QUASAR Framework
The unifying theoretical scaffold for QLT's photonic dimension ladder — from STAR-PHASER linear-circuit chips (Q < 0.2) through QUASAR hybrid and field-dominant machines (Q ≥ 0.2). Coupled state Ψ = (A, Φ, Ξ), the Q-metric, boundary-layer fab, five scaling laws, and a mandatory never-state register — consolidated from internal memos S81–S85.
STAR-PHASER → QUASAR
Same physics, different operating coordinate — the Q-metric separates linear circuits from field computers.
Tap any tile to expand. GEMINI ships today; all other chips are roadmap, research, or speculative-framework unless labeled.
STAR-PHASER
Region I · Q < 0.2 · linear QFP dominatesGEMINI — d=2, path qubits
→v1 shipping product. Dual-rail path encoding on Si₃N₄; Q ≪ 0.05. OPC hygiene only. LDV dominates the action (S81).
d=2 · Region I · Q ≪ 0.05 · TODAY
SOLAR — d=10 composite qudit
→Structured composite — not GF(10). Q ≈ 0.05–0.10. Region I.
Never-state NS-03: never call SOLAR "GF(10)."
GALAXY — QUASAR-transition prototype
→GALAXY is not QUASAR. Rehearses Region II hardware in Region I operation (Q ≈ 0.08–0.15). See §6.
d=16 · GF(16) · Q < 0.2 · roadmap
TETRIS — last pure STAR-PHASER
→d=32, GF(32), Q ≈ 0.12–0.18. Final gate-locked product before QUASAR NOVA at d=64. Law IV AWG stress begins (S84).
QUASAR
Region II+ · Q ≥ 0.2 · hybrid & field-dominantNOVA — first QUASAR chip
→d=64, GF(64), Q ≈ 0.25–0.45. Distributed OPC mesh, 8×8 octet tiling. ~4× OPC margin at Δf=50 GHz.
SUPER — paradigm break
→d=128, Q ≈ 0.35–0.65. 16k phase pairs, ~2.7× OPC margin. OPC + squeeze mandatory.
THETA — field balance
→d=256, Q ≈ 0.85–1.05. Continuous OPC mesh, Δf=25 GHz. Speculative-framework
LOTUS — Region IV horizon
→d=512, Q ≈ 1.2–1.6. Split-band OPC, CV-primary. Research horizon only (NS-08).
Canonical brand rule
STAR-PHASER ⟺ Region I only (Q < 0.2) → GEMINI · SOLAR · GALAXY · TETRIS QUASAR ⟺ Region II+ (Q ≥ 0.2) → NOVA · SUPER · THETA · LOTUS
Lagrangian: Ψ = (A, Φ, Ξ)
Coupled three-sector state with decomposed action — design scaffold, not measured Hamiltonian.
Speculative-framework Entire §2 carries this label. Euler–Lagrange equations recover STAR-PHASER limits at Q ≪ 1 and field-computing limits at Q ≳ 1 — pending THETA/LOTUS telemetry.
GF(2^m) control
Symbol labels on frequency bins. QFP sandwiches, FAU kernels (GFADD/GFMUL). Slow manifold updated at gate clock. Maps to fab L3–L4 (TFLN EOM + QFP mesh).
Spectral manifold
Continuous optical state ψ(f,t). Nonlinear dynamics (FWM, OPC) act primarily on Φ. Maps to L2 SiN bus + comb.
Syndrome manifold
Witness observables — pairwise phases, homodyne quadratures, ChiL scalars, RS syndromes. Maps to L7 readout + digital ECC.
Action decomposition
Five Lagrangian sectorsLDV — discrete-variable kinetic
→L_DV = Σ_k (iℏ/2)(â_k† ∂_t â_k − ∂_t â_k† â_k) − V_QFP(A, Φ)
Standard photonic mode kinetic energy + QFP gate potential. Region I: V_QFP ≫ nonlinear terms.
LCV — continuous-variable squeezing
→L_CV = ∫ df (|∂_t q(f)|² + |∂_t p(f)|²)/2 − V_sqz(q, p; r(f))
Mandatory from SUPER upward. Reduces effective σ_gate in whitepaper §2.3.
