Moments-based quantum computation of the electric dipole moment of molecular systems

This paper demonstrates that the quantum computed moments (QCM) method, utilizing a Lanczos cluster expansion on an IBM Quantum device, can accurately estimate the electric dipole moment of a water molecule with superior noise robustness and higher precision compared to standard VQE approaches.

The Gist

The Big Picture: Measuring the "Charge" of a Water Molecule

Imagine you are trying to measure the electric personality of a water molecule. Specifically, you want to know its electric dipole moment. Think of this as measuring how much the molecule acts like a tiny magnet with a positive end and a negative end. This is a crucial property for understanding how water interacts with everything else.

Scientists are trying to use quantum computers (machines that use the weird rules of quantum physics to solve problems) to calculate this. However, current quantum computers are like "noisy" calculators; they make mistakes easily, especially when doing complex math.

Most experiments have focused on using these noisy machines to find the energy of a molecule (how stable it is). But this paper asks: Can we use these same noisy machines to measure other things, like the dipole moment, accurately?

The Problem: The "Noisy" Measurement

The standard way to measure a property on a quantum computer is to run a specific program (a circuit) and ask the computer, "What is the average value of this property?"

The authors found that if you just ask the computer this directly, the "noise" (static) in the machine makes the answer wrong. It's like trying to hear a whisper in a hurricane; the signal gets lost. In their tests, the direct method gave an error of about 5%.

The Solution: The "Moment" Recipe

The authors used a clever trick called Quantum Computed Moments (QCM).

The Analogy: The Bouncing Ball
Imagine you drop a ball into a dark room and you want to know exactly where it will stop (the ground state).

1. The Direct Method: You just look at the ball once. If the room is foggy (noisy), you might guess the wrong spot.
2. The Moments Method: Instead of just looking once, you bounce the ball off the walls several times and listen to the echoes (the "moments"). Even if the room is foggy, the pattern of the echoes contains hidden information that lets you calculate exactly where the ball should be, filtering out the fog.

In the paper, they use a mathematical framework (Lanczos cluster expansion) to take these "echoes" (mathematical moments of the energy) and combine them to get a much cleaner, more accurate answer. They had previously used this to fix energy calculations, but this paper is the first time they applied it to the dipole moment.

The Secret Sauce: The "Tweak" Trick

To measure the dipole moment, they couldn't just ask the computer directly. They had to use a mathematical rule called the Hellmann-Feynman theorem.

The Analogy: The Slope of a Hill
Imagine the energy of the molecule is a hill. The dipole moment is the slope of that hill at the very bottom.

• To find the slope, you can't just stand at the bottom and look; you need to see how the height changes if you take a tiny step to the left and a tiny step to the right.
• The authors "tweaked" the math of the molecule slightly (adding a small imaginary force, $\lambda$) to create two slightly different versions of the hill.
• They calculated the energy of these two tweaked versions using their "Moments" recipe.
• By comparing the difference between the two, they could calculate the slope (the dipole moment) without ever needing to measure the dipole directly on the noisy machine.

Why this is clever: Because they used the same noisy quantum measurements for both the "left step" and the "right step," the random noise canceled out. It's like weighing yourself on a broken scale that adds 5 pounds randomly. If you weigh yourself, then weigh yourself again immediately after, the error is the same both times. If you subtract the two numbers, the error disappears, leaving you with the true difference.

The Results: A Clearer Picture

When they tested this on a real IBM quantum computer (a superconducting device):

• Direct Method (The "Whisper"): The result was off by about 5%.
• Moments Method (The "Echoes"): The result was off by only 2% (specifically, within 0.03 debye of the perfect theoretical answer).

Even more impressively, this 2% error was achieved even though the direct method was run on a perfect, noise-free computer simulation and still had a 5% error. This proves that the "Moments" technique is not just fixing noise; it's actually a smarter way to extract the answer from the data.

The Takeaway

The paper demonstrates that you don't need a perfect, error-free quantum computer to measure complex chemical properties. By using a "Moments-based" recipe that listens to the echoes of the system's energy, scientists can get accurate results for things like electric dipole moments, even on today's noisy machines. It turns a noisy, blurry picture into a sharp, clear one.

Technical Summary

Technical Summary: Moments-based Quantum Computation of the Electric Dipole Moment

Problem Statement
While quantum computers hold promise for simulating quantum chemical systems, practical demonstrations on near-term devices have largely been restricted to estimating ground-state electronic energies, primarily using the Variational Quantum Eigensolver (VQE). A significant gap exists in the accurate, noise-robust computation of other ground-state properties, such as electric dipole moments. Direct expectation value determination (standard VQE output) for these non-energetic observables is often susceptible to noise and systematic errors, particularly when the trial state is not the exact ground state. Furthermore, existing methods for estimating molecular response properties often rely on excited states or external fields, whereas this work focuses on evaluating non-energetic ground-state properties in the absence of external potentials.

