Optimization
academic program
The platform built to run research on real QPUs
Turn QPU computations into a powerful, practical tool for your research. Hardware performance, optimized routines, reproducibility, collaboration — we build the tools to maximize the impact of your work.
Everything your lab needs, in one place
Write experiments, manage jobs, collaborate with your research team.
Qiskit Interface
Import your Qiskit code and extract maximum performance on QPUs using Haiqu SDK tools. Install the environment and run your circuit with the best possible performance.
QPU Backend Access
Direct access to IBM, IonQ, IQM, Rigetti and OQC quantum processors, classical CPU and GPU simulators in one place.
Job tracking, versioning, analytics
Complete reproducibility of computations: tracked backends, calibrations, seeds, compilation settings and metadata so any result can be independently reproduced at any point in time.
Collaborative work environment
Managed access to complete experiment code, data, and artifacts. Share and collaborate on your quantum projects with your dedicated team.
QPU Budget Control
Estimate QPU cost before the run based on factors including hardware usage time, circuit complexity and provider-specific pricing. Minimize unnecessary costs via automatic batching and redundant circuit execution caching.
Tools to achieve the best possible performance
Circuit compression
QPU noise-aware approximate compiling reduces circuit depth by up to two orders of magnitude.
Error mitigation suite
A set of optimized, computationally efficient and composable Error Mitigation, Suppression and Detection routines to extract the most out of the QPU performance. QML-specific, lightweight EM.
Data loading and state preparation
State-of-the-art tensor methods for quantum state preparation, feature loading, distribution encoding. Shallow and noise-resilient quantum circuits for loading large scale data.
Application subroutines
A suite of optimized subroutines and tools: parameterized quantum circuit pre-training, equivariant QML ansatze, classical and quantum optimization, compressed SKQD and others.
Quantum at Work — Experiments on the platform
Check out some of the experiments run on the platform and learn from them.
SIMULATION
Quantum computational fluid dynamics (CFD) simulation
Data loading
Block vector loading of 64×64 grayscale image
Performance in realistic experimental loops
Haiqu doesn't treat middleware as a stack of sequential black boxes. Our infrastructure understands the structure of your quantum workflow and manages your quantum resources accordingly. See how it performs.
Extend executable circuit depth
Haiqu uses tensor network and predictive ML methods to approximately compile circuits, taking into account hardware topology and device noise, optimizing for execution accuracy on actual noisy QPUs.
Application subroutines
Whatever your research project, the SDK provides powerful building blocks optimized for performance on actual noisy QPUs: tensor network pre-training for QML, compressed Krylov circuits for diagonalization, or efficient Variational Function Tomography (VFT) for state readout in CFD applications, among many others.
Minimize the impact of QPU noise
Haiqu’s lightweight EM suite comprises optimized sets of routines (REM, custom DD, Twirling, ODR, and others) for both observable and bit-string distribution mitigations, allowing up to 10^2 shot budget reduction for the same accuracy on hardware.
Efficiently encode data on QPUs
Initial state preparation is a key bottleneck in quantum algorithms. Haiqu’s tensor network methods provide shallow encoding circuits with controlled approximation on scales of 100s of qubits for quantum chemistry and condensed matter, CFD and statistical distributions. Dense encodings enable addressing large scale QML and optimization.
free for researchers
Apply Now for Free Access
Get the most out of real hardware for your research.
- Fill out the form
- Schedule a call with our team
- Get your access to the platform
- Start your research