LOPC — phase conjugation
→L_OPC = − κ_OPC Σ_k (â_k† â_k* + h.c.) · η_OPC(k)
χ⁽³⁾ FWM with undepleted pump [demonstrated theory]. THETA+: distributed along bus z.
LINT — A–Φ–Ξ interaction
→EOM drives algebra onto field. ChiL witness loop reads Φ, writes Ξ. Digital QEC maps Ξ → A — OPC ≠ QEC compliance.
LFWM — four-wave mixing
→Supplies H_NL bucket. GALAXY: first chip with non-zero L_FWM in production stack. Region II+: duty rises until nonlinear ≈ linear.
Gate instruction → δA → drives δΦ via EOM/QFP Noise / drift → δΦ → observed by δΞ OPC + squeeze → feedback on Φ, witness update on Ξ Digital QEC → discrete jump on A, Ξ (syndrome decode)
| Limit | Condition | Chips |
|---|---|---|
| Linear circuit | Q ≪ 0.2 | GEMINI–TETRIS |
| Hybrid | 0.2 ≤ Q < 1 | NOVA, SUPER |
| Field balance | Q ≈ 1 | THETA |
| Field-dominant | Q > 1 | LOTUS |
Regions I–IV
When does a photonic processor stop behaving like a linear interferometer with nonlinear patches?
Q ≡ (H_NL + H_OPC + H_CV) / H_linear
Q < 0.2
STAR-PHASER. Linear QFP dominates. GEMINI → TETRIS.
0.2 ≤ Q < 1
QUASAR hybrid. Distributed OPC + squeeze. NOVA, SUPER.
Q ≈ 1 ± 0.1
Field balance. THETA. Speculative-framework
Q > 1
Field-dominant. LOTUS. Research horizon only.
| Chip | d | Q target | H_linear | H_OPC | H_CV | H_NL | Region |
|---|---|---|---|---|---|---|---|
| GEMINI | 2 | ≪ 0.05 | ~98% | ~1% | ~1% | ~0% | I |
| SOLAR | 10 | ~0.05–0.10 | ~90% | ~5% | ~5% | ~0% | I |
| GALAXY ★ | 16 | ~0.08–0.15 | ~85% | ~6% | ~4% | ~5% | I |
| TETRIS | 32 | ~0.12–0.18 | ~82% | ~8% | ~5% | ~5% | I |
| NOVA | 64 | ~0.25–0.45 | ~60% | ~15% | ~12% | ~13% | II |
| SUPER | 128 | ~0.35–0.65 | ~50% | ~18% | ~17% | ~15% | II |
| THETA | 256 | ~0.85–1.05 | ~45% | ~28% | ~22% | ~20% | III |
| LOTUS | 512 | ~1.2–1.6 | ~35% | ~30% | ~25% | ~25% | IV |
η_FWM ≈ (γ_eff · P_pump · L_eff)² (low-gain, Δk ≈ 0) H_OPC ∝ N_OPC · η_FWM · P_pump H_NL ∝ N_FWM_pathways · (γ P L)² η_eff = 1 − (1 − η_FWM)^K (K recirculating passes)
Boundary-layer fab & 7-layer cross-section
Nonlinearity as distributed boundary layer co-equal with the SiN backbone.
Tap any layer to expand. L5 boundary NL is the QUASAR differentiator — rehearsed on GALAXY, continuous on LOTUS.
Seven-layer vertical stack
Bottom → top · canonical fab diagramSubstrate — Si handle (725 µm)
→Thermal sink. All QUASAR chips share this foundation.
Backbone — stoichiometric Si₃N₄ (400 nm)
→Damascene bus + rings. 0.05–0.1 dB/cm class. Linear routing stays pure SiN.
EO mix — TFLN heterobond TW-EOM (300 nm)
→Gate drives on A sector. Maps to LDV.
Shaper — SiN microring SPU bank
→16 rings (GALAXY) scaling to 512 rings (LOTUS). Per-bin spectral shaping.
