Achieve better batteries! Mercedes-Benz achieves new breakthrough in quantum chemical simulation

On April 13, PsiQuantum announced a new analysis of how electrolyte molecules in lithium-ion batteries (LiB) are simulated on a fault-tolerant quantum computer.

A technical paper co-authored by PsiQuantum and Mercedes-Benz R&D, "Fault Tolerant Resource Estimation for Quantum Chemical Simulations: The Case of Lithium-Ion Battery Electrolyte Molecules," was published in Physical Review on April 7, 2022 Research [1], the paper explains in detail how fault-tolerant quantum computers will speed up the design of electric vehicle batteries, exemplified by lithium-ion (Li-ion) batteries.

With this technological breakthrough, automakers have been able to perform previously impossible battery chemistry simulations, enabling breakthroughs in next-generation battery designs.

Lithium-ion battery technology bottleneck

Lithium-ion batteries function by moving charges from one electrode to the other through a liquid electrolyte material during charge and discharge cycles. The rollout of electric vehicles depends on the development of faster-charging, longer-lasting battery technology, which is also a key enabler of the transition from internal combustion engines: new and improved electrolytes will have a major impact on all aspects of battery performance, including energy density (efficiency). ), charging speed, battery life, range, cost and safety.

However, the current development of new lithium-ion batteries involves a great deal of trial and error. One goal of battery development is to find additive chemicals that can enhance electrical current, and evaluating potential additives requires accurately modeling how their presence affects electrolyte molecules. In principle, this slow and expensive process of R&D could be greatly accelerated by simulating and validating new chemicals in computers, as is now the norm for applications such as aerodynamics, mechanical design, etc. But classical supercomputers have difficulty simulating the quantum behavior of molecules and related reactions, and quantum computers are expected to overcome this limitation.

Quantum chemical simulations of molecules can help us better understand the electrochemical reactions that occur in liquid electrolytes and serve as useful guides for the design of novel electrolytes. PsiQuantum has been working with Mercedes-Benz for a long time to evaluate how advanced quantum computers can revolutionize the design of lithium-ion (Li-ion) batteries. To understand the quantum part cost required for these research computations, PsiQuantum identified a set of related electrolyte molecules and produced detailed resource estimates for their simulation on specific fault-tolerant quantum computing architectures.

Photon-based fusion quantum computing scheme

The research team first made an abstract fault-tolerance cost estimate with parameters unrelated to fault-tolerant quantum computing architectures, such as the number of qubits and gates, and then compiled logical operations into standard terms for fault-tolerant architectures.

The researchers chose the "fusion-based quantum computing (FBQC)" scheme: the basic building blocks of FBQC computing are resource states, measurements, and feedforward operations; compared with the well-known "measurement-based quantum computing model (MBQC)", it can Efficient realization of nonlinear quantum gates. Another technique required for practice is photonics: physically resource states can be created with resource state generators (RSGs), because RSGs can emit resource states encoded with a limited number of photons. The subsequent required measurements can be made through a process called fusion: with only linear optics and photodetection.

A prominent feature of the photon-based FBQC scheme is that it can be interleaving [2]. Photons can be stored in an extremely long optical delay line without appreciably degrading the photon quality. Thus, a trade-off between device footprint and computation runtime can be easily made, possibly with less engineering overhead when building a large-scale quantum computer; of course, the overall cost of the algorithm will still largely depend on the length of the optical delay line .

That is, with optical fibers as quantum memory, scientists can use interleaving to perform the linear space-time trade-off of fault-tolerant quantum computing.

A schematic diagram of the structure used to generate the above fault-tolerant estimates. The RSGs are arranged in a two-dimensional plane, and different regions are responsible for different tasks. The rectangles in the middle correspond to the regions that produce logical qubits, each producing the temporal propagation of two logical qubits that propagate over time. Each logical region consists of an extraction block and a connection block: the former performs circuit extraction (based on a 15-to-1 protocol), and the latter receives the output magic state of the extraction block and performs the required "lattice surgery" surgery), the output magic state of each stage is delivered to the connector block of the higher stage through interleaved fibers. Periodic boundary conditions are indicated by arrows, using two magic state distillation (MSD) plants.

