Abstract:Strömbergsson and Venkatesh proved that a system of homogeneous linear congruence modulo prime has a positive probability to have a short non-trivial solution. We extend this result and show that the same holds for square-free moduli. In the case of 2-variables single linear congruence, we show that there is a positive probability to have a short solution for all integer moduli as well as positive probability for having short non-trivial solutions which are primitive in a suitable sense.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the existence and probabilistic properties of short solutions of systems of homogeneous linear congruence equations. Specifically, the paper extends the results of Strömbergsson and Venkatesh under prime moduli and proves that for square - free moduli, the probability of the existence of short solutions for systems of homogeneous linear congruence equations is also positive. In addition, for a single linear congruence equation in two variables, the paper further proves that for all integer moduli, the probability of the existence of short solutions is positive, and the probability of the existence of primitive solutions is also positive.
### Main contributions of the paper
1. **Extension of existing results**:
- The original results were limited to prime moduli, while this paper extends them to square - free moduli.
- For a single linear congruence equation in two variables, it is further extended to all integer moduli.
2. **Definition of "short solution"**:
- For a given modulus \( N \), if a solution \( \mathbf{x} \) satisfies \( \|\mathbf{x}\| \ll N^{j/n} \) (where \( j \) is the rank of the system of equations and \( n \) is the number of variables), it is called a "short solution".
- A non - trivial solution refers to a solution where \( \mathbf{x} \neq 0 \mod N \).
3. **Probabilistic properties**:
- It is proved that for a square - free modulus \( N \), the probability that a randomly selected system of homogeneous linear congruence equations has a short solution is positive.
- For a single linear congruence equation in two variables, specific probability values are further given.
### Main results
- **Theorem 2**: For a square - free modulus \( N \), the probability that a randomly selected system of homogeneous linear congruence equations \( A\mathbf{x} \equiv 0 \mod N \) has a short solution is positive.
- **Theorem 4**: For a single linear congruence equation in two variables \( r_1 x + r_2 y \equiv 0 \mod N \), the probability of the existence of a non - trivial short solution is \( \frac{3a}{\pi^2} \), and the probability of the existence of a primitive solution is at least \( \frac{3a}{\pi^2} \left(1 - \frac{1}{\sqrt{2}}\right) \).
### Methods and techniques
- **Lattice theory**: Use lattice theory to correspond systems of homogeneous linear congruence equations to sub - lattices, and derive solutions of the equations by studying the properties of sub - lattices.
- **Hecke operator**: Apply the properties of Hecke operators, especially their actions on \( L^2 \) spaces, to estimate the probability of solutions.
- **Dirichlet box principle**: Use the Dirichlet box principle to find short solutions.
### Applications
- **Applications in number theory**: These results have wide applications in number theory. For example, in the upper bound estimation of exponential sums, these results can be used to improve existing estimation methods.
In summary, through the study of systems of homogeneous linear congruence equations, this paper extends existing theoretical results and provides new probabilistic properties, providing new tools and methods for research in related fields.