Boundary NL — AlGaAs / As₂S₃ χ⁽³⁾ overlay
→200–500 nm overlay where mode overlap exceeds threshold. OPC + FWM execute here. GALAXY: 2–3 cells (~15–20 mm period). LOTUS: continuous mesh (~2–3 mm). BEOL deposition <250 °C.
How we do it: Adiabatic taper cell Lcell = 2–4 mm; comb-derived pump rotation per G45.
CV inject — squeeze coupler + OPO shunt
→Evanescently couples to SiN bus at octet boundaries. L0 boundary (GALAXY optional 3–6 dB) → L3 field (THETA/LOTUS ≥8–10 dB). Co-fabricated windows with L5.
Readout — WDM + SPAD / PD array
→Top BEOL metal. Homodyne witness per octet for ChiL. Maps to Ξ sector.
| Zone | Location | Duty | Function |
|---|---|---|---|
| Z0 passive | I/O routing | 0% NL | Low-loss transport |
| Z1 refresh | Between QFP stages | Low–med | OPC conjugation |
| Z2 compute | Dense QFP blocks | 0% χ⁽³⁾ | SU(d) gates |
| Z3 CV | Octet boundaries | Medium | 3–10 dB squeeze |
| Z4 pump bus | Parallel waveguide | High when active | Pump routing |
| Chip | d | Δf | Bcomb | NFWM |
|---|---|---|---|---|
| GALAXY | 16 | 50 GHz | 750 GHz | ~3.8k |
| NOVA | 64 | 50 GHz | 3.15 THz | ~262k |
| THETA | 256 | 25 GHz | 6.38 THz | ~16.8M |
| LOTUS | 512 | 20 GHz | 10.2 THz | ~134M |
Grid: fk = fref + k·Δf. LOTUS requires split-band factorization (Band A ∥ Band B).
GALAXY — QUASAR-transition prototype
Trials Region II hardware in Region I operation. GALAXY is not QUASAR.
- SiN microcomb 16-line · Δf=50 GHz · Bcomb=750 GHz
- TFLN EOM · dual-octet shapers (2×8 genesis)
- 2–3 AlGaAs distributed OPC cells (~15–20 mm period)
- GF(16) FAU — GFADD/GFMUL/GFINV kernels
- Optional 3–6 dB on-chip squeeze
- Q target: 0.08–0.15 < 0.2 (Region I)
| Element | GALAXY | NOVA+ |
|---|---|---|
| L5 boundary layer | 2–3 AlGaAs cells | 8–100+ mesh nodes |
| Mesh period | ~15–20 mm sparse | 2–5 mm distributed |
| CV injection | 3–6 dB optional | Mandatory ≥6 dB |
| Octet tiling | 2×8 genesis | 8×8 full scale |
| Q target | 0.08–0.15 | ≥0.25 |
| Signoff | T-GALAXY matrix | T-NOVA+ |
T-GALAXY milestone gating chain
T-GALAXY-01 GF(16) gate F_ρ > 0.95 [to-be-tested] T-GALAXY-OPC AlGaAs distributed OPC η [to-be-tested] ──gates──► NOVA OPC claims T-GALAXY-CAL 240-pair calibration σ_φ ≤ 0.05 rad [to-be-tested] T-GALAXY-Q Pump-power vs phase fidelity Q < 0.2 [to-be-tested] T-GALAXY-SQZ 3–6 dB on-chip squeeze [roadmap]
Stability phase diagram
Q vs d — canonical diligence artifact for the STAR-PHASER / QUASAR lineup.
OPC margin = 2×
Crosses SUPER, THETA. Amber curve vs d. Tightens as Bcomb → BOPC.
Npair = 10⁴
SUPER crisis at 16k pairs. ChiL++ mandatory. Linear + periodic OPC breaks.
