Feasibility of performing quantum chemistry calculations on quantum computers

The authors propose two criteria demonstrating that achieving practical quantum advantage in chemistry calculations is unlikely with current technologies, as VQE algorithms require fault-tolerant hardware due to decoherence sensitivity while QPE algorithms suffer from exponentially suppressed success probabilities due to the orthogonality catastrophe.

The Gist

The Big Question: Can Quantum Computers Solve Chemistry?

Imagine you have a brand-new, super-powered race car (the Quantum Computer). Everyone is telling you that this car is going to revolutionize how we design new medicines and fertilizers by simulating how molecules work (this is Quantum Chemistry).

The authors of this paper are like the mechanics checking the engine. They asked: "Is this car actually ready to win the race, or is it just a fancy toy?"

They looked at the two main ways people try to use these cars for chemistry and found some serious potholes on the road.

1. The "Noisy" Approach (VQE)

The Idea: This method tries to use the quantum computers we have right now. These machines are fast but "noisy," meaning they make mistakes easily, like a radio with a lot of static.

The Problem:
Imagine you are trying to tune a radio to hear a very quiet whisper (the chemical energy you want to measure). But you are standing next to a jet engine (the hardware noise).

• In chemistry, the "whisper" is tiny.
• The "jet engine" noise is huge.

The paper shows that the noise in the computer creates a massive amount of "fake energy." Even if you try to fix the static with software tricks (Error Mitigation), the jet engine is just too loud. To hear the whisper, you would need a computer that is perfect (noise-free). But if you have a perfect computer, you wouldn't need this specific "noisy" method in the first place.

The Verdict: Using current noisy computers for chemistry is like trying to weigh a feather on a scale that is vibrating on a truck. The result will be wrong.

2. The "Perfect" Approach (QPE)

The Idea: This method waits for the future, when we have perfect, error-free quantum computers. This is the "gold standard" algorithm.

The Problem:
To use this method, you have to give the computer a starting guess. Imagine you are looking for a specific lost hiker in a massive forest (the molecule).

• If you start searching near where the hiker was last seen, you find them quickly.
• If you start searching on the other side of the country, you will never find them.

The paper found that as molecules get bigger, the "forest" gets exponentially larger. Your starting guess becomes less and less likely to be anywhere near the real answer. This is called the "Orthogonality Catastrophe."

The Verdict: Even with a perfect computer, if you don't know where to start looking, you won't find the answer. As the molecule gets bigger, the chance of guessing right drops to almost zero.

3. The Competition (Classical Computers)

The Idea: We shouldn't forget about regular computers (like the laptop you are using now).

The Comparison:
The authors compared the Quantum approach to a classical method called Variational Monte Carlo (VMC).

• Quantum (VQE): Like a race car that breaks down often and needs a perfect track to work.
• Classical (VMC): Like a very reliable bicycle. It doesn't have the "noise" problem, and it's getting faster every year thanks to better software.

The Verdict: For finding the lowest energy of a molecule, the "bicycle" (Classical Computer) might actually beat the "race car" (Quantum Computer) for a long time to come.

The Conclusion: Don't Give Up, Just Change Tracks

The authors aren't saying quantum computers are useless. They are saying that chemistry might not be the best place to start.

A Better Target:
Instead of trying to find the "deepest valley" (the lowest energy state), maybe we should use quantum computers to watch the "river flow" (how particles move and change over time).

• Finding the valley is hard because classical computers are already good at it.
• Watching the river is hard for classical computers because the water gets too turbulent to track.

Summary:
The paper suggests that while quantum computers are exciting, they might not be the magic wand for chemistry problems just yet. The hardware is too noisy, and the algorithms require guesses that are too hard to make. We might need to find a different "race track" where quantum computers can actually show off their speed.

Technical Summary

Here is a detailed technical summary of the paper "Feasibility of performing quantum chemistry calculations on quantum computers" by Louvet, Ayral, and Waintal.

1. Problem Statement

Quantum chemistry is widely envisioned as a primary application for quantum computers, specifically for finding the ground state energy of molecules. However, extrapolating current hardware capabilities to determine if a genuine "quantum advantage" is achievable remains difficult. The paper addresses the feasibility gap by deriving necessary conditions (criteria) for two leading quantum algorithms:

1. Variational Quantum Eigensolver (VQE): Designed for Noisy Intermediate-Scale Quantum (NISQ) devices.
2. Quantum Phase Estimation (QPE): Designed for future fault-tolerant quantum computers.

The authors argue that existing classical numerical techniques (e.g., Coupled Cluster, DMRG) are already highly efficient, setting a high bar for quantum methods. The core problem is determining whether quantum hardware can meet the precision and resource requirements to outperform these classical methods, particularly for strongly correlated systems.

2. Methodology

The authors employ a theoretical and analytical approach to derive quantitative bounds on hardware performance and algorithmic success, rather than relying solely on numerical simulations.

