⚛️ quantum physics

Empirical Evaluation of QAOA with Zero Noise Extrapolation on NISQ Hardware for Carbon Credit Portfolio Optimization in the Brazilian Cerrado

This study demonstrates that the Quantum Approximate Optimization Algorithm (QAOA) combined with Zero Noise Extrapolation (ZNE) outperforms classical heuristics in optimizing complex carbon credit portfolios for the Brazilian Cerrado, establishing empirical quantum utility on NISQ-era hardware for environmental conservation planning.

Original authors: Hugo José Ribeiro

Published 2026-02-11

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hugo José Ribeiro

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a landscape architect trying to save a massive, diverse ecosystem called the Cerrado in Brazil. You have a limited budget and a list of 88 different towns (municipalities) where you could invest in conservation. Your goal is to pick exactly 28 of these towns to protect.

But here's the catch: You can't just pick the 28 towns that look the "greenest" on a map. You have to balance three things at once:

Carbon: How much pollution can the trees absorb?
Biodiversity: How well do these towns connect to help animals move and survive?
People: How will this help the local communities?

This is a giant puzzle. If you try to solve it by just picking the "best" town one by one (a method called a Greedy Heuristic), you might miss the big picture. You might pick two great towns that are far apart, leaving a gap where animals can't cross, or you might miss a town that isn't the absolute best on its own but is the perfect "glue" that connects two other great areas.

The Quantum Solution: A New Way to Look at the Puzzle

The author of this paper, Hugo José Ribeiro, tried using a Quantum Computer to solve this puzzle. Think of a classical computer (like your laptop) as a very fast, very smart person reading a map and checking one path at a time. A quantum computer, however, is like having a magical ability to look at all possible paths at the same time.

The specific tool they used is called QAOA (Quantum Approximate Optimization Algorithm). It's like a digital explorer that jumps around the landscape of possibilities, looking for the perfect combination of 28 towns.

The Problem: The Quantum Computer is "Noisy"

Here's the tricky part: The quantum computers available today (called NISQ devices) are like a radio with a lot of static. They are powerful, but they make mistakes because of "noise" (interference). If you ask the quantum computer to solve the puzzle, the static often scrambles the answer, making it worse than what a simple human could do with a basic checklist.

The Fix: Zero Noise Extrapolation (ZNE)

To fix the static, the researcher used a clever trick called Zero Noise Extrapolation (ZNE).

The Analogy: Imagine you are trying to guess the exact temperature of a room, but your thermometer is broken and reads slightly too high.

You take a reading with the broken thermometer (Normal Noise).
You then deliberately make the thermometer worse by shaking it or heating it up, and take a second reading (Amplified Noise).
You do it again, making it even worse (Maximum Noise).

Now, you have three data points: "Normal," "Worse," and "Worst." By drawing a line through these points and extending it backward to where the shaking would be zero, you can mathematically guess what the temperature would have been if the thermometer were perfect.

In this paper, the researcher did this with the quantum computer. They ran the same puzzle three times: once normally, once with "extra noise" added, and once with "double extra noise." Then, they used math to "extrapolate" back to what the answer would be if the computer had zero noise.

The Results: Did it Work?

The paper reports some very exciting results from running this experiment on real IBM quantum computers over a 17-day period:

The Classical Baseline: The standard "Greedy" method (picking the best towns one by one) got a score of 44.42.
The Raw Quantum Attempt: Without fixing the noise, the quantum computer scored about 43.55. It was actually slightly worse than the simple method because of the static.
The Quantum + ZNE Result: After using the "Zero Noise Extrapolation" trick to clean up the answer, the quantum computer scored 58.47.

The Takeaway: The quantum method, after cleaning up the noise, was 31.6% better than the standard classical method.

Why Does This Matter?

The most interesting part isn't just the higher score; it's how the quantum computer found the solution.

The "Greedy" method picked towns based on their individual scores. But the quantum method found a town called Chapadão do Céu.

The Greedy Method ignored it because its individual score wasn't the highest.
The Quantum Method picked it because, even though it wasn't the "best" on its own, it was the perfect connector. It had amazing biodiversity links to its neighbors.

The quantum computer saw the "synergy" (the teamwork between towns) that the simple method missed. It found a better portfolio by looking at the whole picture rather than just the individual pieces.

The Bottom Line

This paper doesn't claim that quantum computers are ready to replace all human planners tomorrow. The author is careful to say this is "Empirical Quantum Utility"—meaning, for this specific, real-world problem, the quantum approach worked better than the standard tools they tested.

It proves that even with today's "noisy" quantum computers, if you use the right tricks (like ZNE) to clean up the signal, you can find better solutions for complex environmental problems than traditional methods can. It's a small but significant step toward using quantum magic to help save the planet's most diverse ecosystems.