FWM · AWG · Qmax
NFWM=10⁶ crosses SUPER–THETA. AWG limit at d≈32–64. Red zone Q>1.2 at LOTUS.
| Chip | d | Q (design) | Region | Brand | Annotation |
|---|---|---|---|---|---|
| GEMINI | 2 | ≪ 0.05 | I | STAR-PHASER | TODAY — v1 ships |
| SOLAR | 10 | ~0.08 | I | STAR-PHASER | ℤ₁₀ composite |
| GALAXY ★ | 16 | ~0.12 | I | STAR-PHASER | QUASAR-transition prototype |
| TETRIS | 32 | ~0.15 | I | STAR-PHASER | Last pure STAR-PHASER |
| — — — BOUNDARY Q = 0.2 — — — | |||||
| NOVA | 64 | ~0.35 | II | QUASAR | First QUASAR · WS12 synthesis |
| SUPER | 128 | ~0.50 | II | QUASAR | Crisis zone · OPC+squeeze mandatory |
| THETA | 256 | ~1.0 | III | QUASAR | Field-computing experimental |
| LOTUS | 512 | ~1.4 | IV | QUASAR | Apex · speculative-framework |
Lineup maturity table
Demonstrated vs roadmap vs speculativeDemonstrated / v1 today
→GEMINI: Path qubits, SFWM class — v1 ships. External: Kues D=10 qudit, Lu 64×64 FB unitaries, Fülöp octave microcomb.
Roadmap / designed-target
→SOLAR d=10 stack · GALAXY 16-bin tile + T-GALAXY · TETRIS 32-tooth AWG stress · NOVA G46 tile + ChiL · SUPER T-SUPER OPC+squeeze.
Speculative-framework / experimental
→THETA Q≈1 field computing · LOTUS Region IV Q>1 · continuous LOTUS mesh · field-QEC at d=512. Campaign blockers: single-photon OPC [to-be-tested]; OPC ≠ QEC [never-state].
Five failure mechanisms
Coupled power laws in d — no single knob fixes all. Δf table is primary mitigation authority.
Five scaling laws
Tap to expand each lawLaw I — OPC bandwidth vs comb span
→B_comb = (d − 1) · Δf must satisfy B_comb ≤ B_OPC / m_margin m_margin = B_OPC / B_comb (B_OPC ≈ 12 THz, m_margin ≥ 1.2 minimum)
First warn: SUPER (~2.7×). THETA @ Δf=50 GHz explicitly FAILS (12.75 THz > 12 THz). Mitigation: reduce Δf; split-band OPC (LOTUS).
Law II — Pairwise phase budget
→N_pair = d · (d − 1) (directed pairwise phase relationships)
ChiL stress begins TETRIS (992). SUPER crisis (16,384). THETA (65,280) FAIL without continuous OPC + global φ₀. Mitigation: octet factorization (8× reduction).
Law III — FWM pathway explosion
→N_FWM ≈ d² · (d − 1) scales ~O(d³)
NOVA ~262k pathways (WARN). LOTUS ~134M (FAIL). Failure: gain competition, thermal runaway, crosstalk floors. Mitigation: pump rotation, pathway nulling, reduce γ·P·L in Z1.
Law IV — AWG / routing resolution
→Δλ_bin ≈ λ² · Δf / c @ λ ≈ 1550 nm
TETRIS monolithic AWG flagged (WARN). LOTUS 512-ch at limit (FAIL). Lower Δf helps Law I but worsens Law IV — coupled optimization. THETA Δf=25 GHz is Pareto point.
Law V — Nonlinear duty / Q ceiling
→Q(d) = (H_NL + H_OPC + H_CV) / H_linear → Q > Q_max(d) ⇒ instability
Running GALAXY at NOVA pump duty → Q > 0.2 → thermal crosstalk catastrophe. Region III–IV stability is Speculative-framework. Mitigation: duty caps per zone, SBS guard, per-chip Q setpoint.
