Announcement • Jul 09
MicroCloud Hologram Inc. Announces Practical Approximate Quantum Multiplier Technology For NISQ Devices
MicroCloud Hologram Inc. has made an important breakthrough centered on the NISQ (Noisy Intermediate-Scale Quantum) environment — the practical approximate quantum multiplier technology. This technology takes approximate computing as its core idea and conducts systematic optimization targeting the two key performance bottlenecks of quantum circuit depth and T-gate count, providing a feasible path that balances efficiency and precision for current noisy quantum devices. MicroCloud Hologram Inc. first started from the most basic arithmetic unit — the adder — and performed a structural reconstruction of the traditional quantum adder circuit. Standard quantum adders usually rely on a bit-by-bit carry propagation mechanism, with circuit depth growing linearly with the number of input bits and requiring a large number of T gates to implement non-Clifford operations. In the newly proposed approximate adder, by weakening the carry precision of some low-weight bits and truncating or simplifying the originally strictly executed carry chain, a constant-depth (O(1)) circuit structure is achieved. This design is not a simple deletion of logic, but ensures through probability analysis and error modeling that the impact of errors on the overall computation remains within an acceptable range. At the specific implementation level, the technology proposes four approximate adder circuits with different precision levels. These circuits introduce parameterized control mechanisms during design, allowing users to flexibly choose between precision and resource consumption according to specific application requirements. For example, in application scenarios with low error sensitivity, a highly compressed version of the adder can be selected to achieve extremely low circuit depth and T-gate count; while in scenarios requiring higher computational precision, a more complete carry path can be enabled to improve result accuracy. This adjustable precision design concept enables quantum arithmetic modules to possess performance gear characteristics similar to those in classical computing for the first time. After completing the construction of the approximate adder, MicroCloud Hologram Inc. further used it as a core module to build a complete approximate quantum multiplier. Unlike traditional multipliers that rely on multi-stage addition accumulation structures, this multiplier achieves significant simplification of the overall circuit structure by optimizing some product generation paths and combining them with approximate addition units. Particularly in terms of T-gate count, the technology significantly reduces implementation costs by decreasing the number of high-cost non-Clifford gates used. This is especially important in current quantum hardware architectures, because T gates typically require complex fault-tolerant encoding and magic state distillation processes to implement, with resource overhead far higher than that of Clifford gates. It is worth noting that this approximate multiplier outperforms existing mainstream quantum multiplier schemes in both key metrics of depth and T-count. The reduction in circuit depth means that quantum states are exposed to the noisy environment for a shorter time during evolution, thereby reducing the impact of decoherence; while the reduction in the number of T gates directly lowers the cumulative risk of gate operation errors. This dual optimization enables the multiplier to achieve a higher execution success rate on actual NISQ devices. To verify the practical effectiveness of this technology, MicroCloud Hologram Inc. conducted tests on various quantum simulation environments and real quantum hardware. The experimental results show that, under the same input scale, although the approximate multiplier exhibits slight deviations in result precision, the overall error distribution presents controllable characteristics without systematic bias. At the same time, due to the significant reduction in circuit depth, the fidelity of the final output results is actually superior to that of traditional exact multipliers. This phenomenon indicates that, in noise-dominated NISQ environments, moderate approximation can instead improve overall computational quality. Further analysis shows that this technology has broad potential in error-tolerant applications. For example, in quantum machine learning tasks, the model itself has certain robustness to small perturbations in input data, so the approximate multiplier can fully replace exact arithmetic modules, thereby significantly reducing training costs. In quantum optimization problems, the objective function usually has smooth characteristics, and local errors do not significantly affect the search process for the global optimal solution, which also provides application space for approximate computing. From an engineering implementation perspective, this technology also possesses good scalability. Since its core structure is based on a modular adder design, it can be easily integrated into more complex quantum arithmetic systems, such as quantum dividers, exponential operation modules, and polynomial computation circuits. In addition, the design is compatible with existing quantum compilation frameworks and can be directly mapped for execution on mainstream quantum hardware platforms without requiring additional hardware support. At a more macroscopic level, this technology reflects an important development trend: quantum algorithm design is shifting from theoretical optimality to engineering usability. The introduction of approximate computing is a typical embodiment of this shift. In the future, this technology is expected to become an important component in building practical quantum software stacks. As the scale of quantum hardware gradually expands and algorithm complexity continues to increase, the performance requirements for underlying arithmetic modules will also keep rising. Approximate computing strategies can not only be applied to multiplication operations but can also be extended to a wider range of quantum logic designs, thereby forming a complete set of low-resource quantum computing methodologies. MicroCloud Hologram Inc.'s practical approximate quantum multiplier technology for NISQ devices not only achieves breakthroughs in technical indicators but also provides a new direction of thinking at the methodological level. It demonstrates that, in the current transitional stage of quantum computing development, the limitations caused by insufficient hardware can be effectively compensated by introducing cross-disciplinary ideas and performing engineering optimizations.