Resources needed to simulate chemical problems

Considering the characteristics of FBQC, the researchers conducted a detailed analysis of the resources required to simulate the above chemical problems. The results of this analysis are shown below:

Left: The ratio of magic state extraction (MSD) footprint to total computational footprint for different logical qubit counts and T-counts. Footprints are measured in the number of RSGs required. Assuming a linear data block with two multi-level 15-to-1 factories, resource estimates for other algorithms, such as simulations of the Fermi-Hubbard model, crystalline materials, FeMoco, and cracking RSA encryption, are plotted for ease of comparison. Resource estimates from quantum chemical algorithms are revised when necessary to produce characteristic energies that are chemically accurate. Assuming logical error rate parameters A = 0.45 and B = 1.35 (that is, the average of the one-qubit and two-qubit mechanisms considered by the research team), the initially prepared magic state has a logical Pauli error rate of 0.1%. Right: Footprint and time estimates for fault-tolerant quantum computation of various molecules based on cc-pVTZ and cc-pVDZ. Spatial and temporal resources can be traded off linearly using the interleaving method, as shown. The study plotted resource estimates for interleaving ratios between 1 and 1000, as such ratios can be achieved introducing negligibly high error rates. Assuming the logical error rate parameters A=0.5 and B=1.6, the logical Pauli error rate of the initially prepared magic state is 0.1%.

For the smallest instances that are classically intractable (eg: fully-configured interactive simulation of PF-6 using the cc-pVDZ basis at 1mHartree accuracy), the research team estimates that a total of 16382 logical qubits and

T gates, as shown in the following table:

The T count nT and qubit count nL for each molecule and basis set minimize the computational effort nT×nL.

After compiling, this equates to an estimated 24 million RSGs with a run time of less than 1 day at a moderate error rate, as shown in the following figure:

Assuming the interleaving ratio is Lintl=1, the space and time overhead of fault tolerance. This provides the fastest calculation,The price is the biggest footprint.

Introducing the interleaving technique, which allows a linear space-time trade-off between runtime and footprint. For example, using a stagger ratio of 24 for the same instance above, you can have a footprint of 1 million RSG and a runtime of 2.5 weeks.

Based on this resource estimate, having a magic state factory that produces magic states serially is suboptimal for the system under consideration for the simulation. For small quantum algorithms (i.e., a small number of qubits and gates), the footprint of the factory constitutes a large portion of a quantum computer; however, as the number of logical qubits required by the algorithm increases, the relative size of the magic state factory will change be smaller. Obviously under this mechanism the (relative) additional cost of adding another magic state factory will be small. Although it has been known that this will eventually happen as the number of logical qubits increases, the exact point at which this crossover occurs remains unresolved in the existing literature.

The research team filled that void: the answer depends on the number of qubits and the number of T-gates. If only one, or two, magic-state factories based on the 15-to-1 protocol are used, and logical T-gates are executed sequentially, the footprint of the factory does not exceed 2% of the entire quantum computer. Therefore, the research shows that there are some quantum algorithms with practical application value, which are not suitable for the relatively large-scale system of the magic state factory.

Conclusion

PsiQuantum has now mathematically demonstrated and proposed new ways to perform constant-time PPM: they can achieve breakthroughs in lithium-ion battery design by running this optimized algorithm on its utility-scale quantum computing architecture .

However, the effect of the lateral gates used in this procedure on threshold and subthreshold expansion is unclear; this doubt will also be left to future work. Future research in this regime should explore other Hamiltonian simulation algorithms that may be suitable for massively parallelization, as well as further algorithmic improvements for depth considerations, to realize the potential gain of reducing the computational burden of quantum chemistry problems.

As the auto industry ends its reliance on fossil fuels, society will increasingly rely on improvements in battery technology. The design of electrolyte additives is just one area of ​​battery design accelerated by fault-tolerant quantum computing. From developing novel cathode materials to multi-scale battery simulation, quantum computing can revolutionize a wide range of battery development. In all cases, only optimized fault-tolerant algorithms running on utility-scale quantum computing devices can achieve the breakthroughs carmakers seek.

Link:

[1] https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.4.023019

[2] https://arxiv.org/abs/2103.08612

[3] https://psiquantum.com/news/counting-qubits-for-better-batteries

[4] https://psiquantum.com/news/psiquantum-breakthrough-paves-way-to-dramatic-acceleration-in-ev-battery-design