Editor: Editorial Department HXY
Microsoft announced a record 12 entangled logic qubits, realizing the simulation of end-to-end quantum + AI + cloud chemical molecules for the first time. This initiative means that mankind has entered a new era of quantum computing.
Humanity has entered a new era of quantum computing!
Just now, Microsoft officially announced that it has created the best performance "logic qubits" in history-12.
Moreover, this is the largest number of quantum entangled bits with the highest fidelity.
This breakthrough was made possible by improving Quantinuum's leading quantum computer, H2 (56 physical qubits).
This is a testament to Microsoft's expertise in world-class error correction.
In less than three months, they were able to triple the number of reliable logical qubits.
In addition, when entangled in a complex state of all 12 logical qubits for "deeper" quantum calculations, the circuit error rate is increased by a factor of 22 compared to the corresponding physical qubits.
These results were completed on the Azure Quantum compute platform, and the large-scale computation of logical qubits was realized using ion trap hardware.
Most excitingly, Microsoft showcased the first end-to-end chemical simulation.
AI can process large-scale data, quantum can enable complex calculations and higher precision results, and the direct development process can be compressed from years to just a few days.
In the future, scalable quantum computing will pave the way for scientific discovery, especially in the fields of chemistry, physics, and life sciences.
At the same time, Microsoft also announced that it has joined forces with Atom Computing to build the world's most powerful computer, and the next step is to achieve 1000+ high-performance logic qubits.
Some netizens said that if there is enough energy, from the perspective of quantum computing, the computing cost will be reduced to 0 (especially for training AI models). That's when Scaling Law really came to an end.
We are on the verge of the era of quantum AI.
A record 12 logical qubits
Microsoft and Quantinuum's genesis was made possible by two key elements:
One is the Azure Quantum qubit virtualization platform and the other is the H2 trapped-ion quantum computer.
In 3 months, the number has increased by 3 times
In April this year, Microsoft and Quantinuum teamed up to virtualize the qubit system and use it to capture qubits with H-series ion traps.
Unexpectedly, they created 4 logical qubits from 30 physical qubits on H2, setting a new record at that time.
Moreover, the error rate at the logical level is 800 times stronger than the physical error rate.
Now, Microsoft has extended Microsoft's error correction algorithm and optimized H2. As a result, the improved H2 quantum computer achieves 56 qubits with a two-qubit fidelity of 99.8%.
On top of that, the team unlocked 12 highly reliable logical qubits.
Previously, in the Bell state preparation, they entwined two logical qubits.
And in this study, all 12 logical qubits, entangled in a more complex arrangement, can be called CAT states, or Greenberger-Horne-Zeilinger (GHZ).
When these logic qubits are entangled, they have a circuit error rate of 0.0011.
This is a 22-fold improvement over the circuit error rate of the corresponding physical qubits (0.024).
8 qubits, 5 rounds of repetitive error correction
To further validate, the researchers performed several fault-tolerant calculations using improved logical qubits.
On 8 logical qubits, 5 rounds of repetitive error correction can be successfully performed.
In addition, 8 logical qubits are used in the error correction process to perform fault-tolerant calculations, successfully demonstrating the combination of logical entanglement operations and multiple rounds of quantum error correction.
They have a circuit error rate of 0.002, which is an 11-fold improvement over the corresponding physical qubits.
This is also the first time that Microsoft has demonstrated the power of combining computation and error correction, and the ability of logical qubits to reliably perform deeper quantum calculations, thus paving the way for fault-tolerant quantum computing.
The first end-to-end chemistry simulation to accelerate scientific discovery
So, what is the use of Microsoft's quantum computing experiments?
Ultimately, it's about accelerating scientific discovery.
To do this, the researchers conducted end-to-end chemical simulations, combining cloud HPC, AI, and high-reliability quantum computing to create a hybrid workflow.
Address: https://arxiv.org/pdf/2409.05835
In quantum computing, 2 logical qubits (created by a qubit virtualization system and H1) are used to construct the active space ground state of an important catalytic intermediate, which is then measured
This is shown in Figure 1 below.
Fig.1 The reaction pathways of P-N-N-P iron catalysts predicted by AutoRXN for the first time, which have a significant impact on the function of reaction products
The measurements are then handed over to AI to estimate the ground-state energy of the active space.
This is the first time that HPC, AI, and quantum computing hardware have been combined to solve a scientific problem.
