Tight Fourier Tails for AC 0 Circuits Avishay Tal (IAS)
Bounded Depth Circuits •
Brief History •
Brief History •
Brief History •
Brief Introduction to Discrete Fourier Analysis •
Brief Introduction to Discrete Fourier Analysis •
Tails and Low-Degree Approximation Equivalence •
Comparison of Results in Fourier language • LMN’ 89 Boppana’ 97 Håstad’ 01 Håstad’ 12 IMP’ 12 This Work
Comparison of Results in Polynomial Language • LMN’ 89 Boppana’ 97 Håstad’ 01 This Work
Our Result •
Applications to Pseudorandomness F PRG • F
Summary of Applications
Parity not in AC 0 FF •
Parity not in AC 0 •
Tails for Random Restrictions •
LMN Result
Recall •
Decision Tree
Restriction Tree
Switch-Many Lemma •
Tails for Arbitrary Restrictions •
Tails for Arbitrary Restrictions
Our Result Prob lem
Different Notions of Fourier Concentration •
Exponentially Small Fourier Tails •
Open Questions •
Open Questions • F PRG F
Open Questions 3. (Further) derandomize switching lemma
Open Questions 4. Which distributions fool AC 0? F F In particular, can you find a distribution that log-time quantum algorithms distinguish from the uniform dist. while AC 0 does not.
Open Questions 5. Does exponentially small Fourier tails imply PRGs? Other implications?
Open Questions 1. 2. 3. 4. 5. Better lower bounds for AC 0 Improve PRGs for AC 0 (Further) derandomize the switching lemma Which distributions fool AC 0? Is ESFT(t) enough to derive PRGs? Thank You!