Downloaded 92 times
Uploaded byinside-BigData.com
Quantum Computing and Quantum Supremacy at Google
The document discusses the progress and challenges of quantum computing, highlighting the importance of both the quantity and quality of qubits for achieving quantum supremacy. It outlines Google's strategy for quantum computing, which includes error correction and near-term applications in quantum chemistry. Key metrics such as error rates, speed, and qubit connectivity are emphasized as crucial for the practical realization of quantum technologies.
More Related Content
Similar to Quantum Computing and Quantum Supremacy at Google
Quantum Computing and Quantum Supremacy at Google
- 1. Quantum Computing andQuantum Supremacy John Martinis, Google
- 2. The Quantum SpaceRace Hardware Challenges: Quantity Quality
- 3.
- 4. Quantum Data ( |0〉+|1〉)2=|00〉+|01〉+|10〉+|11〉
- 5. ( |0〉+|1〉 )n n=50:supercomputer n=300: more states than atoms in universe Really Big Data
- 6. Classical bit: Coin ontable Digital: self-correcting Quantum bit: Coin in space Analog: sensitive to small errors Bits and Qubits 1e0 iφ θθψ sincos +=0 or 1
- 7. Classical circuit: Quantumcircuit: Any Logic Built from Universal Gates 1 bit NOT 2 bit AND Wiring fan-out 1 qubit rotation 2 qubit CNOT -No copy- measurek |ψ〉=|0〉 Time (sequence)Space (layout)
- 8. Classical system: Quantumsystem: Full Digital (Clocked) Logic D flip-flop aligns timing to clock Error correction aligns amplitude and phase each round, makes “quantum flip-flop” clock logic error correction (parity) logic Classical error decoding and logic 10-9 errors: 1000 qubits + 0.1% errors
- 9. Need Both Qualityand Quantity 100 101 102 103 104 105 106 108 Number of Qubits 10-1 10-2 10-3 10-4 Limitingerrorrate Classically simulatable ✘ Error correction threshold 0.1110100 “errorcorrectiongain” 107 10-9 errors
- 10. Need Both Qualityand Quantity 100 101 102 103 104 105 106 108 Number of Qubits 10-1 10-2 10-3 10-4 Limitingerrorrate Near-term applications Classically simulatable ✘ ? Error correction threshold 0.1110100 “errorcorrectiongain” 107 ✔ Useful error corrected QC quantity hype 1 2 3 Google strategy 1: Quantum Supremacy 2: Look for near-term apps 3: Error correction 4: Full QC 4
- 11. Quality in QuantumChemistry Experiments H2 Molecule LiH Molecule 2017 Brand 5% Brand 5% 1) Need low errors for accurate predictions 2) Need figure of merit for useful 2015 Google 0. 5% chemical accuracy 20x worse No better than initial trial? Quality:
- 12. Xmon Transmon Qubit quantumcircuit: 6 GHz microwave oscillator 100 𝜇m Flux Bias Self capacitance Readout resonator Bus AC drive Frequency (Z) µwaves (X,Y) Variable L (quantum: stabilize phase to 1 ps)
- 18. Quantum vs. Classical-SupercomputerChallenge 50 qubits - checked with classical supercomputer
- 19. Quantum Supremacy Algorithm:Qubit Speckle Clifford Non-Clifford X, Z, H, X1/2… Z1/4 CZ (Random guess: any outcome k has probability pcl = 1/2n) 2) Run quantum computer, measure k (2n possible outcomes) repeat sampling 100,000 times 4) Correlation: cross entropy S = 〈 ln p(k)/pcl 〉 5) Compare to theory Squ ≅ 0.42 quantum S ≅ -0.58 classical nqubits d (time) measurek |ψ〉=|0〉 6) Try another instance 1 s days 200 drives 1) Choose 1 instance, randomly from gateset 3) Calculate |ψ〉, p(k)= |〈k|ψ〉|2 store in lookup table
- 20. speckle = coherence predict= fidelity
- 21.
- 22. Measuring fidelity Computing 500 states inparallel 0.3% error per gate & cycle
- 23. Quantum Materials (theory) 9lattice sites 10000 T
- 24. Quantum Materials (theory+ experiment)
- 25. “Bristlecone” Architecture 2D gridof nearest neighbor coupled qubits, 72 total Tile
- 27.
- 28. Huge Progress inAlgorithms (Quantum Chemistry) N N2.7 N6 N8 N11 N4 sec day year universe Exact: 100 logical qubits (error corrected) Approximate: 100 physical qubits (?) 1985 Feynman (proposal)
- 29. Huge Progress inAlgorithms (Quantum Chemistry) N N2.7 N6 N8 N11 N4 sec day year universe 1985 Feynman (proposal) Useful QC now possible?
- 30. Summary: Key Metrics 1)Quantity 2) Quality limiting error best and average 3) Speed 106 difference in technologies! 4) Connectivity Error correction at least 2D 9 B: 0.55% M: 0.75% 40 ns 1D UCSB: Google: 72 2D
- 31. Two Qubit GateErrors (most critical) 2-qubit gate error e2 integratedhistogram IBM20 Google5 Google better by ~ 2-10x 10x 2x Google9gmon system perf. phase error better