Dissipative quantum algorithms for excited-state quantum chemistry

This paper introduces a versatile dissipative quantum algorithm that transforms the preparation of electronic excited states into an effective ground-state problem by engineering Lindblad dynamics to make the target state the unique steady state, demonstrating its efficacy through numerical simulations of complex atomic and molecular systems.

The Gist

Imagine you are trying to find a specific, rare room in a massive, dark, and confusing hotel. In the world of quantum chemistry, this "room" is an excited state—a specific, high-energy arrangement of electrons in an atom or molecule. These states are crucial for understanding things like how plants capture sunlight or how certain chemical reactions happen, but finding them on a quantum computer is notoriously difficult.

Usually, to find this room, you need a perfect map (a "good initial guess") to start your search. But often, we don't have a good map. If you start in the wrong place, you might get stuck in a dead end or wander aimlessly.

This paper introduces a new, clever strategy called Dissipative Quantum Algorithms. Instead of trying to walk carefully toward the target room, this method uses a "quantum vacuum cleaner" to suck everything else out of the hotel, leaving only the room you want.

Here is how it works, broken down into simple concepts:

1. The Core Idea: The "Quantum Vacuum"

In physics, "dissipation" usually means losing energy (like a ball rolling down a hill and stopping). The authors flip this idea on its head. They design a special "environment" (a set of rules for the quantum computer) that acts like a one-way street.

• The Analogy: Imagine a hotel where every room has a door that only opens downward. If you are in a room higher up, you can slide down to a lower room. But if you are in the lowest room, you can't go anywhere else; you get stuck there.
• The Trick: The researchers modify the rules of the hotel so that the target excited state (the rare room you want) becomes the "lowest" room in a specific section. Once the system starts moving, it naturally slides down until it gets stuck in that target room. No matter where you start, you eventually end up there.

2. Three Different Ways to Set the Rules

The paper proposes three different "blueprints" for building this one-way street, depending on what information you already have about the target room:

• Strategy A: The "Symmetry" Filter (The VIP Section)

  • The Metaphor: Imagine the hotel has different wings. Some wings are for people with red hats, others for blue hats. If you know your target room is in the "Red Hat Wing," you simply lock the doors to all other wings.
  • How it works: If the excited state has a different "spin" or particle count than the ground state, the algorithm restricts the search to that specific group. The system then just finds the lowest room within that group, which happens to be your target.

• Strategy B: The "Folded Spectrum" (The U-Turn)

  • The Metaphor: Imagine you have a map where the target room is actually on the 10th floor, but you want it to feel like the ground floor. You take the map, fold it in half at the 10th floor, and flip the top half upside down. Now, the 10th floor is the bottom of the new map.
  • How it works: If you know the approximate energy of the target, the algorithm mathematically "folds" the energy levels around that point. The target excited state becomes the new "ground state" (the bottom), and the vacuum cleaner naturally pulls the system down to it.

• Strategy C: The "Spectral Projector" (The Bouncer)

  • The Metaphor: Imagine a bouncer at the door of the hotel who says, "No one below the 5th floor is allowed to enter."
  • How it works: Instead of folding the map (which is computationally expensive), this method acts as a filter. It blocks any path that leads to rooms with energy lower than a certain point. The system is forced to slide down only until it hits that "floor," where it gets stuck. This is often cheaper to run on a computer than the "folded" method.

3. Testing the Vacuum

The authors tested this "quantum vacuum" on several digital simulations:

• Simple Molecules: They successfully found excited states in hydrogen molecules (H2 and H4).
• Atoms: They found specific energy states in atoms like Carbon and Oxygen.
• Complex Molecules: They tackled Benzene (a ring of carbon atoms) and Ferrocene (a sandwich-like molecule with iron). These are tricky because the electrons are highly "entangled" (they move in complex, coordinated ways).

The Results:
In every case, the method successfully "cooled" the system down to the correct excited state. It was accurate enough to predict energy levels with "chemical accuracy" (the gold standard for chemistry). It also proved to be very robust, meaning it didn't break down even when the starting point was messy or when the system was stretched out (like pulling a molecule apart).

4. Why This Matters

Traditional methods often get stuck if you don't have a perfect starting guess. This new approach is like a self-correcting vacuum: it doesn't care where you start; it just keeps pulling until you are in the right place. It avoids the need for complex, error-prone tuning that other quantum algorithms require.

In summary: The paper presents a new way to use quantum computers to find specific, high-energy chemical states by engineering a "one-way" flow that naturally funnels the system into the desired state, regardless of where it starts. It's a flexible, robust tool for simulating complex chemistry.

Technical Summary

Technical Summary: Dissipative Quantum Algorithms for Excited-State Quantum Chemistry

Problem Statement
Accurate and efficient preparation of electronic excited states on quantum devices remains a significant challenge in quantum chemistry. While Quantum Phase Estimation (QPE) is a natural candidate for preparing energy eigenstates, it relies on the availability of high-quality initial states with sufficient overlap with the target eigenstate, which is often difficult to construct for chemically relevant excited states. Furthermore, from a computational complexity perspective, preparing the ground state of a local Hamiltonian is QMA-hard in the worst case, implying that excited-state preparation is at least as difficult. Existing alternatives, such as Variational Quantum Algorithms (VQA), Adiabatic State Preparation (ASP), and Quantum Imaginary-Time Evolution (QITE), face limitations including variational parameter optimization, the need for full state tomography, normalization factors that scale poorly with system size, and susceptibility to mean-field initialization artifacts (e.g., degeneracies in Hartree-Fock references) and noise accumulation.

