Oracle and Classiq have tested a 36-qubit quantum circuit simulation in OCI, showcasing AI-driven code generation for finance.
Oracle Corporation, in collaboration with software developer Classiq, has successfully showcased a method for generating and simulating quantum code using Oracle Cloud Infrastructure (OCI).
This initiative, part of a "proof of concept," focused on a portfolio optimization task based on the Markowitz model, encompassing 12 assets. A notable aspect of the experiment was Classiq's AI agent, which produced executable code from a natural language prompt in less than 15 minutes.
To test the algorithm, a node from the NVIDIA DGX A100 supercomputer was utilized within OCI. Leveraging the NVIDIA cuQuantum library, engineers managed to simulate a 36-qubit circuit, surpassing Classiq's typical limit of 29 qubits.
The simulations were executed on eight A100 GPUs, requiring around 512 GB of video memory for the complete state vector model of 36 qubits. The application of a hybrid QAOA algorithm with a chain depth of 730 enabled the exploration of 68.7 billion combinations. The entire simulation process lasted five hours, with results deviating from classical market benchmarks by just 4.63%.
Classiq clarified that the aim of the project was not to achieve quantum supremacy but to illustrate the synergy between AI-driven code generation and scalable cloud computing in sectors like finance and logistics.
There remains no indication that this approach outperforms existing classical methods. Instead, it serves as a demonstration of a technological framework rather than a leap in computational efficiency.
