## Can Negative Dissipation Rates Make NISQ Hardware More Useful?

Researchers at the University of Luxembourg have published a quantum simulation method that deliberately incorporates negative dissipation rates into the Gorini–Kossakowski–Sudarshan–Lindblad (GKSL) master equation — a move that is counterintuitive by design. Rather than treating environmental noise as a problem to suppress, Kasturi Ranjan Swain and Adolfo del Campo use Lyapunov-based feedback to engineer that noise into a control resource, steering quantum systems toward target ground states with improved convergence. The paper was accepted on June 27, 2026, and published August 3, 2026 in *npj Quantum Information*.

The core result is that negative dissipation rates — normally excluded from standard Lindblad treatments because they sit outside the Markovian regime — can be implemented within a quantum algorithm designed to calculate ground-state properties. The method extends the simulation framework into genuinely non-Markovian territory: dynamics where the system's future evolution depends on its history, not just its current state. That extension comes with a cost the authors explicitly acknowledge: increased exponential sampling overhead compared to conventional approaches.

For [NISQ](https://quantumintel.tech/glossary/nisq) hardware operators and algorithm designers, the trade-off framing matters more than the headline result. This is not noise cancellation; it is noise redirection.

---

## What the GKSL Extension Actually Does

The Lindblad master equation is the workhorse of open quantum system theory. In its standard Markovian form, all dissipation rates are non-negative — a mathematical requirement that enforces complete positivity and ensures the density matrix remains physical at every time step. The moment you allow negative rates, you step outside that guarantee and into the more complex domain of non-Markovian dynamics, where memory effects in the environment feed back into the system's evolution.

The Luxembourg team's contribution is a concrete algorithmic framework for doing this on a quantum computer in a controlled way. Lyapunov-based feedback — borrowed from classical control theory — provides the stabilizing mechanism. By designing the feedback to drive the system toward a specific target state, the researchers show that even when the effective dissipation rates go negative, the dynamics can still converge. The convergence improvement they report is the key claim that distinguishes this from merely demonstrating that negative rates are simulable.

The physical interpretation is significant. Real-world quantum systems rarely satisfy the Markovian approximation cleanly. Superconducting transmon qubits coupled to lossy resonators, trapped-ion chains with phonon-bath interactions, and solid-state spin systems all exhibit non-Markovian signatures at timescales relevant to computation. A simulation framework that can faithfully represent those dynamics — rather than projecting them onto a Markovian approximation — has direct utility for benchmarking hardware and developing more accurate noise models.

The [decoherence](https://quantumintel.tech/glossary/decoherence) problem in NISQ devices is, at its root, an open-systems problem. Methods that engage with that reality rather than paper over it represent a more honest path toward useful near-term computation.

---

## The Sampling Overhead Problem Is Not Trivial

The authors are transparent about the cost: the method incurs increased exponential sampling overhead. In quantum simulation, sampling overhead is not an engineering inconvenience — it can determine whether a method is practically viable on any near-term device, regardless of how elegant the underlying theory is.

Exponential overhead means that as system size grows, the number of circuit shots required to estimate expectation values to a fixed precision scales badly. This is the same fundamental challenge that limits many quantum error mitigation techniques, including probabilistic error cancellation, which also introduces sampling costs that grow with circuit depth and error rate.

The authors argue that for certain applications, the improved convergence and access to non-Markovian physics justify the additional overhead. That is a defensible position for small-scale proof-of-concept studies. Whether it survives contact with systems of industrially relevant size is an open question the paper does not resolve — and that honest gap is worth flagging for any enterprise buyer or algorithm developer evaluating this approach.

The research team also credited Peter Zoller for "numerous insightful suggestions," and acknowledged Kazutaka Takahashi, Pablo Martínez-Azcona, and András Grabarits for useful discussions during development. Zoller's involvement as a sounding board signals engagement from one of the field's central figures in open-system quantum optics and simulation.

