Qubitclient

Overview

qubitclient is the unified Python package for quantum calibration data analysis. It combines traditional curve fitting, neural network analysis, LLM review, and real-time measurement control into a single, coherent system.

Installation

# Basic installation
pip install 'qubitclient'

# Full installation (all features)
pip install 'qubitclient[full]'

Requirements: Python 3.10+

Initialization

Configuration File

Initialize configuration files in your project directory:

qubitclient init

This generates qubitclient.json and .mcp.json in the current directory.

qubitclient.json structure:

{
  "url": "http://0.0.0.0:9801",
  "api_key": "your-proxy-api-key",
  "llm": {
    "api_key": "your-vllm-api-key",
    "base_url": "http://xx.xx.xx.xx:9091/v1",
    "model": "nv-community/Ising-Calibration-1-35B-A3B"
  },
  "generate": {
    "api_key": "your-vllm-api-key",
    "base_url": "http://xx.xx.xx.xx:9091/v1",
    "model": "Qwen/Qwen-Image-Edit"
  },
  "license": {
    "token": "your-license-token"
  }
}

Client Initialization

from qubitclient import QubitScopeClient, QubitNNScopeClient

# Numerical fitting client (auto-loads qubitclient.json)
scope_client = QubitScopeClient()

# Neural network client
nn_client = QubitNNScopeClient()

Sub-Skills

qubitclient-scope

Numerical curve fitting and parameter extraction for quantum experiments.

Key capabilities:

• S21 peak detection (single/multi-peak)
• T1/T2 relaxation time fitting
• Rabi oscillation analysis
• DRAG pulse optimization
• Randomized Benchmarking (RB)
• Single-shot readout fidelity
• Power shift characterization

qubitclient-nnscope

Neural network based analysis for spectrum and curve segmentation.

Key capabilities:

• S21 peak detection with deep learning
• 2D spectrum segmentation (COSINE, POLY curve types)
• S21 vs Flux parameter curve extraction
• Power shift curve segmentation

qubitclient-vqa-review

LLM-powered visual question answering for quantum calibration results.

Key capabilities:

• Automated analysis review
• Calibration result interpretation
• Visual inspection of measurement data

qubitclient-control

Real-time measurement control via MCP (Model Context Protocol).

Key capabilities:

• Live measurement streaming
• Parameter updates during measurement
• Automated calibration workflows

Usage

# Scope tasks (numerical fitting)
from qubitclient import QubitScopeClient, TaskName
client = QubitScopeClient()

# NNScope tasks (neural network)
from qubitclient import QubitNNScopeClient, NNTaskName, CurveType
nn_client = QubitNNScopeClient()

# MCP Control tasks
from qubitclient.ctrl import MCPClient
mcp_client = MCPClient()

Task Categories