Section Q&A

Rapid-fire questions an interviewer asks around Logic & Aptitude Puzzles. Answer out loud, then check yourself.

Q1. Why is the 9-ball puzzle solved in two weighings, not three?

Answer: A balance yields three outcomes per weighing. Two weighings provide 3^2=9 distinguishable outcome paths, matching nine candidates exactly.

Trap: Treating each weighing as binary and missing ternary information capacity.

Q2. In biased-coin fair-bit generation, why do we reject HH and TT?

Answer: Only HT and TH are equiprobable for any unknown bias p. Keeping HH/TT would skew outputs unless p were known.

Trap: Assuming unknown-bias outcomes are automatically symmetric.

Q3. Does adding one synchronizer flop improve CDC reliability linearly?

Answer: No. MTBF scales exponentially with added settling time (approximately exp(deltaT/tau)), which is why one stage can give huge gains.

Trap: Assuming CDC improvement is a fixed additive margin per flop.

Q4. A mismatch appears only on some seeds. What should be checked before editing DUT RTL?

Answer: First rule out TB observation races (sampling region, clocking block skew, handle reuse). Intermittent seed-only failures are often checker timing bugs.

Trap: Jumping directly into DUT logic changes without proving monitor/checker correctness.