"Assumptions kill an investigation." - Reacher
EE 400, University of Washington Seattle. Ten weeks, four quantum technology experiments and one extension.
PHYS 225 gave me quantum mechanics as theory. I took this lab course to see how Rabi frequency, quantum key distribution, and entanglement behave on a bench. What I did not anticipate was how much of the real work goes into identifying and quantifying uncertainty. That looks obvious in retrospect, but I only learned it by measuring.
Each experiment below links to a retrospective write-up on my site, where I work back through the analysis with more care than the lab period allowed.
| # | Experiment | Description | Note |
|---|---|---|---|
| 1 | Quantum State Control | Find the qubit resonant frequency, drive Rabi oscillations, and analyze performance using the spin of a defect in diamond. Demonstrates basic single-qubit functionality. | Nitrogen Vacancy (NV) center qubit. |
| 2 | Quantum Entanglement | Violate Bell's inequalities using an entangled photon source based on parametric down-conversion. Proves that non-local correlations exist in quantum systems. | |
| 3 | Quantum Key Distribution (BB84) | Implement BB84 and analyze its performance using a single-photon source based on parametric down-conversion. Demonstrates fundamentally secure cryptographic key distribution. | |
| 4 | Tomography and Teleportation | Explore quantum state tomography and teleportation on cloud quantum hardware. | |
| 5 | Quantum Magnetometry | Extend the Quantum State Control setup beyond the standard procedure, using the diamond defect spin as a magnetic field sensor. | Extension of Experiment 1 |
The objectives:
- Determine the qubit energy levels
- Drive coherent rotations between two qubit levels and measure Rabi frequency and qubit decay time
- Experimentally determine the relationship between Rabi frequency and applied MW power
I learned to drive a full ODMR measurement chain: a SpinCore PulseBlaster sequencing TTL pulses to the laser, the MW switch, and the lock-in reference; a Windfreak generator setting microwave frequency and power; and an Ametek lock-in amplifier pulling the photoluminescence signal out of noise. Sweeping 2750 to 3000 MHz resolved the eight transitions predicted from the four NV orientations. Pulsed-ODMR narrowed the resonance to 2788 MHz, and Rabi oscillations gave 2.22 MHz at -15 dBm and 2.86 MHz at -12 dBm.
Retrospective: full analysis on my site
The objective:
- Empirically quantify and demonstrate CHSH Bell Test
I learned to operate a quED SPDC source: a half-wave plate setting the pump to 45°, two Type-I crystals producing polarization-entangled pairs, motorized polarizers, and fiber-coupled APDs feeding a coincidence counter. Polarization carried the correlation. Fitting the coincidence curves in Python separated the two signatures: the unentangled state factorizes as cos²α·cos²β, while the entangled state depends only on the angle difference, ½cos²(β−α). CHSH gave S = 2.241 ± 0.040 entangled and S = 1.404 ± 0.043 with the half-wave plate removed, straddling the classical bound of 2. Uncertainties came from Poisson counting statistics propagated through the S calculation.
Retrospective: full analysis on my site
The objective:
- Demonstrate BB88 Quantum key distribution protocol
I learned to run BB84 end to end on a quED system: a fiber beamsplitter and three APDs generating quantum random bits, motorized half-wave plates and polarizers encoding them, and quApp's BB84 mode transmitting and measuring. Python handled the rest, sifting matched-basis detections into a 1024-bit key and XORing it against a 32×32 image. Comparing Alice's and Bob's keys gave a bit error rate of 4.30%, 44 mismatched bits, which appear directly as 44 flipped pixels in the decrypted image. The key rate came to 23.51% against an ideal 25%. The RNG showed a 41.7% bias toward zeros.
Reference: Quantum Computation and Quantum Information by Isaac Chuang and Michael Nielsen page 582
Retrospective: full analysis on my site
I learned to work inside IBM's Qiskit stack: building the three-qubit teleportation circuit, transpiling it through a preset pass manager, and running it on three backends, the ideal AerSimulator, the FakeSherbrooke noise model, and the real ibm_sherbrooke QPU. Swap gates moved Bob's qubit up to ten positions away to stand in for distance. State tomography reconstructed both density matrices from ⟨X⟩, ⟨Y⟩ and ⟨Z⟩ over 10,000 shots per basis, and state fidelity scored the transfer. Fidelity held at 1.000 in simulation, swung between 0.95 and 0.25 under the noise model, and between 0.76 and 0.30 on hardware.
Retrospective: full analysis on my site
All the projects were completed in Spring 2025 Quarter.