Parallel Quantum Chemistry on Noisy Intermediate-Scale Quantum Computers

arXiv:2202.02417 · quant-ph, cond-mat.str-el, physics.chem-ph · Submitted 2022-02-04 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: I'm Kai, and with me are Mira and Lev, guest researcher.

Mira: Today's paper: "Parallel Quantum Chemistry on Noisy Intermediate-Scale Quantum Computers".

Kai: A novel parallel hybrid quantum-classical algorithm for solving the quantum-chemical ground-state energy problem on gate-based quantum computers has been presented,

Mira: First, who's behind it and why it matters.

Paper summary: Kai: So, to recap, we've established that this paper proposes a hybrid quantum-classical algorithm based on RDMFT decomposition using an adaptive cluster approximation to solve ground-state energy problems. The central thesis is that by breaking the full system into coupled subsystems, the problem becomes inherently parallelizable.

Mira: And what they claim is that this approach allows for treating much larger molecules than traditional VQE methods because it manages the computational scaling effectively through these approximations. They argue that this method introduces a new level of parallelization suitable for NISQ devices while retaining noise tolerance in its convergence behavior.

Lev: From an error correction perspective, the paper’s main contribution here is showing a viable pathway for applying quantum chemistry to realistic molecular systems, not just tiny toy models. It suggests that if we can implement this decomposition efficiently, it provides a structured way to tackle the complexity inherent in electronic structure problems on quantum computers.

Kai: Right, Lev; and the paper focuses on demonstrating this by introducing a hybrid quantum-classical algorithm designed specifically to compute the reduced density matrix functional on quantum hardware. It outlines the steps from preparing an ansatz state to performing the constrained minimization classically.

Mira: And what matters most is that they don't just propose a mathematical trick; they detail concrete techniques for reducing qubit count and program depth, such as using symmetries and local approximations of the interaction Hamiltonian to manage complexity.

Lev: I think the real significance lies in how they address the noise aspect directly in Section VI, where they compare noiseless simulations with genuine runs on IBM hardware. That comparison is vital because it grounds this theoretical framework in the reality of current experimental constraints for quantum computing.

Kai: Exactly, Lev; and when you look at their results for the half-filled Hubbard chain simulation, they show rapid convergence in just ten outer iterations in noise-free runs, which sets a benchmark for what's achievable on ideal systems. This gives us a baseline to compare against how this algorithm performs under the noise models they test later.

Mira: And then they move into genuine NISQ simulations and compare the convergence behavior there to those noiseless cases, which leads them to conclude something about the representability of the many-particle state on noisy quantum computers.

Lev: So, in essence, they're not just presenting a new formula; they are demonstrating a concrete methodology for how to approach ab-initio molecular dynamics simulations using quantum hardware that incorporates explicit considerations for noise limitations.

Conclusion: Kai: So, wrapping up this discussion on "Parallel Quantum Chemistry on Noisy Intermediate-Scale Quantum Computers," we have to consider the scope of the work presented by Schade, Bauer, Tamoev, Mazur, Plessl, and K¨uhne. The authors are clearly aiming at solving a problem that's far beyond what VQE can handle on current systems.

Mira: Indeed; the title itself signals the focus on intermediate-scale quantum computers, which means they aren't just targeting theoretical models; they are looking at practical implementation challenges and scaling limits imposed by real hardware constraints. It’s about bridging the gap between complex theory and practical application.

Lev: I think the implication is that if this approach works as described, it offers a scalable structure for tackling molecular problems ab-initio, which could eventually lead to more accurate force evaluations in molecular dynamics simulations using these quantum computers.

Kai: That’s right; and in simple terms, they are showing us how to leverage the structure of RDMFT decomposition to manage the complexity, making it feasible for larger molecules on current gate-based machines. It’s about finding a way to make these systems computationally tractable by leveraging quantum structure.

Mira: Ultimately, if this methodology proves robust under noisy conditions, it suggests that we have a more reliable path forward for using NISQ devices in high-level quantum chemistry calculations compared to purely variational methods.

Lev: And from an error correction standpoint, this work provides a concrete algorithm that could be integrated with existing quantum error-correction research to build systems capable of running these kinds of simulations effectively.