• VQE Analysis (Noise Criterion):

  • They model the effect of hardware noise on the variational energy estimation.
  • They define a fidelity $F$ between the ideal state and the noisy density matrix $\rho$.
  • They derive a criterion for the maximum tolerable gate error ($\epsilon_{max}$) as a function of the target chemical precision ($\eta_{chem}$), the number of gates ($N_g$), and the energy difference between the noise-induced state and the variational state ($E_{noise} - E_V$).
  • Key Physical Insight: They analyze the relationship between the Target Hamiltonian ($H$, the molecule) and the Hardware Hamiltonian ($H_{hw}$, the quantum processor). They argue that because $H_{hw}$ is generic and uncorrelated with $H$, hardware noise (e.g., relaxation) populates high-energy eigenstates of the molecule that are not physically screened. This leads to a noise-induced energy shift that scales quadratically with the number of electrons ($N^2$).

• QPE Analysis (Overlap Criterion):

  • They analyze the success probability of QPE, which depends on the overlap ($\Omega$) between the input state and the true ground state.
  • They propose a method to estimate this overlap using the Overlap Index ($I_\Omega$), derived from the energy and energy variance of the input state without needing the exact ground state.
  • They investigate the "Orthogonality Catastrophe," where the overlap between two states decreases exponentially with system size.

• Classical Comparison:

  • They compare VQE against the classical Variational Monte Carlo (VMC) algorithm, specifically focusing on neural-network-based wavefunctions.

3. Key Contributions

1. VQE Noise Tolerance Criterion: The derivation of Equation (8), which sets an upper bound on hardware error rates. It highlights that achieving chemical accuracy requires gate errors far below current physical limits because the noise-induced energy bias scales unfavorably ($N^2$) with system size.
2. Hardware-Molecule Decoupling: The identification that gate-based quantum computers lack physical constraints (symmetries) shared with the target molecule. Consequently, hardware noise creates macroscopic electric dipoles in the molecular energy spectrum, leading to large energy errors ($\sim 10$ Hartrees) compared to the target precision ($\sim 1.6$ mHa).
3. QPE Overlap Index: A practical criterion ($I_\Omega \approx (E_\Phi - E_0)^2 / 2\sigma^2_\Phi$) to estimate the success probability of QPE based on the energy variance of the ansatz. This quantifies the Orthogonality Catastrophe in variational state preparation.
4. VQE vs. VMC Benchmark: A detailed comparison showing that VMC may remain superior to VQE even on perfect quantum hardware due to VMC's lack of decoherence, variance reduction capabilities, and efficient gradient calculation.

4. Key Results

• VQE Feasibility:

  • For a benchmark molecule like Benzene (30 electrons, 108 orbitals), achieving chemical accuracy requires a gate error rate of $\epsilon \le 10^{-12}$.
  • Current state-of-the-art hardware operates at $\epsilon \approx 10^{-3}$ (two-qubit gates).
  • Error mitigation techniques (e.g., probabilistic error cancellation) can reduce bias but increase the variance of the estimator exponentially, requiring an exponential number of measurement shots ($N_s$) to compensate.
  • The noise-induced energy error scales as $E_{noise} \approx aN + bN^2$, making larger basis sets (needed for accuracy) significantly more susceptible to noise.

• QPE Feasibility:

  • The overlap index $I_\Omega$ scales linearly with system size ($N$), implying the success probability $\Omega \approx e^{-\alpha N}$ decays exponentially.
  • This "Orthogonality Catastrophe" means that for large systems, the probability of preparing a state with sufficient overlap for QPE becomes vanishingly small, requiring exponentially many repetitions.

• VQE vs. VMC:

  • VMC avoids decoherence entirely.
  • VMC benefits from variance reduction: as the ansatz approaches the ground state, the variance of the local energy vanishes, allowing for infinite precision with finite samples. VQE does not possess this property.
  • Gradient calculation in VMC is $O(1)$ (via backpropagation), whereas in VQE it is $O(N_\theta)$ (number of parameters).

5. Significance and Conclusion

• Reality Check: The paper provides a skeptical but rigorous assessment of the current trajectory of quantum chemistry on quantum computers. It suggests that the "early disruptive application" narrative may be overly optimistic given the hardware constraints.
• Hardware Requirements: It establishes that relevant chemistry calculations likely require fault-tolerant quantum computers, not just NISQ devices with error mitigation. The required error rates are orders of magnitude beyond current capabilities.
• Strategic Shift: The authors suggest that ground state energy estimation might not be the optimal target for quantum advantage due to the high quality of classical methods and the specific difficulties of quantum hardware (decoherence, orthogonality).
• Alternative Path: They propose that quantum dynamics (time evolution) might be a more suitable "soft target" for quantum computers, as classical algorithms struggle with sign problems and entanglement growth in dynamical simulations, whereas quantum hardware naturally handles unitary evolution. However, this still requires error correction to be viable.

In summary, the paper argues that while quantum chemistry remains a theoretical goal, the specific constraints of noise scaling and state preparation overlap create formidable barriers that current and near-future hardware cannot overcome without significant algorithmic or hardware breakthroughs.