Methodology
The authors introduce a general dissipative algorithm framework that recasts the problem of preparing a specific excited state as an effective ground-state problem. The core mechanism utilizes engineered Lindblad dynamics, where the system is coupled to a designed environment (quantum channel) such that the target excited state becomes the unique steady state of the evolution. The dynamics are governed by the Lindblad equation:
$$\frac{d\rho}{dt} = -i[H, \rho] + \sum_k \left( K_k \rho K_k^\dagger - \frac{1}{2} \{K_k^\dagger K_k, \rho\} \right)$$
where $H$ is the system Hamiltonian and $\{K_k\}$ are jump operators constructed from coupling operators and a filter function. The jump operators are designed to annihilate the target state ($K_k |\psi_{\text{target}}\rangle = 0$), ensuring it is a fixed point, while inducing transitions from other states toward the target.

To address the challenge of targeting specific excited states, the paper proposes three complementary strategies tailored to different types of prior information:

1. Symmetry-Based Approach: If the target excited state resides in a distinct symmetry sector (e.g., defined by particle numbers $N_\alpha, N_\beta$ or spin) from the ground state, the coupling operators are restricted to preserve this symmetry. This confines the dissipative evolution to the specific sector, effectively reducing the problem to finding the ground state within that subspace. States in other sectors become "dark" and are never populated.
2. Folded-Spectrum Approach: When an approximate energy $\mu$ of the target state is available, the Hamiltonian is transformed via $H \mapsto (H - \mu I)^2$. This transformation "folds" the spectrum around $\mu$, making the target excited state the effective ground state of the modified operator. While effective, this approach squares the Hamiltonian, significantly increasing the spectral radius and inverse gap, thereby raising the cost of Hamiltonian simulation.
3. Spectral Projector Approach: To avoid the high cost of squaring the Hamiltonian, this method constructs an operator $P_\mu(H)$ that projects out eigenstates with energies below a reference $\mu$. This projector is applied to both the jump operators and the initial state. This restricts the dissipative dynamics to the subspace above $\mu$, allowing the system to converge to the lowest-energy state in that subspace (the target excited state) without simulating the squared Hamiltonian.

The coupling operators are primarily chosen from a set of Hermitian one-body operators ($S_{II}$ or a reduced nearest-neighbor version $S'_{II}$) to ensure locality. The authors also address ergodicity issues where the system might get trapped in "dark states" due to disconnected configuration spaces by augmenting the coupling set with higher-order operators (e.g., quartic terms) when necessary.

Key Results
The framework was validated through numerical simulations using classical Monte Carlo trajectory methods to simulate the unraveled Lindblad dynamics across various systems:

• Hydrogen Systems ($H_2$ and $H_4$): All three protocols successfully prepared triplet excited states with high accuracy. The symmetry-based method was efficient for states in different symmetry sectors, while the folded-spectrum and spectral projector methods handled states in the same sector as the ground state. Resource estimation confirmed that the folded-spectrum method is significantly more expensive due to the Hamiltonian squaring, whereas the spectral projector method offers a favorable trade-off.
• Atomic Spectra (Li to O): The folded-spectrum method achieved excited-state energy estimates with errors well below the chemical accuracy threshold relative to Full Configuration Interaction (FCI) references. The method often outperformed Equation-of-Motion Coupled Cluster (EOM-CCSD/UCCSD) methods in accuracy. A specific challenge with the Carbon atom's $^5S$ state demonstrated the necessity of augmenting the coupling operator set to restore ergodicity and avoid trapping in spurious dark states.
• Small Molecules (BH, CH+, H2O, HCl): The dissipative method achieved chemical accuracy for systems with both single-reference and strong multi-reference character (e.g., stretched geometries of BH and CH+ where EOM-CCSD failed). The method demonstrated robust convergence of overlap, energy, and spin multiplicity.
• Complex Systems (Benzene and Ferrocene): Using active-space models (6e, 6o for benzene; 10e, 7o for ferrocene), the method successfully prepared $\pi$-$\pi^*$ and $d$-$d$ transition states. The results showed agreement with experimental data and CASCI references, with discrepancies attributed to the limitations of the active-space model (lack of dynamical correlation) rather than the dissipative algorithm itself.

Significance and Claims
The paper claims to present the first dissipative dynamics-based method specifically designed for excited-state quantum chemistry. Its primary significance lies in:

1. Independence from High-Quality Initial States: The method can drive the system to the target excited state from arbitrary initial states, even those with zero overlap, provided the dynamics are ergodic.
2. Robustness: Unlike Adiabatic State Preparation, the dissipative approach is not limited by artifacts arising from mean-field initialization (such as Coulson-Fischer points or degeneracies) and is intrinsically more robust to certain types of noise (e.g., depolarizing noise) due to the contractive nature of the dynamics.
3. Versatility: The three proposed strategies allow the framework to adapt to available prior information (symmetry, approximate energy), making it applicable to a wide range of excited states, including those with strong multi-reference character.
4. Scalability: The authors suggest that while strict ergodicity to a single eigenstate in large systems may be challenging, the dissipative protocol can serve as a robust cooling mechanism to confine the system to a low-energy manifold, which can then be refined using phase-estimation-type techniques.

The work concludes that dissipative state preparation offers a flexible and potentially scalable primitive for the quantum simulation of excited-state chemistry, particularly for strongly correlated electronic systems where traditional methods struggle.