Kai: So, the big picture here is that the work points toward a method where we can use the structure of quantum mechanics itself to manage complexity on these intermediate-scale machines, opening up new avenues for ab-initio molecular dynamics calculations.

Robert Schade, Carsten Bauer, Konstantin Tamoev, Lukas Mazur, Christian Plessl, Thomas D. K¨uhne

Paderborn Center for Parallel Computing Paderborn University Department for Computer Science

quant-ph, cond-mat.str-el, physics.chem-ph

Submitted: 2022-02-04

Updated: 2022-08-11

Comments: 17 pages, 13 figures, 1 table

Journal ref: Phys. Rev. Research 4, 033160 (2022)

DOI: 10.1103/PhysRevResearch.4.033160

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 92/100

The gist: A novel parallel hybrid quantum-classical algorithm for solving the quantum-chemical ground-state energy problem on gate-based quantum computers has been presented, offering a path to treat much

Key concepts

Reduced Density-Matrix Functional Theory (RDMFT)
This is a way to describe the electronic structure of a molecule by focusing on reduced density matrices instead of the full system's density matrix. It breaks down complex problems into smaller, more manageable subsystems, which makes the overall calculation much simpler and inherently parallelizable for quantum computers.
Adaptive Cluster Approximation (ACA)
ACA is a technique used to evaluate local or semi-local reduced density matrices by creating a smaller effective system. It converges very quickly, meaning it achieves high accuracy with only a small number of required cluster levels, significantly reducing the computational cost and qubit count.
Hybrid Quantum-Classical Algorithm
This method combines quantum computation and classical computation. The quantum computer handles preparing states and measuring specific properties (like density matrix elements), while the classical computer performs the complex constrained minimization needed to find the best parameters for those measurements.

Terminology

Summary

A novel parallel hybrid quantum-classical algorithm for solving the quantum-chemical ground-state energy problem on gate-based quantum computers has been presented, offering a path to treat much larger molecules than traditional Variational Quantum Eigensolvers (VQE) by leveraging reduced density-matrix functional theory (RDMFT) and the adaptive cluster approximation (ACA). This approach is significant because it introduces a new level of parallelization, allowing for the computational treatment of large systems on NISQ devices while retaining noise tolerance.

The Core Theoretical Framework

The approach is based on the reduced density-matrix functional theory (RDMFT) formulation of the electronic structure problem, which decomposes the density-matrix functional of the full system into an indirectly coupled sum of density-matrix functionals for all its subsystems using the adaptive cluster approximation to RDMFT. This decomposition drastically reduces qubit count and makes the problem inherently parallelizable. The solutions for these effective subsystems involve a constrained minimization over many-particle states approximated by parametrized trial states on the quantum computer, similar to VQE.

The Hybrid Quantum-Classical Algorithm

The proposed algorithm is a hybrid quantum-classical approach designed to compute the reduced density matrix functional (RDMF) on quantum hardware. The process involves:

  1. Preparing a parametrized ansatz state, denoted as Ψ(u)i, on the quantum computer, where the parameters are defined by a vector of real parameters "u."

  2. Evaluating relevant observables on the quantum computer, specifically the elements of the one-particle reduced density matrix ρ(1)β,α(u) and elements of the two-particle reduced density matrix ρ(2), i.e., ρ(2)α,β,γ,δ(u).

  3. Performing a constrained minimization to solve for the parameters "u," which is done on a classical computer. This minimization seeks to minimize an upper bound to the exact constrained minimum of the RDMF in Eq. (6).

Qubit Reduction Techniques

The paper details several techniques aimed at reducing qubit count and program depth:

  1. Employing symmetries, as noted for VQE, but also introducing an additional degree of freedom by means of an unitary transformation of the non-interacting states optimized in every step to minimize the augmented Lagrangian. This makes a given hardware-efficient trial state more flexible.

  2. Using the local approximation approach where the interaction Hamiltonian is decomposed into local terms and non-local terms, allowing for evaluation via a sum of local RDMFs: F Wˆ[ρ(1)] ≈ X i F Wˆ local,i [ρ(1)] + F Wˆ non-local [ρ(1)].

  3. Utilizing the Adaptive Cluster Approximation (ACA) to evaluate a local or semi-local RDMF by creating a smaller effective system. The ACA converges quickly, with the additive error converging like O(e-n), implying that the required ACA level 'n' is in O(log(-1/epsilon)).

