TECH Signal 468
A partial digestion of the HRT counterexample
↗
A new counterexample shows that Schwartz functions can satisfy nontrivial linear relations among their time-frequency shifts, disproving the Heil-Ramanathan-Topiwala conjecture in that class.
For engineers who rely on the independence of time-frequency shifts, such as in Gabor frame design or sparse time-frequency representations, this result means that the Schwartz assumption no longer guarantees independence, so additional checks may be needed. The construction uses AI-assisted search and numerical verification, illustrating a hybrid proof approach that could be adopted in other hardness-of-analysis problems. The finding does not extend to functions with analytic or super-exponential decay, where the conjecture remains true, limiting the scope of the impact.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
The counterexample provides explicit complex coefficients, distinct time-frequency points, and a nonzero Schwartz function that sum to zero under shifts.
The proof strategy was initially generated with AI assistance, then refined by hand and verified with a traditional numerical computation.
The result sits just beyond prior positive theorems: it holds for Schwartz functions but fails for those with super-exponential or analytic decay.
THE CLUSTER