⚛️ quantum physics

GCAMPS: A Scalable Classical Simulator for Qudit Systems

This paper introduces GCAMPS, a scalable classical simulator that generalizes the Clifford Augmented Matrix Product State (CAMPS) method to qudit systems, demonstrating significant performance improvements over conventional tensor network approaches, particularly for qutrit simulations.

Original authors: Ben Harper, Azar C. Nakhl, Thomas Quella, Martin Sevior, Muhammad Usman

Published 2026-01-28

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Ben Harper, Azar C. Nakhl, Thomas Quella, Martin Sevior, Muhammad Usman

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 trying to simulate a complex quantum system on a regular computer. Think of the computer's memory as a backpack.

The Problem: The Backpack is Too Small
Standard ways of simulating quantum systems are like trying to pack a mountain of rocks into a tiny backpack. As you add more "quantum particles" (called qudits) to your simulation, the amount of information needed to describe them explodes. For a standard computer, this is like trying to carry a whole mountain in a backpack that can only hold a few pebbles. Eventually, the backpack rips, and the simulation crashes. This is especially true for systems that have more than two states (like a light switch that is just on/off). These are called qudits (think of them as multi-colored switches instead of just black and white).

The Old Solution: The "Magic" Shortcut
Scientists have developed a clever trick called the Stabilizer Method. Imagine this as a special map that only works for systems that are "simple" or "predictable" (called Clifford circuits). If your quantum system is simple, this map is tiny and fits easily in your backpack. However, if your system gets complicated (adding "magic" or non-Clifford gates), the map becomes useless, and you have to go back to the heavy mountain of rocks.

The New Solution: GCAMPS (The Hybrid Backpack)
The authors of this paper introduced a new method called GCAMPS (Generalised Clifford Augmented Matrix Product State). Think of this as a hybrid backpack that combines two strategies:

The Map (Stabilizer): It keeps the "simple" parts of the system on a tiny, efficient map.
The Rocks (Tensor Network): It keeps the "complicated" parts as a compressed stack of rocks (a Matrix Product State).

The genius of GCAMPS is that it constantly tries to turn the "complicated" rocks back into "simple" map instructions. When a complicated operation happens, the system breaks it down, pushes the complicated bits onto the rock stack, and then immediately tries to find a "magic key" (a specific Clifford operation) to turn those rocks back into a simple map again. This keeps the backpack light.

The Big Discovery: It Works Even Better for "Multi-Color" Switches
The authors took this hybrid backpack and upgraded it to handle qudits (systems with 3 or more states, like a qutrit which has three states: 0, 1, and 2).

The Challenge: Simulating these 3-state systems is much harder for old methods because the "rocks" get huge and heavy very quickly.
The Result: When they tested GCAMPS on these 3-state systems, it didn't just work; it performed better than it did on standard 2-state systems.

Why?
Imagine trying to carry a pile of heavy bricks (the standard method).

For 2-state systems, the bricks are small. The hybrid backpack helps, but the improvement is okay.
For 3-state systems, the bricks are massive boulders. The old method fails almost immediately. However, the GCAMPS hybrid backpack is so good at turning those massive boulders back into a tiny map that the improvement is huge. It saves so much more memory and time for the 3-state systems than it does for the 2-state ones.

The Bottom Line
The paper claims that GCAMPS allows scientists to simulate complex quantum systems with 3 states (qutrits) on regular computers much more efficiently than before. It proves that this "hybrid backpack" strategy works for these more complex systems, opening the door to studying complex physics (like specific types of magnetic chains) that were previously impossible to simulate without a real quantum computer.

What they did NOT claim:

They did not claim this solves medical problems or clinical issues.
They did not claim this builds a working quantum computer.
They did not claim it works for all possible quantum systems (specifically, finding the "magic keys" to compress the data gets harder as the system gets very large, so there are still limits).

Technical Summary: GCAMPS: A Scalable Classical Simulator for Qudit Systems

Problem Statement
Classical simulation of quantum systems is a computationally intractable problem that scales poorly with system size. Conventional methods face distinct limitations: state vector and density matrix simulations scale exponentially with the number of qudits ( $O(d^n)$ ), while tensor network methods (like Matrix Product States, MPS) scale with entanglement, and stabilizer methods scale with "magic" (non-Clifford operations). While hybrid approaches like Clifford Augmented Matrix Product State (CAMPS) have successfully combined stabilizer and tensor network techniques to simulate qubit systems ( $d=2$ ) with high efficiency, these methods have not been extended to higher-dimensional quantum systems (qudits, $d>2$ ). Furthermore, extending stabilizer simulators to qudits is non-trivial because universal extended stabilizer simulators typically scale exponentially with non-Clifford operations regardless of the state's actual complexity.

