Sums of two coprime cubes of the form 2^a 3^b C^p
Year: 2026
Type: preprint
Abstract: I study A^3 + B^3 = c C^p in coprime nonzero integers, for c = 2^a 3^b with 0 <= a <= 5, 0 <= b <= 2 and p >= 5 prime, by the modular method with the Frey curve of Kraus. For c in {16,32,48,96,144,288}, and for every c when C is even (for c = 1, p >= 17 due to Darmon and Granville), there is no solution. For c in {4,8} there is no solution with 3 not dividing C. For c in {3,4,6,8,12,18,24,36} solutions are excluded for p in explicit congruence classes by the symplectic criteria of Freitas and Kraus. A computer elimination with auxiliary primes, in the refined form of Jaswal, shows that for c in {3,4,6,8,12,18,24,36,72} there is no solution for 7 <= p <= 1000, and, for c != 3, none for p = 5; for (c,p) = (24,5) the proof is a descent to y^2 = x^5 - 2^4 3^9 (Jacobian of rank 0). (18,7) has a second proof by descent to Y^2 = X^7 - 3 and Chabauty. The coefficient 4: A^3 + B^3 = 4C^5 and 4C^7 have no solution (p = 7 with 3 not dividing C), and the only primitive solutions of y^3 + z^3 = 4w^4 are (+-1,-+1,0). Status: many steps are computer-assisted (PARI/GP 2.19.0 and 2.17.3, SageMath 10.7; an independent Magma Online check for (24,5)) and are marked [computational] in the text; the scripts are in code.zip. Case A of y^3 + z^3 = 4w^4 is formalized in Lean 4 (O1CaseA.lean, standard axioms). The elimination code has been checked by independent re-runs but not by a second human. Prior art: the row c = 3 is due to Jaswal (preprint), who solves x^3 + y^3 = 3z^n for all n >= 3, so my results for c = 3 are not new; the case c = 1 is due to Darmon and Granville; the method rests on Kraus, Darmon-Granville, Freitas-Kraus, Bruin, Stoll and Siksek. The value c = 9 is not treated, and for c = 2 only solutions with C even or 3 | C are treated.
Author: Bali, Mohammad
Language: English
Subfield: Geometry and Topology
Field: Mathematics
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7229748742
