AI summary by TwinMind
Quantum Supremacy: What is it and what does it mean?
Sabine Hossenfelder
6m 22s
Quantum Supremacy Definition and Timeline
- Google expected to demonstrate quantum supremacy later this year
- Quantum supremacy: when quantum computer outperforms conventional computer for first time on specific task
- Must be specific task because quantum computers are special purpose machines
Conventional vs Quantum Computers
- Conventional computers work with quantum mechanics at component level (electron balance)
- Operations performed are not quantum
- Store/handle information as bits (1 and 0, up and down)
- Quantum computers store information as qubits (quantum bits)
- Can take any combination of 0 and 1
- Operations can entangle qubits
- Allows solving certain problems much faster
Quantum Computer Applications
- Calculating properties of molecules/materials
- Conductivity, rigidity, color
- Can be calculated from atomic buildup
- Know equations but cannot solve with conventional computers (takes too long)
Computational Scaling Problem
- Can simulate quantum computer on conventional computer by solving quantum mechanics equations
- Computational burden increases exponentially with number of qubits
- 2-4 qubits: personal computer capable
- 50 qubits: requires cluster of supercomputers
- Beyond 50 qubits: cannot be calculated in reasonable time
Testing Quantum Supremacy
- Proposed by Scott Aaronson
- Set up quantum computer to produce probabilistic distributions of measurable variables
- Simulate measurement outcomes on conventional computer (takes very long)
- Demonstrates quantum computer does something classical computer cannot do
- Declaration point somewhat ambiguous
- Could argue better conventional computers/algorithms exist
- For practical purposes doesn't matter much
Current Limitations and Reality Check
- Most molecules have quantum processes exceeding present supercomputer capabilities
- Random variable generation for quantum supremacy testing not useful for actual calculations
- Makes existing quantum computers seem like "new toys for physicists"
Requirements for Useful Quantum Computing
- Estimates vary: 500,000 to 1 billion qubits
- Depends on definition of "useful" and algorithm improvement optimism
- Realistic estimate: few million qubits needed
Scaling Challenges
- No one knows when few million qubit computers will exist
- Dominant approaches unlikely to scale:
- Superconducting qubits
- Ion traps
- No ideas for getting beyond few hundred qubits
- Both engineering and cost problems
NISQ (Noisy Intermediate Scale Quantum) Computers
- Term invented to make investors believe practical applications coming in next decades
- Plausible they will be practically feasible soon
- Problem: no one knows how to calculate useful things with them
Pessimistic Outlook
- Speaker not optimistic about practical applications anytime soon
- Worried quantum computing will follow nuclear fusion path
- Remain "forever promising but never quite work"
- Nevertheless, quantum supremacy demonstration will be significant milestone
