: e.g. the practical numbers (numbers whose divisors “cover” all smaller integers). There is a conjecture, now theorem, that every even number is sum of two practical numbers – proven by Melfi (1996). Likewise, one can consider Goldbach’s problem for lucky numbers (a certain sieve-defined sequence); that remains open, but any techniques there might reflect back to primes. By exploring these analogs, researchers can test strategies in a toy environment. The ultimate idea is to find a principle that is true in all these worlds – perhaps some structural reason primes behave like a “sum-heavy” set – and then bring that principle (with new tools) into the integer world.

Prerequisite Themes

Familiarity with abstract algebra (finite fields, polynomial rings); understanding of how primes generalize to other domains (e.g. irreducibles in $\mathbb{F}_q[T]$) and how analytic methods translate (like using the Weil conjectures for polynomial zeta functions); basics of algebraic geometry or function field arithmetic might be needed to read the proofs by Sawin & Shusterman. Also, knowledge of analogous conjectures (twin primes vs. polynomial twin primes, etc.) and their status.

Dependencies

This path stands somewhat apart – it’s more about insight than directly solving the integer problem. However, it can feed into any of the above paths by suggesting what should be true. For example, the function field proof might inform a new approach to the circle method or a new sieve idea by showing how to circumvent a difficulty in a simpler setting. It also heavily depends on Path 3 concepts since function field proofs often use their own RH analog (Weil’s theorem).

Signs of Progress

A clear correspondence between the function field results and integer results – e.g. identifying a certain statistic or identity that holds in function fields and attempting to prove an integer analogue. If one can, say, prove an averaged form of Goldbach for all even $N$ (like an asymptotic formula for the number of representations) by imitating the function field method, that would be major progress. Another sign is if new conjectures come from analogies that can be tested numerically: for instance, the function field proof might hint that any counterexample to Goldbach (if it existed) would have to be astronomically large and of a special form – providing a target for computational search or a narrowing of possibilities (though no one has found any counterexample up to $4\cdot10^{18}$). In summary, progress on this path is measured by cross-pollination: when techniques or theorems from the analog setting lead to partial results or simplified scenarios of Goldbach in the integer setting.

Base Camp 5.1: Goldbach in Function Fields – Understanding the Polynomial Analogue

Scope: Study the case of primes in $\mathbb{F}_q[T]$, irreducible polynomials, and the result that every polynomial of sufficiently large degree $N$ over $\mathbb{F}_q$ (for large $q$) is sum of two irreducibles (the function field Goldbach analog). What you must be able to do: understand the correspondence between integers and monic polynomials, primes and irreducibles, degree and logarithm size, etc. Prove simpler analogues first: e.g. over $\mathbb{F}_q[T]$, prove the weak Goldbach analog (every polynomial of sufficiently large degree is sum of 3 irreducibles) – this can be easier due to the simpler prime distribution (Chebotarev’s theorem in this context is easier). Then examine the hard part: reducing 3 to 2 irreducibles. Sawin & Shusterman’s 2022 paper actually solves twin primes and Goldbach analog by advanced algebraic geometry. While reproducing that is tough, try to glean key ideas: perhaps using the geometry of curves over finite fields and the fact that $\mathbb{F}_q[T]$ has a simpler zeta function (the Weil conjectures proven by André Weil give RH in function fields which helps dramatically).

Stepping-stones: (1) Show using simple combinatorics that over $\mathbb{F}_q$ with $q$ large, the number of irreducible polynomials of degree $n$ is $\approx \frac{q^n}{n}$ (exact formula by prime polynomial theorem). Use that to reason: number of ways to write a polynomial of degree $N$ as sum of two irreducibles of degrees $i$ and $N-i$ is about $\frac{q^i}{i}\frac{q^{N-i}}{N-i}$ – sum that over i and get $\sim N \frac{q^N}{N^2}$ which tends to infinity for large $N$. So random heuristic works even better here (no nearly as strong logarithmic slowdown). (2) Formulate the Goldbach polynomial conjecture and see that for large $q$, it should hold by heuristic and indeed can be proven by a big sieve or something (maybe not trivial but doable with algebraic geometry providing equidistribution of zeros of $L$-functions of curves). (3) Read Hayes 1965 proof for polynomials with integer coefficients – an even closer analog in some ways (though that seems to be a different interpretation, likely an older attempt in a restricted integer scenario or using generating functions). (4) Identify differences: in $\mathbb{F}_q[T]$, there’s a lot more regularity – primes (irreducibles) are fairly well-distributed by degree and no large oscillations, plus one can leverage combinatorial methods because the setting is pseudo-random (there’s even a straightforward sieve in function fields that works better because of polynomial regularity).

