n conjecture
In-game article clicks load inline without leaving the challenge.
In number theory, the n conjecture is a conjecture stated by Browkin & Brzeziński (1994) as a generalization of the abc conjecture to more than three integers.
Formulations
Given n ≥ 3 {\displaystyle n\geq 3}, let a 1 , a 2 , . . . , a n ∈ Z {\displaystyle a_{1},a_{2},...,a_{n}\in \mathbb {Z} } satisfy three conditions:
(i) gcd ( a 1 , a 2 , . . . , a n ) = 1 {\displaystyle \gcd(a_{1},a_{2},...,a_{n})=1}
(ii) a 1 + a 2 + . . . + a n = 0 {\displaystyle a_{1}+a_{2}+...+a_{n}=0}
(iii) no proper subsum of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} equals 0 {\displaystyle 0}
First formulation
The n conjecture states that for every ε > 0 {\displaystyle \varepsilon >0}, there is a constant C {\displaystyle C} depending on n {\displaystyle n} and ε {\displaystyle \varepsilon }, such that:
max ( | a 1 | , | a 2 | , . . . , | a n | ) < C n , ε rad ( | a 1 | ⋅ | a 2 | ⋅ … ⋅ | a n | ) 2 n − 5 + ε {\displaystyle \operatorname {max} (|a_{1}|,|a_{2}|,...,|a_{n}|)<C_{n,\varepsilon }\operatorname {rad} (|a_{1}|\cdot |a_{2}|\cdot \ldots \cdot |a_{n}|)^{2n-5+\varepsilon }}
where rad ( m ) {\displaystyle \operatorname {rad} (m)} denotes the radical of an integer m {\displaystyle m}, defined as the product of the distinct prime factors of m {\displaystyle m}.
Second formulation
Define the quality of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} as
q ( a 1 , a 2 , . . . , a n ) = log ( max ( | a 1 | , | a 2 | , . . . , | a n | ) ) log ( rad ( | a 1 | ⋅ | a 2 | ⋅ . . . ⋅ | a n | ) ) {\displaystyle q(a_{1},a_{2},...,a_{n})={\frac {\log(\operatorname {max} (|a_{1}|,|a_{2}|,...,|a_{n}|))}{\log(\operatorname {rad} (|a_{1}|\cdot |a_{2}|\cdot ...\cdot |a_{n}|))}}}
The n conjecture states that lim sup q ( a 1 , a 2 , . . . , a n ) = 2 n − 5 {\displaystyle \limsup q(a_{1},a_{2},...,a_{n})=2n-5}.
Stronger form
Vojta (1998) proposed a stronger variant of the n conjecture, where setwise coprimeness of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} is replaced by pairwise coprimeness of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}}.
There are two different formulations of this strong n conjecture.
Given n ≥ 3 {\displaystyle n\geq 3}, let a 1 , a 2 , . . . , a n ∈ Z {\displaystyle a_{1},a_{2},...,a_{n}\in \mathbb {Z} } satisfy three conditions:
(i) a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} are pairwise coprime
(ii) a 1 + a 2 + . . . + a n = 0 {\displaystyle a_{1}+a_{2}+...+a_{n}=0}
(iii) no proper subsum of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} equals 0 {\displaystyle 0}
First formulation
The strong n conjecture states that for every ε > 0 {\displaystyle \varepsilon >0}, there is a constant C {\displaystyle C} depending on n {\displaystyle n} and ε {\displaystyle \varepsilon }, such that:
max ( | a 1 | , | a 2 | , . . . , | a n | ) < C n , ε rad ( | a 1 | ⋅ | a 2 | ⋅ … ⋅ | a n | ) 1 + ε {\displaystyle \operatorname {max} (|a_{1}|,|a_{2}|,...,|a_{n}|)<C_{n,\varepsilon }\operatorname {rad} (|a_{1}|\cdot |a_{2}|\cdot \ldots \cdot |a_{n}|)^{1+\varepsilon }}
Second formulation
Define the quality of a 1 , a 2 , . . . , a n {\displaystyle a_{1},a_{2},...,a_{n}} as
q ( a 1 , a 2 , . . . , a n ) = log ( max ( | a 1 | , | a 2 | , . . . , | a n | ) ) log ( rad ( | a 1 | ⋅ | a 2 | ⋅ . . . ⋅ | a n | ) ) {\displaystyle q(a_{1},a_{2},...,a_{n})={\frac {\log(\operatorname {max} (|a_{1}|,|a_{2}|,...,|a_{n}|))}{\log(\operatorname {rad} (|a_{1}|\cdot |a_{2}|\cdot ...\cdot |a_{n}|))}}}
The strong n conjecture states that lim sup q ( a 1 , a 2 , . . . , a n ) = 1 {\displaystyle \limsup q(a_{1},a_{2},...,a_{n})=1}.
Hölzl, Kleine and Stephan (2025)harvtxt error: no target: CITEREFHölzl,_Kleine_and_Stephan2025 (help) have shown that for n ≥ 5 {\displaystyle n\geq 5} the above limit superior is for odd n {\displaystyle n} at least 5 / 3 {\displaystyle 5/3} and for even n {\displaystyle n} is at least 5 / 4 {\displaystyle 5/4}. For the cases n = 3 {\displaystyle n=3} (abc-conjecture) and n = 4 {\displaystyle n=4}, they did not find any nontrivial lower bounds. It is also open whether there is a common constant upper bound above the limit superiors for all n ≥ 3 {\displaystyle n\geq 3}. For the exact status of the case n = 3 {\displaystyle n=3} see the article on the abc conjecture.