The adelic five-point string amplitude: exact form and admissible regularization of the product over all places

For every prime $p$, evaluating the Koba-Nielsen integral of open-string tachyon scattering over the p-adic numbers instead of the reals defines a p-adic string amplitude, and at four points the Veneziano amplitude times the product of all p-adic amplitudes formally reduces to a constant via the Euler product for the Riemann zeta function. Marinari and Parisi computed the p-adic five-point amplitude in closed form and found that the analogous adelic product is formally divergent (five powers of the divergent factor $\zeta(1)$) and that, after removal of the divergent factors, the residual finite product still fails to be a constant because its uncanceled zeros are not matched by the real five-point amplitude. The open problem is to determine this adelic five-point amplitude: give a regulator-independent, or canonically regularized, value to the product over all places at generic kinematics; determine its exact functional form; and establish whether it admits a physically consistent interpretation. Jepsen (2023) evaluated the product exactly, in terms of ratios of the Riemann zeta function, in nine special (unphysical) regimes where it converges without a regulator, showed the result is piecewise analytic and non-constant, and showed that regularizations based on splitting the infinite product into separately analytically continued subproducts are generally invalid. The determination at generic kinematics and a principled regularization remain open.

Background

A string scattering amplitude has a remarkable arithmetic shadow. For a fixed prime number $p$, the field $\mathbb{Q}_p$ of p-adic numbers is the completion of the rationals with respect to the p-adic norm $|x|_p=p^{-n}$, where $p^n$ is the highest power of $p$ dividing $x$; the adeles are the ring that combines the real numbers with all the $\mathbb{Q}_p$ at once, and one speaks of the various fields $\mathbb{Q}_v$ ($v=\infty$ or $v=p$) as the "places" of the rationals. Volovich, and independently Freund and Olson, proposed in 1987 to build string amplitudes by evaluating the Koba-Nielsen integrals — the standard integral representations of open-string tachyon amplitudes (the tachyon being the lowest-lying string excitation), in which the integrand is a product of powers of distances between vertex positions on the boundary of the string worldsheet — over $\mathbb{Q}_p$ rather than over the reals, using the p-adic norm and the translation-invariant Haar measure (the standard measure on a locally compact field). Spacetime coordinates and momenta stay real; only the worldsheet variables become p-adic. Zabrodin showed that the resulting worldsheet is the Bruhat-Tits tree, a $(p+1)$-valent tree whose boundary is $\mathbb{Q}_p$. The p-adic amplitudes are far simpler than the real ones — only finitely many poles, all tachyonic — which made the p-adic string a useful toy model.

At four points, with Mandelstam invariants $s$ and $t$ (the Lorentz-invariant scalar products of the external momenta), the p-adic amplitude evaluates to a product of local zeta functions $\zeta_p(z)=(1-p^{-z})^{-1}$, the Euler factors of the Riemann zeta function $\zeta(z)=\prod_p \zeta_p(z)$, an Euler product convergent for $\mathrm{Re}\,z>1$. Freund and Witten observed that the real Veneziano amplitude — the four-point open-string tachyon amplitude, a sum of Euler beta functions $B(a,b)=\int_0^1 t^{a-1}(1-t)^{b-1}\,dt$ — can itself be written in terms of $\zeta$, so that formally

$$A_4^{(\infty)}(s,t)\times\prod_{p\in\mathbb{P}} A_4^{(p)}(s,t) \;\leftrightarrow\; \text{constant},$$

an "adelic product formula" equivalent to the functional equation of $\zeta$. This suggested that the real amplitudes might be recovered from, or arithmetically constrained by, their p-adic counterparts.

Marinari and Parisi tested this at five points. The five-point Koba-Nielsen amplitude over a place $v$ is the double integral

$$A_5^{(v)}=\int_{\mathbb{Q}_v\times\mathbb{Q}_v} dx\,dy\;|x|_v^{s_{12}}\,|y|_v^{s_{13}}\,|y-x|_v^{s_{23}}\,|1-y|_v^{s_{34}}\,|1-x|_v^{s_{24}},$$

