| name | distributed-quantum-computing |
| description | Distributed Quantum Computing architecture and patterns. Apply when designing multi-QPU systems, quantum communication protocols, or scaling quantum computing beyond single device limitations. |
| metadata | {"openclaw":{"emoji":"⏛","source":"arxiv:2212.10609,arxiv:2404.01265","authors":["Caleffi et al.","Barral et al."],"year":2024}} |
Distributed Quantum Computing
Framework from arxiv:2212.10609 & arxiv:2404.01265 - scaling quantum computing via distributed paradigm.
Core Problem
Single QPU Limitation:
- Current: ~100-1000 noisy qubits
- Need: ~10,000-1,000,000 noise-free qubits for practical quantum advantage
- Solution: Distributed quantum computing - multiple QPUs communicating and cooperating
Architecture
┌─────────────────────────────────────────────────────────────┐
│ Distributed Quantum Computing System │
├─────────────────────────────────────────────────────────────┤
│ │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ │
│ │ QPU-1 │ │ QPU-2 │ │ QPU-3 │ │
│ │ 100 qubits│ │ 100 qubits│ │ 100 qubits│ │
│ └──────────┘ ┌──────────┘ └──────────┘ │
│ │ │ │ │
│ └──────────────┼──────────────┘ │
│ │ │
│ ┌───────▼───────┐ │
│ │ Quantum Network│ │
│ │ (Entanglement │ │
│ │ Distribution)│ │
│ └───────┬───────┘ │
│ │ │
│ ┌───────▼───────┐ │
│ │ Distributed │ │
│ │ Quantum Gates│ │
│ └───────┬───────┘ │
│ │ │
│ ┌───────▼───────┐ │
│ │ Scheduler │ │
│ │ (Task Distribution)│ │
│ └───────┴───────┘ │
│ │
└─────────────────────────────────────────────────────────────┘
Key Components
1. Quantum Communication Protocols
| Protocol | Purpose |
|---|
| Teleportation | Transfer quantum state between QPUs |
| Entanglement swapping | Create entanglement across non-directly connected QPUs |
| Quantum routing | Route qubits through quantum network |
2. Distributed Quantum Gates
- Non-local gates: Gates acting on qubits across different QPUs
- CAT gates: Communication-Assisted Teleportation gates
- Telegate protocol: Teleport gate execution to remote QPU
3. Scheduler
- Task decomposition across QPUs
- Minimize communication overhead
- Balance QPU workload
Challenges
| Challenge | Description | Current Solutions |
|---|
| Entanglement distribution | Create/maintain entanglement across QPUs | Quantum repeaters, entanglement swapping |
| Noise propagation | Errors spread across distributed system | Distributed error correction |
| Communication overhead | Teleportation requires classical communication | Minimize non-local gates |
| Synchronization | QPUs must be synchronized | Distributed quantum clock |
| Scalability | Network topology limits scaling | Hierarchical architecture |
Design Patterns
Pattern 1: Quantum Circuit Partitioning
def partition_circuit(circuit, n_qpus):
"""Partition quantum circuit across multiple QPUs."""
partitions = []
for i in range(n_qpus):
partition = extract_local_gates(circuit, qpu_range=i)
non_local_gates = extract_non_local_gates(circuit, qpu_range=i)
partitions.append({
'local': partition,
'non_local': non_local_gates,
'communication': estimate_teleportation_cost(non_local_gates)
})
return optimize_partition(partitions)
Optimal Gate-Packing (arXiv:2606.18494): The primary cost metric for multi-QPU partitioning is e-bit count (maximally entangled qubit pairs between QPUs).
- Build interaction graph from two-qubit gates (nodes=qubits, edges=gates)
- Apply gate-packing: group commuting gates before partitioning to reduce cross-QPU edges
- Use graph partitioning (METIS, spectral, ILP) to assign qubits to QPUs
- Rewrite cross-QPU gates via teleportation, preserving unitary equivalence
- Validate with unitary matrix comparison and hardware execution
Pitfall: Always optimize the circuit before partitioning. Partitioning an unoptimized circuit wastes e-bits on gates that could be eliminated. For QFT circuits, exploit the hierarchical structure of controlled-phase gates to reduce e-bit cost beyond generic partitioning.
Pattern 2: Entanglement Distribution Network
QPUs connected via quantum network:
- Direct links: High-fidelity entanglement
- Indirect links: Entanglement swapping via repeaters
- Topology: Minimize shortest path between any two QPUs
Federated Quantum Learning
- See
q-anchor-federated-quantum-learning skill for QFL with ZNE-guided correction addressing double-drift (client drift + hardware bias)
Key Pattern: Distributed Quantum Error Correction
- Surface code adapted for distributed QPUs
- Parity checks across QPU boundaries
- Requires entangled ancilla qubits
Metrics
| Metric | Target |
|---|
| Entanglement fidelity | > 0.99 for distributed gates |
| Communication latency | < 1ms for teleportation |
| QPU utilization | > 80% parallel execution |
| Error rate | < 0.001 per distributed operation |
Applications
- Distributed Shor's algorithm: Factor large numbers across QPUs
- Distributed QAOA: Optimize large-scale optimization problems
- DQAOA simulator (arxiv:2606.26297): Qiskit-compatible simulator for multi-QPU QUBO. Canonicalizes QUBO → cost Hamiltonian → variable allocation → circuit construction. Runtime optimizations: circuit reuse, batched eval, parallel multi-start. Power unit commitment case: distributed recovers same optimal as monolithic. Cross-QPU couplings require remote operations → more demanding. See
distributed-qaoa-simulator skill.
- Quantum simulation: Simulate larger quantum systems
- Quantum machine learning: Train larger quantum models
Relation to OpenClaw
OpenClaw's distributed agent architecture parallels DQC:
- Multiple QPUs → Multiple agent instances
- Quantum communication → Agent communication protocols
- Distributed gates → Cross-agent tool calls
- Scheduler → Agent orchestrator
Sources: arxiv:2212.10609 (Caleffi et al., 2024), arxiv:2404.01265 (Barral et al., 2024)
