Blind Transpiler: An open-source library for universally blind and homomorphic quantum computations

Overview

Blind quantum computation (BQC) allows a client with limited resources to delegate complex quantum tasks to a remote server without revealing the input data or the algorithm being executed. While theoretically robust, the lack of standardized software tools has historically hindered the rapid prototyping and experimental validation of these protocols. This paper introduces the first open-source library, blindtranspiler, designed to bridge this gap by automating the transpilation of standard Qiskit circuits into their blind counterparts.

Approach and Architecture

The library is built with a modular, four-layer architecture that separates core Qiskit primitives from the logic of blind transpilation. At its core, the library takes a standard quantum circuit as input and converts it into a BQCInstruction object, which contains a sequence of modified gate objects (BOPs). These objects encapsulate the necessary classical control and delegation instructions required to maintain blindness.

By decoupling the transpilation controllers from the specific translation rules, the library supports multiple BQC primitives, including:

• Quantum Homomorphic Encryption (QHE) for data security.
• Universal Blind Quantum Computation (UBQC) for both data and algorithmic security.

This design allows researchers to easily swap between different protocols or add new translation rules without refactoring the entire codebase. The library also includes built-in support for recursive decryption of parametric gates, which is essential for efficient variational quantum algorithms.

As quantum computing moves toward a cloud-based service model, ensuring the privacy of proprietary algorithms and sensitive data is critical. The blindtranspiler library provides a standardized, reproducible way for researchers to simulate and validate secure delegation protocols. By lowering the barrier to entry for BQC research, this tool facilitates the development of secure quantum applications, such as private cloud computing, secure federated learning, and blind variational classifiers, in both simulation and hardware-ready environments.