---

## Industry Trajectory: Noise as a Feature, Not a Bug

The broader framing of this work — harnessing noise rather than fighting it — sits within a growing cluster of approaches that challenge the assumption that pre-[fault-tolerant quantum computing](https://quantumintel.tech/glossary/fault-tolerant-quantum-computing) hardware is too dirty to be useful. Dissipative quantum computing, environment-assisted quantum transport, and noise-assisted optimization have all attracted serious theoretical attention over the past several years.

What makes the Luxembourg result worth watching is the combination of three elements: a theoretically principled extension of the Lindblad framework, a concrete feedback-control implementation strategy, and a publication venue (*npj Quantum Information*) with rigorous peer review. That combination moves it out of the speculative category.

For the hardware companies building [NISQ](https://quantumintel.tech/glossary/nisq) devices today — and for the algorithm teams at firms like [Quantinuum](https://quantumintel.tech/companies/quantinuum), [IonQ](https://quantumintel.tech/companies/ionq), and [Rigetti Computing](https://quantumintel.tech/companies/rigetti-computing) who must extract useful computation from imperfect qubits — methods that reframe noise as a controllable resource rather than an obstacle deserve attention. The sampling overhead problem is real and must be solved before any of this scales, but the theoretical foundation laid here is sound.

---

## Key Takeaways

- **Kasturi Ranjan Swain and Adolfo del Campo** at the University of Luxembourg developed a quantum algorithm incorporating negative dissipation rates in the GKSL master equation, published in *npj Quantum Information* on August 3, 2026.
- The method uses **Lyapunov-based feedback control** to engineer environmental noise into a steering mechanism, achieving improved convergence toward target ground states.
- It extends quantum simulation into **non-Markovian regimes** — dynamics with environmental memory effects — that the standard Lindblad framework cannot capture.
- The improvement comes with a documented cost: **increased exponential sampling overhead**, which limits near-term scalability.
- The approach is relevant to any NISQ platform where decoherence and non-Markovian bath coupling degrade simulation fidelity — which is effectively all current hardware.
- Peter Zoller provided input during manuscript development, lending additional credibility to the theoretical foundations.

---

## Frequently Asked Questions

**What is the GKSL master equation and why does it matter for quantum computing?**
The Gorini–Kossakowski–Sudarshan–Lindblad (GKSL) master equation describes how an open quantum system evolves when it interacts with an environment. It is the standard theoretical tool for modeling decoherence and dissipation in qubits. Most quantum hardware operates as an open system — qubits are never perfectly isolated — so the GKSL equation is the relevant framework for understanding and simulating realistic noise.

**What are negative dissipation rates and why are they unusual?**
In the standard Markovian Lindblad formalism, dissipation rates must be non-negative to ensure the quantum state (density matrix) remains physically valid at all times. Negative rates arise in non-Markovian dynamics, where memory effects in the environment violate the Markovian approximation. They are counterintuitive because they can temporarily represent information flowing back from the environment into the system, rather than being lost irreversibly.

**What is Lyapunov-based feedback in this context?**
Lyapunov-based feedback is a control strategy borrowed from classical control theory. A Lyapunov function acts as an energy-like quantity that the feedback is designed to minimize, driving the system toward a stable target state. In the Luxembourg team's framework, this feedback stabilizes the quantum dynamics even when negative dissipation rates would otherwise destabilize standard Lindblad evolution.

**What is the practical limitation of this method?**
The method incurs exponential sampling overhead — meaning the number of measurements required grows exponentially with system size or simulation complexity. This makes it computationally expensive and limits near-term applicability to small or moderate system sizes. The authors acknowledge this trade-off explicitly.

**How does this relate to quantum error mitigation strategies on current hardware?**
Rather than correcting errors after the fact (as in quantum error correction) or statistically canceling them (as in probabilistic error cancellation), this approach attempts to redirect noise into a productive control mechanism. It is complementary to, not a replacement for, standard error mitigation techniques, and could in principle be combined with them for certain open-system simulation tasks.