with $s_{ij}=2\alpha' k_i\cdot k_j$ built from the five external momenta $k_i$ ($\alpha'$ is the inverse string tension); over the reals it evaluates to a sum over channel orderings of Euler beta-function terms. Marinari and Parisi computed the p-adic version in closed form (independently also obtained by Brekke, Freund, Olson and Witten): a finite combinatorial sum of the elementary terms $1/(p^{1+s_{ij}}-1)$ with Haar-measure counting factors. They then factorized the local amplitude into a rapidly convergent remainder and explicit Euler-type factors, and found that the prime product of those factors formally produces five powers of $\zeta(1)$ — divergent, since $\zeta$ has a pole at $1$ — and that, after the divergent factors are discarded by hand, the residual finite product develops zeros near the symmetric kinematic point that are not matched by any zero or pole of the real amplitude. Hence the adelic five-point product cannot be a constant, contrary to the four-point identity. The paper left open whether a canonical, regulator-independent prescription gives the five-point prime product a finite value; whether the divergence and the uncanceled zeros are artifacts of the Koba-Nielsen starting point or a genuine obstruction; and what the exact functional form of the combined product is. The conclusion was immediately contested: Frampton, Okada and Ubriaco criticized the arguments against the five-particle adelic formula and argued that a simple regularization makes the adelic five-point amplitude equal to one. Jepsen recently showed that in certain kinematic regimes (all unphysical: one pair of momenta orthogonal, with sign conditions on the remaining invariants) the five-point prime product converges without any regulator and equals explicit ratios of the Riemann zeta function; that the resulting amplitude is piecewise analytic, neither constant nor a single analytic function; and that any regularization built by splitting the infinite product into separately continued subproducts — the mechanism behind the $\zeta(1)^5$ bookkeeping — is generally invalid, because redistributing a conditionally convergent infinite product into separately analytically continued subproducts can change its value (a multiplicative-anomaly-type failure). At generic kinematics the product is still formally divergent, and its regularized value is undetermined.

Problem Statement

Consider the five-point open-string tachyon amplitude in Koba-Nielsen form over a local field,

$$A_5^{(v)}(s)=\int_{\mathbb{Q}_v\times\mathbb{Q}_v} dx\,dy\;|x|_v^{s_{12}}\,|y|_v^{s_{13}}\,|y-x|_v^{s_{23}}\,|1-y|_v^{s_{34}}\,|1-x|_v^{s_{24}},$$

where $v=\infty$ denotes the real place, $v=p$ runs over the prime numbers (the set of all primes is denoted $\mathbb{P}$), $|\cdot|_v$ is the ordinary absolute value or the p-adic norm accordingly, $dx\,dy$ is the translation-invariant Haar measure on $\mathbb{Q}_v\times\mathbb{Q}_v$, and the dimensionless invariants $s_{ij}=2\alpha' k_i\cdot k_j$ are built from the five conserved external tachyon momenta $k_i$, the tachyon being the lowest-lying string excitation and $\alpha'$ the inverse string tension (equivalently, in Marinari-Parisi's conventions, $p_{ij}=2p_i\cdot p_j$; over the reals the double integral evaluates to a sum over channel orderings of Euler beta-function terms). The task is to determine the adelic five-point amplitude

$$\Big\{\prod_{p\in\mathbb{P}} A_5^{(p)}(s)\Big\}\times A_5^{(\infty)}(s),$$

namely:

1. Provide a regulator-independent — or canonically regularized — value for this formally non-convergent infinite product at generic values of the kinematic invariants $s_{ij}$, that is, away from the codimension-one regimes in which one pair of external momenta is orthogonal (some $s_{ij}=0$). The prescription must be stated explicitly, with its domain of validity delimited. A prescription that works by splitting the product into separately analytically continued subproducts must be justified against the known multiplicative-anomaly obstruction: redistributing a conditionally convergent infinite product into separately analytically continued subproducts can change its value.

2. Determine the exact functional form of the resulting amplitude, including any finite normalization factors and its dependence on the invariants $s_{ij}$.

3. Establish the qualitative character of the result — constant, piecewise-analytic function, or ill-defined — and whether it admits a physically consistent interpretation, comparing with the four-point adelic identity (the Freund-Witten constant-product identity for the four-point amplitude) and with the demands of factorization and unitarity.

These three deliverables jointly constitute a single objective: the determination of the adelic five-point string amplitude. The determination must account for regularization schemes that do not rest on illegitimate splitting of the infinite product. A construction yields a determination of this same adelic product only if the local amplitudes it induces agree with the Koba-Nielsen p-adic integrals above — possibly up to momentum-independent normalization coefficients whose choice is justified from first principles; a construction whose induced local amplitudes differ from the Koba-Nielsen integrals does not by itself answer the question, however precisely the difference is documented.

