LWE with polynomial ratio ⇒ SIS

Fully-black-box reduction · standard model · folklore · security loss: Constant factor: for prime , with one call to the SIS solver.

Statement

Let be prime, the discrete Gaussian over of width , and with . If decision LWE is hard for , then SIS is hard for , with a constant-factor loss — folklore. Under LWE with polynomial modulus-to-noise ratio this gives SIS in the regime , : take , a prime (so ), and , which is at most for all large ; more generally with .

Sketch

Given with , run the SIS solver on , which is uniform in , and output “LWE” iff it returns with , and . For , is subgaussian with parameter , so it leaves with probability at most ; for uniform it is uniform modulo , since and is prime, and lands in the interval with probability at most .

Notes

  • Composed with SIS ⇒ CRHF at the same , LWE with polynomial modulus-to-noise ratio gives collision-resistant hashing keyed by a uniform matrix — standard.