Next, we will explain how Microsoft can use such a hybrid workflow to achieve practical applications in the field of chemistry.
First, the research team used AutoCAS and AutoRXN to perform high-performance computing (HPC) simulations in Azure Quantum Elements to identify the active space and reaction pathways of catalysts, respectively.
They then optimized the error detection code used in a custom quantum algorithm to simulate the quantum behavior of the active space on two logical qubits.
Next, a method called classical shading is used to measure logical qubits to generate classical data.
This method uses the measurement results as classical data to train AI models to understand the quantum properties of molecules.
This "shadow data" is then combined with an AI model to determine the chemical properties of the catalyst and estimate the chemically accurate energy of the active space ground state (Figure 2).
To facilitate comparison, the researchers performed similar calculations on physical qubits.
Figure 2 Comparison of the accuracy achieved in estimating ground state energy in active space using physical and logical qubits. Logical qubit calculations produce better ground-state energy estimates with a 97% chance. The chemical accuracy is limited to 1.6 mHa of the classical calculation of the true ground state energy
Through the classical calculation method, the ground state energy of the catalyst active space is obtained, and the accuracy of the results based on hybrid quantum and AI is finally evaluated.
Using qubits to solve this problem does not show the advantages of scientific quantum, as classical computers can also derive it.
However, for some complex chemical problems, classical computational amplification is unlikely to be solved with high-precision methods, so quantum computers are the best choice.
The above is the whole process of the whole chemical simulation, and this concept is verified, which illustrates two problems:
- The first end-to-end workflow demonstration using quantum computing, high-performance computing (HPC), and AI to simulate and solve chemical problems.
- Quantum mechanics problems are solved with a high degree of accuracy. Logical qubits have better ground-state energy estimation than physical qubits.
These achievements demonstrate continued progress in scientific quantum superiority, and quantum computing milestones are achieved when quantum-classical hybrid supercomputers are able to solve scientific problems that classical computers cannot solve alone.
Quantum computing, ushering in a new era
Outside of Google, Microsoft has made a big bet on quantum computing.
Nadella has publicly stated that mixed reality, artificial intelligence and quantum computing are three groundbreaking technologies that will "shape" the world in the coming years.
Microsoft is ushering in a new era of computing, and its breakthrough in quantum computing is about unlocking the scientific potential to tackle some of the world's most pressing challenges.
That's one of the reasons why Azure Quantum, the first reliable quantum computing platform, was born.
They hope that through this platform, quantum computing can be scaled up.
Five months ago, Microsoft and others demonstrated logical qubits that are 800 times more reliable than physical qubits, and claimed that they had entered the next stage of solving practical problems with reliable quantum computers.
The main problem with today's noisy medium-scale quantum computers (NISQ) is that physical qubits are too noisy, error-prone, and unreliable in real-world applications.
Moreover, simply by increasing the physical qubits, it is not enough to achieve quantum error correction of the file.
That's why, there is a need to transition to reliable, high-fidelity logic qubits.
The combination of common multiple physical qubits can resist noise and maintain coherence over long computational periods.
However, quantum computing does not exist in isolation, it needs to be deeply integrated with the power of cloud HPC.
As a result, a new generation of hybrid quantum applications can solve humanity's most pressing challenge today – pioneering more sustainable energy solutions and life-saving treatments.
The Azure Quantum compute platform provides quantum computing in a variety of hardware architectures and supports industry-leading hybrid quantum applications.
All programs are housed in a secure, unified, and scalable cloud environment.
Create the next generation of hybrid quantum applications
By combining high-performance computing, AI, and quantum technologies on the same cloud platform, Azure Quantum, Microsoft is pioneering a new computing paradigm.
Azure Quantum leverages multiple qubit architectures and multiple chips to enable quantum applications to execute seamlessly, providing accelerated computing for a variety of application areas, such as generative chemistry and density functional theory (DFT).
The advantages of AI in large-scale data processing complement the unprecedented precision of quantum in complex computing, forming a powerful computing foundation that provides a secure, unified, and scalable hybrid computing environment.
This allows innovators to develop the best solutions to problems that are difficult or even impossible to solve on classic computers.
In this differentiated computing stack, researchers use quantum tools at the appropriate stage, plus Copilot's workflow collaboration, developer tools, classical supercomputing, AI collaborative reasoning, and multimodal models, in an iterative cycle.
In Microsoft's view, this approach can be hugely powerful, perhaps shortening the engineering cycle for R&D and solutions from years to days.