References

1. I.V. Volovich, "p-adic string", Classical and Quantum Gravity 4 (1987) L83-L87. DOI: 10.1088/0264-9381/4/4/003, https://doi.org/10.1088/0264-9381/4/4/003
2. P.G.O. Freund and M. Olson, "Non-archimedean strings", Physics Letters B 199 (1987) 186-190. DOI: 10.1016/0370-2693(87)91356-6, https://doi.org/10.1016/0370-2693(87)91356-6
3. P.G.O. Freund and E. Witten, "Adelic string amplitudes", Physics Letters B 199 (1987) 191-194. DOI: 10.1016/0370-2693(87)91357-8, https://doi.org/10.1016/0370-2693(87)91357-8
4. E. Marinari and G. Parisi, "On the p-adic five-point function", Physics Letters B 203 (1988) 52-54. DOI: 10.1016/0370-2693(88)91569-9, https://doi.org/10.1016/0370-2693(88)91569-9
5. L. Brekke, P.G.O. Freund, M. Olson and E. Witten, "Non-archimedean string dynamics", Nuclear Physics B 302 (1988) 365-402. DOI: 10.1016/0550-3213(88)90207-6, https://doi.org/10.1016/0550-3213(88)90207-6
6. I.Ya. Aref'eva, B.G. Dragović and I.V. Volovich, "On the adelic string amplitudes", Physics Letters B 209 (1988) 445-450. DOI: 10.1016/0370-2693(88)91171-9, https://doi.org/10.1016/0370-2693(88)91171-9
7. P.H. Frampton and Y. Okada, "p-adic string N-point function", Physical Review Letters 60 (1988) 484-486. DOI: 10.1103/PhysRevLett.60.484, https://doi.org/10.1103/PhysRevLett.60.484
8. P.H. Frampton, Y. Okada and M.R. Ubriaco, "On adelic formulas for the p-adic string", Physics Letters B 213 (1988) 260-262. DOI: 10.1016/0370-2693(88)91757-1, https://doi.org/10.1016/0370-2693(88)91757-1
9. L. Brekke, P.G.O. Freund, E. Melzer and M. Olson, "Adelic string N-point amplitudes", Physics Letters B 216 (1989) 53-58. DOI: 10.1016/0370-2693(89)91368-3, https://doi.org/10.1016/0370-2693(89)91368-3
10. P. Ruelle, E. Thiran, D. Verstegen and J. Weyers, "Adelic string and superstring amplitudes", Modern Physics Letters A 4 (1989) 1745-1752. DOI: 10.1142/S0217732389001970, https://doi.org/10.1142/S0217732389001970
11. A. Zabrodin, "Non-archimedean string action and Bruhat-Tits trees", Modern Physics Letters A 4 (1989) 367-374. DOI: 10.1142/S0217732389000447, https://doi.org/10.1142/S0217732389000447
12. L. Brekke and P.G.O. Freund, "p-adic numbers in physics", Physics Reports 233 (1993) 1-66. DOI: 10.1016/0370-1573(93)90043-D, https://doi.org/10.1016/0370-1573(93)90043-D
13. V.S. Vladimirov, "On the Freund-Witten adelic formula for Veneziano amplitudes", Letters in Mathematical Physics 27 (1993) 123-131. DOI: 10.1007/BF00750680, https://doi.org/10.1007/BF00750680
14. V.S. Vladimirov and T.M. Sapuzhak, "Adèlic formulas for string amplitudes in fields of algebraic numbers", Letters in Mathematical Physics 37 (1996) 233-242. DOI: 10.1007/BF00416026, https://doi.org/10.1007/BF00416026
15. V.S. Vladimirov, "Regularized adelic formulas for string and superstring amplitudes in one-class quadratic fields", Theoretical and Mathematical Physics 164 (2010) 1101-1109. DOI: 10.1007/s11232-010-0088-1, https://doi.org/10.1007/s11232-010-0088-1
16. B. Dragovich, "Towards effective Lagrangians for adelic strings", Fortschritte der Physik 57 (2009) 546-551. DOI: 10.1002/prop.200900030, https://doi.org/10.1002/prop.200900030
17. M. Bocardo-Gaspar, H. García-Compeán, E.Y. López and W.A. Zúñiga-Galindo, "Local Zeta Functions and Koba-Nielsen String Amplitudes", Symmetry 13 (2021) 967. DOI: 10.3390/sym13060967, https://doi.org/10.3390/sym13060967
18. C.B. Jepsen, "Adelic amplitudes and intricacies of infinite products", Nuclear Physics B 987 (2023) 116094. DOI: 10.1016/j.nuclphysb.2023.116094, arXiv:2211.01611, https://arxiv.org/abs/2211.01611
19. A. Huang and C. Jepsen, "One-loop p-adic string theory and the Néron local height function", arXiv:2604.00970 (2026), https://arxiv.org/abs/2604.00970