Methodology
The authors employ the Quantum Computed Moments (QCM) method, based on the Lanczos cluster expansion, to estimate the electric dipole moment of the water molecule ($H_2O$). The core approach involves adapting the Hellmann-Feynman theorem to moments-based energy estimation.

1. Hamiltonian Perturbation: The standard Hamiltonian $H$ is modified by adding a perturbation term $\lambda \mu$, where $\mu$ is the electric dipole moment operator and $\lambda$ is a scalar parameter ($H_\lambda = H + \lambda \mu$).
2. Hellmann-Feynman Theorem: The dipole moment is related to the derivative of the ground-state energy $E_\lambda$ with respect to $\lambda$ at $\lambda=0$: $\mu_0 = \frac{\partial E_\lambda}{\partial \lambda}|_{\lambda=0}$.
3. Finite Difference Approximation: The derivative is approximated using a finite difference scheme: $\mu_0 \approx \frac{1}{2\Delta}[E_L(\Delta) - E_L(-\Delta)]$, where $E_L$ is the moments-corrected energy estimate.
4. Moments-Based Correction: Instead of using raw VQE energies, the authors calculate Hamiltonian moments ($\langle H^p \rangle$) from the Reduced Density Matrix (RDM) of a trial state (Unitary Coupled Cluster Doubles, UCCD). These moments are used in an analytic formula derived from the Lanczos recursion to produce a corrected energy estimate $E_L$.
5. Noise Mitigation Strategy:
   • Reference-State Calibration: A depolarizing noise model is assumed. The noise level is estimated by comparing the noisy moments of the trial state against the classically exact moments of a Hartree-Fock reference state. This calibration is applied to the modified moments $\langle H_\lambda^p \rangle$.
   • Post-Mitigated vs. Pre-Mitigated: The paper evaluates different strategies for applying this calibration. "Post-mitigated moments" (Method B) apply error mitigation to the combined modified moments $\langle H_\lambda^p \rangle$. "Pre-mitigated moments" (Method C) apply mitigation to the individual operator coefficients before constructing the modified moments. The authors find Method B to be more stable against noise, as Method C suffers from error propagation in higher-order terms (specifically quadratic in the dipole operator).
   • Finite Difference Advantage: Crucially, the parameter $\lambda$ is introduced in post-processing. This allows the same quantum measurements (RDM) to be reused for both $\lambda = +\Delta$ and $\lambda = -\Delta$, significantly reducing stochastic sampling noise compared to running separate circuits.

Key Contributions

• Extension of QCM to Non-Energetic Observables: The paper demonstrates that moments-based energy correction techniques, previously validated for ground-state energies, can be successfully adapted to estimate non-energetic properties like the electric dipole moment via the Hellmann-Feynman theorem.
• Experimental Verification on Hardware: The method is implemented on an IBM superconducting quantum device (ibm_torino). The study utilizes an 8-qubit system representing the active space of the water molecule (STO-3G basis, frozen core).
• Noise Robustness Analysis: The work provides a detailed analysis of how different error mitigation strategies (pre- vs. post-mitigation) affect the stability of the results, identifying that applying mitigation to the full modified moment operator yields superior stability compared to mitigating individual coefficients.
• Efficiency: The approach avoids increasing quantum circuit depth, as the perturbation is handled classically during post-processing, while still incorporating wavefunction response information that standard expectation value calculations miss.

Results

• Accuracy: The noise-mitigated moments-based estimate for the water molecule's dipole moment agrees with Full Configuration Interaction (FCI) calculations to within 0.03 ± 0.007 debye (2% ± 0.5%).
• Comparison to Direct Estimation: In contrast, direct expectation value determination (standard VQE) yields errors on the order of 0.07 debye (5%), even when performed without noise.
• Variance: The variance of the moments-corrected estimate is notably smaller than that of the direct expectation value, demonstrating that the noise-robustness property of the energy correction transfers to non-energetic properties.
• Convergence: The results converge to the statevector simulation results for small values of the finite difference step size $\Delta$. The study notes that while the percentage error for the dipole moment is higher than for energy (due to the sensitivity of off-diagonal RDM elements to noise), the relative improvement over direct estimation is significant.

Significance and Claims
The paper claims that moments-based techniques offer a viable path to improving the accuracy of non-energetic ground-state property estimation on noisy intermediate-scale quantum (NISQ) devices. By leveraging the Hellmann-Feynman theorem and reusing quantum data for finite differencing, the method circumvents the limitations of direct expectation value evaluation. The authors assert that this approach broadens the potential range of chemical applications where quantum devices could provide an advantage over classical computation, specifically by transferring the inherent accuracy and noise-robustness of corrected energy values to other observables. The work serves as an experimental verification that the QCM method produces superior estimates for such quantities compared to direct evaluation in chemical systems.