Best resources:

Base Camp 5.2: Other Algebraic Analogues – Goldbach in Number Fields or Thin Sets

Scope: Consider Goldbach-type questions in other rings or domains: e.g., the ring of Gaussian integers $\mathbb{Z}[i]$, or more exotic structures. What you must be able to do: understand how “sum of two primes” translates: in $\mathbb{Z}[i]$, one could ask if every sufficiently large Gaussian integer (in norm) is sum of two Gaussian primes? However, unique complexities arise (like units, etc.). Another angle: look at Waring’s problem analogs in additive combinatorics – though Goldbach is about primes (multiplicative concept) rather than powers. Possibly inspect models in random graphs or hypergraphs that simulate constraints similar to primes (like random 2-colorings and monochromatic solutions – akin to Schur’s problem vs. Goldbach’s is like a 2-coloring with primes/composites). Though not directly algebraic, these analogies build intuition.

Stepping-stones: (1) Examine what Goldbach’s conjecture would mean in $\mathbb{Z}[i]$: It actually becomes easier or trivial? For instance, every Gaussian even (divisible by 1+i perhaps) as sum of two Gaussian primes – needs exploring. (2) Consider a simpler number field: $\mathbb{Q}(\sqrt{2})$, etc., and ask if every sufficiently large norm integer is sum of two prime norms? (This might tie to norms being represented by forms.) (3) Understand results like Heegner’s theorem on sum of two squares – which is a different question but highlights that primes splitting in extensions have structure (e.g., an integer prime $p$ splits in $\mathbb{Z}[i]$ if $p=a^2+b^2$, so a “two squares Goldbach” is trivial or false depending on representation). So analogies can break down if the domain’s prime structure is too rigid. (4) Look at thin sequences: e.g., the sequence of prime numbers is thin in the naturals. Goldbach is about sumset of two thin sets (primes + primes). If one had another thin set with some structure (like squares, triangular numbers, etc.), are they additive bases? For squares: “sum of two squares covers many numbers but not all (e.g., $7$ cannot be expressed as sum of two squares of naturals)”. So maybe primes are special due to distribution. Realize any analog likely requires some equidistribution property of the sequence.

Best resources:

Base Camp 5.3: Heuristics from Physics and Randomness – Cross-Disciplinary Insights

Scope: See if methods from statistical physics, random matrix theory, or ergodic theory provide insight. What you must be able to do: interpret prime problems in terms of spectra (via random matrix theory, where zeros of zeta ~ eigenvalues of random matrices, which via Montgomery’s conjecture and Hardy–Littlewood leads to prime pair correlations). If zeros spacings have the predicted GOE distribution, then asymptotically Goldbach’s conjecture should hold (because that would validate the Hardy–Littlewood heuristic fully). So random matrix theory is a cross-field analogy linking quantum physics to primes. Similarly, Bohmian mechanics anecdote (mentioned by the user): not directly relevant, but conceptually – outlandish theories sometimes find vindication decades later. Perhaps someone might find a physical model where primes naturally arise (there have been attempts like Polya’s suggestion “the Riemann zeros have a spectral interpretation”). Being open to that is worthwhile.

Stepping-stones: (1) Learn the Montgomery pair correlation conjecture: how the non-trivial zeros of zeta have pair correlation $1 - (\sin \pi x / (\pi x))^2$. Understand that if true, it implies strong results about primes (via explicit formulas) like the distribution of prime gaps and also subtle cancellation that underpins the Hardy–Littlewood constants. (2) See Odlyzko’s numerical experiments on zeta zeros – how they match random matrix eigenvalues. This builds confidence in the spectral approach. (3) Check if any analogies from dynamical systems exist: e.g. primes as “skipping orbits” mod 1 (there’s a viewpoint in ergodic theory about equidistribution mod 1 of sequences related to primes). (4) Explore any literature on using the “Ising model” or other physical systems to simulate primes (a bit far-fetched, but there have been amusing attempts to encode primes in ground states of some systems – not particularly fruitful yet).

Best resources:

Foundational across camps: Rosen’s Number Theory in Function Fields and the Sawin–Shusterman paper are key references repeated for Path 5, as they thoroughly address the function field analog which is arguably the crown jewel of this path’s approach.

Full Bibliography (Path 5)

BC5.1 Function Fields:

BC5.2 Other Analogues:

BC5.3 Physics & Randomness:

Each resource above is chosen for clarity or necessity. Classic sources (Hardy–Littlewood 1923, Schnirelmann 1930 in secondary references, etc.) ground the historical context; modern papers (Helfgott 2013 on arXiv, Sawin–Shusterman 2022) show cutting-edge progress. This bibliography will guide a comprehensive study regime and support the construction of a multi-pronged attack on the “Kangchenjunga” of number theory – Goldbach’s conjecture.

← Back to Kangchenjunga – Goldbach’s Conjecture