dqpu

A Web3-Powered (Near), Decentralized Quantum Simulator with Verifiable Computation.

Workflow

The DQPU system is composed of 3 actors:

Clients: users who need to perform a quantum sampling paying a reward
Verifiers: delegates who check for data validity and detect cheating users; they receive a reward for checking quantum sampling result validity
Samplers: users who run quantum samplers (either simulator or real quantum computers) and receive a reward for doing sampling

The following process outlines how clients can submit quantum circuits for sampling using the DQPU contract:

Client Submits Job: A Client sends a quantum circuit along with a reward to the DQPU smart contract. The circuit data is uploaded to a distributed file storage system like IPFS. The smart contract adds the job to a queue in a 'pending-validation' state with the associated reward.
Verifier Validates Circuit: A Verifier¹ validates the submitted circuit. This might involve checks for syntax errors or ensuring the circuit is within allowed parameters. The verifier also adds special verification elements (traps) into the circuit and add the new circuit to the contract². Once validated, the job moves to a 'waiting' state, becomes 'invalid' otherwise.
Simulation or Hardware Execution: A Sampler retrieves a job from the waiting list. It then either simulates the circuit on a software program or executes it on real quantum hardware, depending on the job requirements and available resources. The simulation or execution result is submitted back to the smart contract with a cautional deposit (a percentage of the reward). The job status changes to 'validating'.
Verifier Checks Result: The same Verifier from step 2 examines the returned result. The Verifier specifically checks the traps inserted earlier to ensure the result hasn't been tampered with. If the trap verification succeeds, the job status is updated to 'executed' and the Sampler account receives the reward, while the Verifier receives a percentage of this reward. If the trap verification fails, the job returns in 'waiting' state (and the Verifier receives the cautional deposit of the Sampler).
Client Receives Result: Once the job is marked as 'executed', the Client can retrieve the final result from the smart contract.

Installation

python setup.py install

Install IPFS:

https://docs.ipfs.tech/install/command-line/#install-official-binary-distributions

Usage: running a sampling job

The workflow described before is hidden to the final user: DQPU can be used seamleassy as any other quantum backend as any other quantum sampler. Currently DQPU implements a qiskit wrapper, a low level library for accessing the system primitives and a cli tool.

Qiskit example

import time
import qiskit
from qiskit.providers.jobstatus import JobStatus 
from dqpu.qiskit import DQPUProvider, DQPUService

# Create a quantum circuit
qc = qiskit.QuantumCircuit(2)
qc.h(0)
qc.cx(0, 1)

# Inizialize the service by providing the Near account
service = DQPUService()
service.setAccount("...")

# Run the sampling
backend = DQPUProvider().get_backend('dqpu_simulator')
job = backend.run(qc, reward="0.001")

# Wait for the job
while job.status() is JobStatus.RUNNING:
    print(f'Job {job.job_id()} is still running, please wait')
    time.sleep(1)

# Get the result
counts = job.result.get_counts()
print(counts)

Low-level example

from dqpu import *

# TODO

Cli tool usage

Generic params:

```-n/--network`: Default: near-testnet
```-a/--account`: Account name / uri

Commands:

Submit a circuit job

$ dqpu-cli -a dqpu_alice.testnet submit --file ~/test.qasm --shots 1024 --reward 0.0001
JOBID

Submit a random circuit: job

$ dqpu-cli -a dqpu_alice.testnet submit-random
JOBID

Submit a job result

$ dpqu-cli -a dqpu_bob.testnet submit-result -i 8 -rf ~/test.qasm

Remove a job

$ dpqu-cli remove -i JOBID

Get job information

$ dpqu-cli info -i JOBID
Job: JOBID
Status: WAITING
Qubits: 4
Depth: 9
Circuit uri: ipfs://.../test.qasm

Get job status

$ dpqu-cli status -i JOBID
EXECUTED

Get job result

$ dpqu-cli get-result -i JOBID
{ "0010": 1024 }

Set job validity

$ dpqu-cli -a dqpu_owner.testnet set-validity -i 9 -v false

Set job result validity

$ dpqu-cli -a dqpu_owner.testnet set-result-validity -i 8 -v true

Usage: running a sampler node

A sampler node continuously pool the DQPU smart contract waiting for new job. When a new job appear, the sampler checks if it can perform the sampling with its hardware.

Every sampler node can implement its own Sampler class, adding supports to other simulators or to real quantum hardware. DQPU package offer 3 implementation:

AerSimulator: statevector simulator from qiskit
DaskSimulator: statevector simulator using Dask distributed computing library
NumpySimulator: statevector simulator using Numpy

After every sampled job, the node receives the reward.

dqpu-sampler -a sampler_account --max-deposit 0.1 --sampler aersimulator --max-qubits 21

Usage: running a verifier node

A verifier node continuously pool the DQPU smart contract waiting for new 'pending-validation' and 'validating-result' jobs. When a new job appear, the verifiers:

'pending-validation' job are checked for quantum circuit validity, and trap qubits are inserted
'validating-result' job are checked for trap verification

After every validation, the verifier receives a percentage of the job reward.

Verifier are special users initially selected by the smart contract creator; this will change in the future.

dqpu-verifier -a verifier_account

Contributing

Read CONTRIBUTING for details.

License

This software is licensed with Apache License 2.0.

Cite

@software{dqpu2024,
  author = {Davide Gessa},
  title = {dqpu: A Web3-Powered, Decentralized Quantum Simulator with Verifiable Computation },
  url = {https://github.com/dakk/dqpu},
  year = {2024},
}

About the author

Davide Gessa (dakk)

In this first version of the contract, Verifiers are trusted entities designated by the smart contract creator. ↩
This step will become private in future versions of the protocol. ↩

vekexasia/dqpu