REVIEW
A conditional arithmetic obstruction to the prime numbers as a spectrum of a probability measure
Number Theory · Submitted 9 Aug 2026 · 15 references
Pith's verdict: A short, correct conditional proof that, under a plausible prime-difference hypothesis, the primes cannot be the spectrum of any probability measure; the unconditional corollary is modest but the argument is worth knowing.
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
Under the hypothesis that every sufficiently large even integer is a difference of two primes, no Borel probability measure has the prime numbers as a spectrum.
The reading
What carries the argument
The load-bearing object is the sequence $a(k)=|\hat{\mu}(k)|^2$, the squared modulus of the Fourier transform of the measure at integer frequencies. Orthogonality of the exponential basis forces $a(p-q)=0$ whenever $p$ and $q$ are distinct primes, while completeness forces the translational tiling identity $\sum_p a(n-p)=1$ for every integer $n$. The contradiction is produced by evaluating the tiling identity at sufficiently negative odd integers, where all prime differences are large even numbers and hence vanish under ED, leaving $a(n-2)=1$ and therefore $a(k)=1$ for all large odd $k$; summing these ones over infinitely many large odd primes contradicts the identity at $n=0$.
Extended reading notes
Core claim
The central claim is Theorem 1.5: assuming the eventual even difference hypothesis (ED), the set of primes cannot be a spectrum of any Borel probability measure. The proof reduces spectrality to two conditions on $a(k)=|\hat{\mu}(k)|^2$: $a(p-q)=0$ for all distinct primes $p,q$, and $\sum_{p} a(n-p)=1$ for every integer $n$. Under ED, the first condition kills all sufficiently large even arguments, which forces $a(k)=1$ for all sufficiently large odd $k$ through the tiling identity at negative odd $n$; then the identity at $n=0$ demands infinitely many terms equal to $1$, an impossibility. A direct corollary, independent of ED, is that any probability measure with prime spectrum would force infinitely many even integers to be absent from the prime difference set.
Load-bearing premise
The argument hinges on the unproved hypothesis that every sufficiently large positive even integer occurs at least once as a difference of two primes; if that hypothesis fails, the main theorem carries no force and only the unconditional corollary remains.
Editorial extensions
If this is right
- If the Polignac conjecture holds, the prime numbers cannot be the spectrum of any Borel probability measure on $\mathbb{R}$.
- Unconditionally, spectrality of the primes would imply infinitely many positive even integers are missing from $\mathbb{P}-\mathbb{P}$, so the prime spectrum problem cannot have a positive answer unless the prime difference set has infinitely many even holes.
- Any measure admitting the primes as a spectrum, if one exists, must be singular continuous with no atoms and no absolutely continuous part.
- The obstruction is purely combinatorial: it applies to any nonnegative even function on $\mathbb{Z}$ satisfying the zero-set and tiling conditions, not to any special feature of measures beyond their Fourier transforms.
Reading between the lines
- The same two-condition clash may obstruct other sparse sets whose difference set covers all sufficiently large elements of one parity class; the argument uses little beyond infinitude of the set and the oddness of all elements except one.
- A frame spectrum, where Parseval equality weakens to an inequality, escapes this counting argument because the tiling identity no longer holds exactly; Question 4.1 asks whether primes can be a frame spectrum, and the paper's method gives no obstruction there.
- If ED is ever established unconditionally, the nonexistence conclusion would follow in full; if ED fails, the prime spectrum question remains open and would likely require a fundamentally different construction or obstruction.
Original abstract
Let $\Pp=\{2,3,5,7,\ldots\}$ denote the set of prime numbers. We prove that if every sufficiently large positive even integer can be represented as a difference of two primes, then there is no Borel probability measure $\mu$ on $\R$ for which \(\left\{e^{2\pi i p x}:p\in\Pp\right\}\) is an orthonormal basis of $L^2(\mu)$. In particular, under the Polignac conjecture, the prime numbers $\Pp$ cannot be a spectrum (i.e., the set of frequencies of an exponential orthonormal basis) of any probability measure on $\R$.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
Let $\Pp=\{2,3,5,7,\ldots\}$ denote the set of prime numbers. We prove that if every sufficiently large positive even integer can be represented as a difference of two primes, then there is no Borel probability measure $\mu$ on $\R$ for which \(\left\{e^{2\pi i p x}:p\in\Pp\right\}\) is an orthonormal basis of $L^2(\mu)$. In particular, under the Polignac conjecture, the prime numbers $\Pp$ cannot be a spectrum (i.e., the set of frequencies of an exponential orthonormal basis) of any probability measure on $\R$.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Let $\Pp=\{2,3,5,7,\ldots\}$ denote the set of prime numbers. | background 1.00 | 0.12 | 0.02 | asserted 0.93 |
| s1 | We prove that if every sufficiently large positive even integer can be represented as a difference of two primes, then there is no Borel probability measure $\mu$ on $\R$ for which \(\left\{e^{2\pi i p x}:p\in\Pp\right\}\) is an orthonormal basis of $L^2(\mu)$. | result 0.97 | 0.92 | 0.85 | asserted 1.00 |
| s2 | In particular, under the Polignac conjecture, the prime numbers $\Pp$ cannot be a spectrum (i.e., the set of frequencies of an exponential orthonormal basis) of any probability measure on $\R$. | interpretation 0.90 | 0.75 | 0.56 | asserted 0.99 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
References
How the literature uses it
What this paper does with the arXiv papers it cites
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.