Measurement Program Optimization

The evaluation of the RDMF requires measuring individual elements of the one-particle reduced density matrix, which can be done efficiently. The number of measurements scales linearly with system size when using ACA: O(Nχlocal i=1 (n + 1) squared Nint,i!, where Nlocal is the number of local RDMFs. Furthermore, the paper shows that by exploiting the general commutativity of observables, all elements can be measured with only O(Nχ) quantum programs instead of O(N 2χ).

Simulation and Noise Handling on NISQ Devices

The algorithm's performance is demonstrated using the Hubbard model as a test case. The results are shown for:

  1. Noise-free quantum simulations, where the augmented Lagrangian converges rapidly, typically in no more than 10 outer iterations.

  2. Genuine simulations on NISQ hardware (e.g., IBM quantum computers), which include noise models like depolarizing gate errors and thermal relaxation errors. The convergence behavior during these noisy executions is compared to noiseless simulations, suggesting that the observed behavior points to a representability issue of the many-particle state on the noisy quantum computer.

Conclusion and Outlook

The work concludes that the evaluation of RDMFs is possible on a quantum computer, and this formulation allows for novel approximations like ACA to drastically reduce the qubit requirements of the problem. The approach is particularly suitable for ab-initio molecular dynamics calculations because forces can be evaluated straightforwardly from available quantities like the one- and two-particle reduced density matrix. Future research will focus on "the investigation of the representability of fermionic quantum states on noisy gate-based quantum computers, the optimization of the measurement programs as well as the integration with molecular dynamics programs like CP2K.

Improvements for AI systems

As a fastidious and diligent AI researcher, I have analyzed this paper, Parallel Quantum Chemistry on Noisy Intermediate-Scale Quantum Computers, focusing on how its proposed hybrid quantum-classical algorithm based on Reduced Density-Matrix Functional Theory (RDMFT) can be leveraged to improve AI systems.

The core contribution is a method to evaluate the quantum mechanical ground-state energy problem (electronic structure) using gate-based quantum computers, specifically designed to be parallelizable and noise-tolerant for Near-Term Intermediate Scale Quantum (NISQ) devices.

Here are the specific improvements that can be made to AI systems, categorized by application:

  1. Improvement in Materials Science and Drug Discovery (Quantum Chemistry Simulation)

The primary improvement is the ability to perform highly accurate, correlated electronic structure calculations for molecular systems that are intractable for classical methods due to exponential scaling.

Specific Improvements:

  1. Accurate Ground-State Energy Prediction: The system can calculate the ground-state energy of complex molecules (like those in the Hubbard model used in the paper) with high fidelity, overcoming the limitations of traditional Density Functional Theory (DFT) when dealing with strong electronic correlations.

  2. Ab Initio Molecular Dynamics (AIMD): The algorithm allows for the calculation of analytical nuclear forces by taking derivatives of the energy functional with respect to atomic positions. This enables a quantum-enhanced AIMD simulation pipeline, predicting not just static properties but also time-dependent behavior (e.g., reaction pathways, phase transitions) with higher accuracy than classical force fields.

  3. Modeling Correlated Systems: By utilizing the RDMFT framework and the Adaptive Cluster Approximation (ACA), the system can accurately model strongly correlated materials, which are crucial for understanding phenomena like metal-insulator transitions in condensed matter physics and designing novel high-temperature superconductors or catalysts.

What the Improved System Can Do:

This improved system can act as a Quantum Chemistry Surrogate Model. It can rapidly screen vast chemical spaces to identify stable molecular structures, predict reaction barriers, and determine the electronic properties (like band gaps or magnetic moments) of novel materials with high confidence, accelerating the discovery process for new pharmaceuticals and battery materials.

  1. Improvement in Quantum Computing Algorithm Design (Hybrid Algorithm Development)

The paper presents a novel parallel hybrid quantum-classical algorithm tailored for NISQ hardware constraints.

  1. Improvement in Quantum Simulation and Error Mitigation (NISQ Error Handling)

The paper explicitly addresses the challenge of noise on NISQ devices (IBM quantum computers).

Sources

Related papers