Quantum Mechanics 103
Quantum Implications for Computing
Schrödinger and Uncertainty Going back to Taylor’s experiment, we see that the
wavefunction of the photon extends through both slits Therefore the photon has “traveled” through both
openings simultaneously The wavefunction of a “particle” will contain every
possible path the particle could take until the particle is “detected” by scattering or being absorbed
These paths can interfere with each other to produce diffraction-like probability patterns
BUT, Schrödinger took this explanation to an extreme
Schrödinger’s Famous Cat
Suppose a radioactive substance is put in a box with a cat for a period of time
• If the Geiger Counter triggers, a gun is discharged and the cat is killed
• During that time, there is a 50% chance that one of the nuclei will decay and trigger a Geiger Counter
Schrödinger’s Famous Cat Until an observer opens the box to make a “measurement”
of the system,
• The nucleus remains both decayed
• The Geiger counter remains both triggered
and undecayed
and untriggered• The gun has both fired
and not fired• The cat is both dead
and alive Disclaimer: To be truly indeterministic, this experiment must be performed in a sound-proof room with no window
Paradox? Paradoxical as it may seem, the concept of “superposition
of states” is borne out well in experiment Like superposition of waves producing
interference effects
Quantum Mechanics is one of the most-tested and best-verified theories of all time
But it seems counter-intuitive since we live in a macroscopic world where uncertainty on the order of is not noticeable
Quantum paradox #2
Einstein-Podolsky-Rosen (EPR) paradox Consider two electrons emitted from a
system at rest; measurements must yield opposite spins if spin of the system does not change
We say that the electrons exist in an “entangled state”
More EPR
If measurement is not done, can have interference effect since each electron is superposition of both spin possibilities
But, measuring spin of one electron destroys interference effects for both it and the other electron;
It also determines the spin of the other electron How does second electron “know” what its spin is
and even that the spin has been determined
Interpreting EPR
Measuring one electron affects the other electron!
For the other electron to “know” about the measurement, a signal must be sent faster than the speed of light!
Such an effect has been experimentally verified, but it is still a topic of much debate
Interference effects Remember this Mach-Zender Interferometer? Can adjust paths so that light is split evenly between top U
detector and lower D detector, all reaches U, or all reaches D – due to interference effects
Placing a detector (either bomb or non-destructive) on one of the paths means 50% goes to each detector ALL THE TIME
Interpretation
Wave theory does not explain why bomb detonates half the time
Particle probability theory does not explain why changing position of mirrors affects detection
Neither explains why presence of bomb destroys interference
Quantum theory explains both! Amplitudes, not probabilities add - interference Measurement yields probability, not amplitude - bomb
detonates half the time Once path determined, wavefunction reflects only that
possibility - presence of bomb destroys interference
Quantum Theory meets Bomb
Four possible paths: RR and TT hit upper detector, TR and RT hit lower detector (R=reflected, T=transmitted)
Classically, 4 equally-likely paths, so prob of each is 1/4, so prob at each detector is 1/4 + 1/4 = ½, independent of path length difference
Quantum mechanically, square of amplitudes must each be 1/4 (prob for particular path), but amplitudes can be imaginary or complex! This allows interference effects
What wave function would give 50% at each detector?
Must have |a|2 = |b|2 = |c|2 = |d|2 = 1/4 Need |a + b|2 = |c+d|2 = 1/2
TTRRRTTR dcba
TT22
1RR
22
1RT
22
1TR
22
1 iiii
2
1
8
4
22
22
2 ba
2
1
8
4
22
22
2 dc
If Path Lengths Differ, Might Have
Lower detector:
Upper detector:
TT22
1RR
22
1RT
2
1TR
2
1 ii
02
1
2
12
2
122
22
22
1
22
122
2
iii
Voila, Interference!
When Measure Which Path,
Lower detector:
Upper detector:
TT2
1TR
2
1
2
1
2
12
2
Voila, No Interference!
RT2
1RR
2
1
2
1
2
12
2
Quantum Storage Consider a quantum dot capacitor, with sides 1 nm
in length and 0.010 microns between “plates” How much energy required to place a single
electron on those plates? Can make confinement of dot dependent upon
voltage Lower the voltage, let an electron on –> 1 Lower voltage on other side, let the electron off -> 0
What must a computer do?Deterministic Turing Machine still good model Two pieces:
Read/write head in some internal state “Infinite” tape with series of 1s, 0s, or blanks
Follows algorithms by performing 3 steps: Read value of tape at head’s location Write some value based on internal state
and value read Move to next value on tape
Can we improve this model? Probabilistic Turing Machine sometimes better Multiple choices for internal state change Not 100% accurate, but accuracy increases with
number of steps Can solve some types of problems to sufficient
accuracy much more quickly than deterministic TM can
Similar concept to Monte Carlo integration
Limits on Turing Machines Some problems are solvable in theory but
take too long in practice e.g., factoring large numbers
Can label problems by how the number of steps to compute grows as the size of the numbers used grows addition grows linearly multiplication grows as the square of digits Fourier transform grows faster than square factoring grows almost exponentially
Examples of factoring time MIP-year = 1 year of 1 million processes per
second Factoring 20-digit decimal number done in 1964,
requiring only 0.000009 MIP-years 45-digit decimal number (1974) needs 0.001 MIP-
years 71-digit decimal number (1984) needs 0.1 MIP-
years 129-digit decimal number (1994) needs 5000
MIP-years
Quantum Cryptography
Current best encryption uses public key for encoding
Need private key (factors of large integer in public key) to decode
Really safe unless Someone can access your private key Quantum computers become prevalent
Quantum Cryptography II
Quantum Computers can factor large numbers near-instantly, making public key encryption passe
But, can send quantum information and know whether it has been intercepted
What problems face QC? Decoherence: if measurement made, superposition
collapses Even if measurement not intentional! i.e., if box moves, cat becomes alive or dead, not both
Quantum error correction No trail of path taken (or else no superposition) Proven to be possible; that doesn’t mean it’s easy!
HUGE Technical challenges electronic states in ion traps (slow, leakage) photons in cavity (spontaneous emission) nuclear spins in molecule (small signal in large noise)