Technical Summary: Empirical Evaluation of QAOA with Zero Noise Extrapolation on NISQ Hardware for Carbon Credit Portfolio Optimization

Problem Definition
This study addresses the multi-objective optimization of carbon credit portfolios within the Brazilian Cerrado biome, a region characterized by high biodiversity but significant agricultural encroachment. The core challenge involves selecting a subset of municipalities to maximize carbon sequestration, biodiversity connectivity, and social impact simultaneously, subject to a fixed budget (cardinality constraint). The problem is formulated as a Quadratic Unconstrained Binary Optimization (QUBO) problem with $n=88$ decision variables (municipalities) and a target selection of $k=28$ . The search space, approximately $1.45 \times 10^{22}$ combinations, renders exhaustive enumeration intractable. Classical approaches, such as greedy heuristics, often fail to capture complex spatial synergies and nonlinear interactions between territories, potentially leading to suboptimal conservation outcomes or "greenwashing" through inefficient allocations.

Methodology
The authors implement a hybrid quantum-classical workflow utilizing the Quantum Approximate Optimization Algorithm (QAOA) combined with Zero Noise Extrapolation (ZNE) on Noisy Intermediate-Scale Quantum (NISQ) hardware.

Problem Formulation: The objective function integrates linear terms (individual scores for carbon, biodiversity, and social metrics) and quadratic terms (spatial adjacency and synergy matrices). A cardinality constraint ( $k=28$ ) is enforced via a quadratic penalty term in the QUBO formulation.
Quantum Algorithm: The QAOA ansatz employs a single layer ( $p=1$ ) to remain within the coherence limits of current hardware. To strictly respect the cardinality constraint during evolution, an XY-mixer Hamiltonian is used instead of the standard transverse-field mixer, ensuring transitions only occur between states with the same number of active qubits.
Initialization: A "warm-start" strategy initializes the quantum state using the solution from a classical greedy heuristic, with parameters $\gamma=0.05$ and $\beta=0.20$ empirically calibrated to preserve the greedy structure while allowing local exploration.
Error Mitigation: Zero Noise Extrapolation (ZNE) is applied to mitigate gate noise. Circuits are executed at three noise scale factors ( $\lambda \in \{1, 2, 3\}$ ) via gate folding (specifically on two-qubit gates). The results are extrapolated to the zero-noise limit ( $\lambda=0$ ) using linear, quadratic, and Richardson extrapolation models.
Experimental Setup: Seven independent runs were conducted over a 17-day period (January 17 – February 2, 2026) on two IBM Quantum Heron processors: ibm_torino (r1) and ibm_fez (r2). Each run involved 8,192 shots per noise scale, totaling 24,576 shots per ZNE protocol.

Key Contributions

Real-World Application: This work presents one of the first applications of QAOA to a real-world, multi-objective environmental optimization problem using empirical geospatial data from 88 municipalities in the Cerrado, moving beyond simulated or synthetic benchmarks.
Statistical Rigor: The study provides a statistically robust validation through seven independent hardware executions, demonstrating temporal stability against hardware calibration drifts.
Error Mitigation in Sustainability: It is the first application of Zero Noise Extrapolation to multi-objective environmental sustainability problems, demonstrating that error mitigation can recover solution quality superior to classical heuristics on current NISQ devices.
Reproducibility: The complete workflow, including QUBO construction, parameter calibration, and raw experimental data, is made publicly available, establishing a reproducible template for high-precision environmental conservation policies.

Results

Performance: The QAOA+ZNE workflow consistently outperformed a classical greedy baseline (score 44.42). The mean portfolio score achieved by the quantum method was 58.47 ± 6.98, representing a 31.6% improvement over the greedy baseline.
Statistical Significance: The improvement was statistically significant with a one-sided paired t-test yielding $p = 0.0009$ and a large effect size (Cohen's $d = 2.01$ ). The method achieved a 100% success rate across all seven runs, meaning every ZNE-extrapolated score exceeded the greedy baseline.
Robustness: Even the most conservative linear extrapolation model yielded a mean score of 49.12, still exceeding the baseline by 10.6%. A leave-one-out analysis confirmed that the significance was not driven by a single high-performing run.
Solution Quality: The quantum approach identified non-trivial territorial synergies. For instance, the municipality of Chapadão do Céu, which was excluded by the greedy heuristic due to a moderate individual score, was consistently selected by the quantum method (6/7 runs) due to its exceptional biodiversity connectivity (98th percentile).
Hardware Stability: The methodology remained stable across a 13-day interval and two different hardware backends, with no statistically significant correlation found between execution time and ZNE scores, indicating resilience to calibration drift.

Significance and Claims
The paper explicitly frames its findings as establishing empirical quantum utility rather than formal quantum advantage. The authors do not claim that their method surpasses optimal classical solvers (e.g., Gurobi or CPLEX) or achieves a superpolynomial speedup. Instead, the significance lies in demonstrating that current NISQ devices, when coupled with rigorous error mitigation (ZNE) and appropriate initialization (warm-start), can consistently identify complex territorial synergies that myopic classical heuristics overlook.

The study validates that quantum optimization can serve as a practical decision-support tool for environmental policy, offering a scalable methodological template for allocating resources in high-stakes conservation contexts. The results suggest that the NISQ era is capable of delivering tangible improvements over widely used operational heuristics in data-driven territorial planning, provided that error mitigation and statistical validation are rigorously applied.