Canonical Δf recommendation table (WS13-MANIFEST authority)
| Chip | d | Δf | Bcomb | OPC margin @12 THz | Primary law | Status |
|---|---|---|---|---|---|---|
| GEMINI | 2 | N/A | — | — | — | v1 today |
| SOLAR | 10 | 50 GHz | 450 GHz | ≫ (~27×) | — | PASS |
| GALAXY | 16 | 50 GHz | 750 GHz | ≫ (~16×) | Law II rehearsal | PASS |
| TETRIS | 32 | 50 GHz | 1.55 THz | ~7.7× | Law IV WARN | PASS |
| NOVA | 64 | 50 GHz | 3.15 THz | ~3.8× | Law III WARN | PASS |
| SUPER | 128 | 35 GHz | 4.45 THz | ~2.7× | Laws II+III | WARN |
| THETA | 256 | 25 GHz | 6.38 THz | ~1.9× | Law I WARN | WARN |
| LOTUS | 512 | 20 GHz | 10.2 THz | ~1.2× | Law I + split-band | WARN* |
* LOTUS ~1.2× assumes split-band OPC — single-band operation FAILS Law I. Bcomb = (d−1)·Δf.
| Law | Failure mode | Scales as | First warn | First fail |
|---|---|---|---|---|
| I | OPC band overflow | Bcomb = (d−1)·Δf | SUPER | THETA @ Δf=50G |
| II | Phase calibration | Npair = d(d−1) | TETRIS | SUPER (16k) |
| III | FWM crosstalk | NFWM ≈ d²(d−1) | NOVA | LOTUS (134M) |
| IV | AWG/demux resolution | Δλ ≈ λ²·Δf/c | TETRIS | LOTUS |
| V | Q / thermal runaway | Q(d, duty) | NOVA | mis-run GALAXY |
Never-state list
Prohibited public language — content review gate on all chip and framework pages.
Never-state register NS-01–NS-15
Tap to expand · correct framing includedShipping · encoding · QEC · squeeze
→- NS-01: ~~"GALAXY/NOVA/SUPER ships today"~~ → only GEMINI is v1; others are roadmap/research
- NS-02: ~~"GF(d) = base-d decimal"~~ → use GF(2^m) notation
- NS-03: ~~"SOLAR is GF(10)"~~ → structured composite qudit (ℤ₁₀ ring)
- NS-04: ~~"OPC replaces QEC"~~ → OPC pre-layer + digital QEC
- NS-05: ~~"Squeezing optional from SUPER up"~~ → OPC + squeeze mandatory
STAR-PHASER · THETA · photon count · FT
→- NS-06: ~~"STAR-PHASER handles d≥64"~~ → QUASAR from NOVA
- NS-07: ~~"THETA/LOTUS demonstrated"~~ → speculative-framework / experimental
- NS-08: ~~"Region IV product"~~ → long-term research horizon
- NS-09: ~~"16/64/512 bins = 16/64/512 photons"~~ → single-photon qudit
- NS-10: ~~"Fault-tolerant qudit processor today"~~ → roadmap
Gates · bases · AlGaAs · OPC band · extrapolation
→- NS-11: ~~"Native GF quantum gates on silicon"~~ → GF algebraic control layer
- NS-12: ~~"QUASAR = different base"~~ → dimension d qudit system
- NS-13: ~~"AlGaAs OPC demonstrated at QLT"~~ → [to-be-tested] T-GALAXY-OPC
- NS-14: ~~"Single-band OPC covers LOTUS"~~ → split-band OPC
- NS-15: ~~"Linear circuit extrapolates to THETA"~~ → field-computing architecture
- S81 — QUASAR Lagrangian Ψ=(A,Φ,Ξ)
- S82 — Q-metric Regions I–IV
- S83 — Boundary-layer fab 7-layer stack
- S84 — Five scaling laws + Δf table
- S85 — Stability diagram + never-states
Research corpus: 7-Research-Campaign/WS13-StarPhaser-Chip-Lineup/S81–S85. Promotion to canonical docs requires CAMPAIGN-MANIFEST ¶7 human approval.
Framework visuals for asset production
Fig Q1
Fig Q2
Fig Q3
Fig Q4
Fig Q5
Fig Q6From framework to chips
QUASAR framework consolidates WS13 diligence. Explore the GALAXY prototype, today's GEMINI product, and the GF(64) encoding roadmap.