Methodology
The authors introduce GCAMPS (Generalised Clifford Augmented Matrix Product State), a hybrid classical simulation framework that generalizes the CAMPS method to arbitrary quantum degrees of freedom ( $d$ ). The method represents the quantum state as $|\psi\rangle = C|\text{MPS}\rangle$ , where $C$ is a Clifford operation represented efficiently by a stabilizer tableau, and $|\text{MPS}\rangle$ is a Matrix Product State that ideally remains close to a product state.

The simulation workflow involves three primary components:

Qudit Stabilizer Formalism: The authors define the generalized Pauli group and Clifford group for $d$ -level systems. They establish update rules for the stabilizer tableau under generalized Clifford gates (including $H_d$ , $S_d$ , and the entangling $SUM_d$ gate). The tableau entries are integers modulo $d$ , and phases are tracked as elements of the multiplicative group generated by $\omega = \exp(2\pi i/d)$ .
Operator Decomposition and Commutation: When a non-Clifford gate (e.g., a $T$ gate) is applied, it is decomposed into a sum of Pauli strings. These strings are then commuted through the Clifford operator $C$ using the stabilizer tableau. This process transforms the local non-Clifford operation into a potentially non-local operator acting directly on the MPS.
Entanglement Reduction (Disentanglers): To prevent the MPS bond dimension from exploding, the framework employs a heuristic search for "disentanglers"—specific Clifford operations ( $Q$ ) that reduce the entanglement in the MPS after a non-Clifford operation. The Clifford $C$ is then updated as $\tilde{C} = CQ^\dagger$ . For low dimensions ( $d=2, 3$ ), the authors perform an exhaustive search over unique two-qudit entangling Cliffords to find optimal disentanglers.

Key Contributions

Generalization to Qudits: The paper provides the first concerted effort to extend the CAMPS hybrid simulation method to support arbitrary quantum degrees of freedom, specifically demonstrating the extension to qutrits ( $d=3$ ).
Qudit Stabilizer Implementation: It details the mathematical framework for qudit stabilizer simulation, including the definition of generalized Pauli operators, the structure of the stabilizer tableau over $\mathbb{Z}_d$ , and the specific update rules for generalized Clifford gates.
Disentangler Search: The authors outline a procedure to generate and canonicalize two-qudit Clifford tableaus to identify unique disentanglers, enabling the optimization of the MPS bond dimension in the GCAMPS framework.

Results
The authors benchmarked GCAMPS against conventional MPS simulations using random $T$ -doped Clifford circuits for both qubits ( $d=2$ ) and qutrits ( $d=3$ ).

Bond Dimension Scaling: For qutrit systems, GCAMPS maintains a constant bond dimension regime for a significant portion of the circuit depth, resulting in linear memory scaling with system size. This behavior mirrors the scaling observed in qubit simulations.
Performance Improvement: In the regime where the number of $T$ gates is less than the number of qudits, GCAMPS demonstrates significant improvements in both runtime and memory usage compared to conventional MPS.
Qutrit vs. Qubit Efficiency: Notably, the performance gain of GCAMPS over conventional MPS is greater for qutrits than for qubits. This is attributed to the fact that conventional MPS simulations for qutrits require decomposing significantly larger matrices ( $3^{N/2}$ vs $2^{N/2}$ ) and saturate bond dimensions much faster. In contrast, GCAMPS keeps the effective bond dimensions low (approx. 3 for qutrits vs. 2 for qubits), yielding a larger relative speedup.
Limitations: The exhaustive search for disentanglers scales exponentially with $d$ , currently restricting the practical application of this specific search method to low quantum degrees of freedom ( $d \leq 3$ ).

Significance
The paper claims that GCAMPS enables the classical simulation of quantum problems that were previously intractable without access to quantum hardware, specifically for systems with higher-dimensional state spaces. By successfully generalizing hybrid simulation techniques to qudits, the work opens avenues for studying complex many-body physics, such as spin-one systems (e.g., bilinear-biquadratic spin chains) and topological phase transitions beyond spin-1/2 systems. The results suggest that hybrid stabilizer-tensor network methods are a viable and highly efficient path for simulating high-dimensional quantum systems, particularly where the "magic" content of the circuit is moderate.

More like this