The transmon, grown into a small universe of the hardware around it — the junction and the capacitor it's made of, the ladder it is, the resonator and pulses that read and drive it, the coherence clock it races, and its sibling qubits (the quantum dot, the exciton, the ion trap). Fully cited; two live instruments; and an honest read of which of David's 'hardware inventions' actually belong here.
the repo's standing architectural question, kept
what a transmon is made of — Cooper pairs, a Josephson junction, a shunt capacitor
Two electrons bound into a boson by the lattice (BCS): below T_c they condense into a single coherent state with zero resistance. The transmon's currency — what tunnels across the junction is a Cooper pair.[7]
A thin insulator between two superconductors: Cooper pairs tunnel through it as a supercurrent. It behaves as a NONLINEAR inductor — the one circuit element that is nonlinear AND non-dissipative — and that nonlinearity is the whole game.[1]
Shunt the junction with a big capacitor so the Josephson energy dominates the charging energy: E_J/E_C ≈ 50–100. This 'transmon regime' exponentially FLATTENS the charge dispersion — the qubit stops caring about stray charge noise — at the small cost of reduced anharmonicity.[1]
the anharmonic oscillator carved into a qubit — unequal rungs, ω01, and the ionization at the top
The junction warps the harmonic well so the rungs are UNEQUAL — the anharmonicity α ≈ −E_C ≈ −200 MHz. Because the |1⟩→|2⟩ gap differs from |0⟩→|1⟩, a drive tuned to ω01 addresses only the qubit transition. You carve a two-level system out of an infinite ladder by spacing.[1]
The |0⟩↔|1⟩ transition sits at ω01/2π ≈ 4–6 GHz — microwave, the band of everyday cavities and the reason these chips live in a dilution fridge near 10 mK (so kT ≪ ℏω and the qubit starts in |0⟩).[2]
Drive it hard enough and the state climbs the rungs and ESCAPES the cosine well entirely into the continuum — 'ionization.' Not a metaphor: a real, studied limit on how fast you can drive, and the demo's 256→257 escape.[6]
how you drive it and look at it without breaking it — the resonator, DRAG pulses, and the coherence clock
Couple the qubit to an LC / transmission-line RESONATOR. In the dispersive regime the qubit's state PULLS the resonator's frequency by ±χ; bounce a probe tone off the resonator and its phase tells you |0⟩ vs |1⟩ — a quantum non-demolition read. (This is where an LC resonator — e.g. the 'toroid' calculator — actually belongs in a transmon.)[4][3]
Qubits are driven by shaped microwave pulses (in-phase I and quadrature Q). Because the ladder is only weakly anharmonic, a naive pulse leaks population to |2⟩; DRAG adds a quadrature derivative term that cancels the leak. Gates are µs-to-ns shaped tones.[5]
T1 (energy relaxation, |1⟩→|0⟩) and T2 (dephasing) set the clock: dielectric loss, quasiparticles, and two-level-system defects eat coherence. Modern transmons reach tens to hundreds of µs — every gate races the decay.[2]
the other quantum-hardware modalities — the quantum dot, the exciton, the ion trap
Shrink a semiconductor to 2–10 nm and the electron's wavefunction is squeezed until its energy levels go DISCRETE — an 'artificial atom' whose level spacing (and emission colour) is set by SIZE, not species. Bawendi's hot-injection synthesis (1993) made them monodisperse and manufacturable.[10]
An electron's SPIN (up/down) trapped in a silicon quantum dot, used as the qubit — Loss–DiVincenzo. Discrete atom-like levels, charging energy, gate control. A different modality from the transmon that shares the 'artificial atom' idea.[8]
The platform's wager: build qubits in a CMOS FOUNDRY and ride the entire semiconductor industry's fabrication. In 2022 three groups crossed >99% two-qubit gate fidelity in silicon spin qubits (the surface-code threshold), and imec/Diraq now make them on industrial lines — the transmon's main rival for scaling.[9][8]
Kane's 1998 idea, the original 'silicon quantum computer': implant a single ³¹P phosphorus atom (a DONOR) into isotopically-pure ²⁸Si and use its nuclear (or electron) spin as the qubit, controlled by surface gates through the hyperfine interaction. Arsenic is the same donor family. The silicon band sets where the donor level sits — David's silicon-band instrument shows the doping shift live.[11]
A photon absorbed in a semiconductor lifts an electron and leaves a hole; the bound electron–hole pair is an EXCITON — a light-matter quasiparticle. Excitons in quantum dots are the basis of single-photon sources and optical qubit readout, the coupling where light and matter exchange a quantum. (The live 'exciton' instrument renders this.)
Hold a single charged atom in oscillating electric fields and cool it with lasers; its internal states are the qubit. The longest-coherence, highest-fidelity qubits — and the natural CONTRAST to the transmon: atoms are identical and slow, transmons are fabricated and fast. (Named because David's folder had a 'teaching trap' — which turned out to be a security sandbox, not this.)
David asked me to find everything in all of C:\Davids files that fits the transmon universe, exhaustively. I swept the entire repo by content (grepping for josephson · transmon · superconduct · anharmonic · cooper-pair) and read every physics/quantum/hardware candidate. Honest verdict (the fluff-call): the genuine fits are few — the one substantive new one is the silicon spin-qubit series, now built in. The rest of the giant repo is cosmology, energy/reactors, materials, math, classical & AI computing — named plainly. IN = genuine superconducting-qubit content · ADJACENT = quantum-hardware-adjacent / already-live · OUT = not transmon physics
the founding transmon paper, the engineer's guide, circuit QED & dispersive readout, DRAG, transmon ionization, the first charge qubit, and the quantum-dot spin-qubit proposal
a gap where two regimes meet — the only question is whether anything crosses it