Scalable Quantum Tomography and Simulation with Matrix Product States

This direction is split into other directions: matrix product state representation. Its last version stays readable and citable here.

Version 1 of 180 papers93 conceptsfrozen 2026-10-04

The fundamental challenge of quantum mechanics is the "exponential wall." As you add qubits to a system, the size of the Hilbert space—the mathematical "room" needed to describe the state—doubles with every particle. To fully describe just 50 qubits, you would need more parameters than there are atoms in the Earth.

However, nature is rarely "randomly" entangled. Most physical states of interest, such as the ground states of molecules or the outputs of quantum circuits, occupy a tiny, highly structured corner of this vast space. Researchers like J. Ignacio Cirac and Mari Carmen Bañuls have pioneered a way to exploit this structure using Matrix Product States (MPS) and Matrix Product Operators (MPO). By decomposing global quantum information into a chain of local tensors, we can simulate and verify quantum systems that were previously thought to be computationally unreachable.

The Intuition: Compression via the Area Law

The reason MPS works is rooted in the Area Law of Entanglement. In most physical systems, especially in one dimension, the entanglement between a subregion and the rest of the system doesn't scale with the volume of that region, but only with its boundary (the "area").

Imagine a chain of atoms. Instead of trying to write down one giant coefficient for every possible configuration of the whole chain, we represent the state as a sequence of small tensors—one for each atom—linked together by "virtual bonds." These bonds act as the wires through which entanglement flows. If the entanglement is low (as the area law suggests), these bonds can be kept small. This "bond dimension" ($\chi$) is the compression knob: by keeping it finite, we turn an exponential problem into a linear one.

As shown in Large-Scale Quantum Dynamics with Matrix Product States by Alberto Baiardi, this allows us to project complex quantum equations onto a manageable mathematical manifold, enabling the simulation of molecular systems with dozens of degrees of freedom that would otherwise crash a supercomputer.

Bypassing the Exponential Wall in Tomography

One of the most critical applications of this idea is Quantum State Tomography (QST)—the process of "photographing" a quantum state to verify a quantum computer's performance. Traditionally, QST is impossible for large systems because it requires $3^N$ different measurement settings.

Recent work by Zhen Qin, Casey Jameson, and Zhihui Zhu has demonstrated that if we assume the target state has an MPO structure, we can reconstruct it using only $O(N)$ measurements. In Sample-Optimal Quantum State Tomography for Structured Quantum States in One Dimension, they prove that we can bypass the exponential wall entirely for 1D systems.

This works because the MPO constraint acts as a powerful "prior" in a machine learning sense. By treating the local tensors as trainable parameters and using algorithms like Projected Gradient Descent, researchers can find the "best fit" state that matches the observed data without ever needing to store the full, exponentially large density matrix.

From Physics to Machine Learning

The power of the MPS representation extends beyond pure physics into generative modeling and optimization. Because an MPS can efficiently represent high-dimensional joint probability distributions, it can be used to build "Quantum-inspired" neural networks.

For example, Wanda Hou introduced a Tensor Network Born Machine that uses MPS to model RNA sequences, outperforming standard models by optimizing how data is embedded into the quantum Hilbert space. Similarly, Weitao Du and colleagues have used MPS to model high-order many-body interactions in molecules, reducing the computational complexity of multi-atom relationships from exponential to linear.

Simulating the Unsimulatable

Perhaps the most striking evidence for the effectiveness of this direction is its ability to solve the "sign problem" in gauge theories. In Non-Abelian string breaking phenomena with matrix product states, researchers used MPS to observe the real-time dynamics of "string breaking"—where the vacuum snaps to create new particles.

While traditional Monte Carlo methods fail due to mathematical oscillations (the sign problem), the Hamiltonian-based MPS approach thrives. By managing entanglement growth during time evolution, it provides a window into the fundamental forces of nature, like the interactions between quarks, using the same tensor-chain logic that Emanuele Dalla Torre and Ehud Altman used to identify hidden topological orders in Bose insulators.

Whether it is verifying a 512-qubit optimization problem on a single GPU—as demonstrated by Taylor L. Patti and Animashree Anandkumar—or mapping the electronic landscape of nitrogenase, the core insight remains the same: by focusing on the local structure of entanglement, we can compress the infinite complexity of the quantum world into a chain of manageable matrices.

Papers in this version

Showing 60 of 80 papers.

History & citationVersion 1 · 2026-10-04

Split into other directions · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_scalable_quantum_tomography_and_simulati_9c4bce2c,
  title = {Scalable Quantum Tomography and Simulation with Matrix Product States},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/scalable-quantum-tomography-and-simulation-with-matrix-product-states/rdr_9c4bce2cf5fd459d825de8afe1237b53},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 1, frozen text):

@misc{lacuna_scalable_quantum_tomography_and_simulati_9c4bce2c_v1,
  title = {Scalable Quantum Tomography and Simulation with Matrix Product States},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/scalable-quantum-tomography-and-simulation-with-matrix-product-states/rdr_9c4bce2cf5fd459d825de8afe1237b53/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}