A computational investigation into whether the lattice theta function can be evaluated in polynomial time for general complex symmetric matrices of large dimension.
Summary of experimental results across dimensions n = 1 to 15
Interactive charts from computational experiments
Numerical results from all experiments
| n | Terms | Time (s) | Time LLL (s) |
|---|---|---|---|
| 1 | 7 | 0.0001 | 0.0001 |
| 2 | 25 | 0.0001 | 0.0002 |
| 3 | 125 | 0.0003 | 0.0004 |
| 4 | 625 | 0.0024 | 0.0034 |
| 5 | 3,125 | 0.3869 | 0.1014 |
| 6 | 15,625 | 0.0340 | 0.9601 |
| 7 | 78,125 | 3.8339 | 3.2964 |
| t | |Theta| | Est. min norm2 |
|---|---|---|
| 0.5 | 1.085 | -0.026 |
| 1.0 | 3.23e-1 | 0.180 |
| 2.0 | 5.91e-2 | 0.225 |
| 5.0 | 7.28e-4 | 0.230 |
| 10.0 | 5.30e-7 | 0.230 |
| 20.0 | 2.80e-13 | 0.230 |
| 50.0 | 4.16e-32 | 0.230 |
| 100.0 | 1.73e-63 | 0.230 |
True CVP distance: 0.230 (brute-force computed)
Evidence regarding the open problem of polynomial-time computability