TensorMixedStates: a Julia library for simulating pure and mixed quantum states using matrix product states
Pith reviewed 2026-05-22 14:27 UTC · model grok-4.3
The pith
A Julia library enables simulation of mixed quantum states and Lindblad evolution using matrix product states.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
TensorMixedStates implements the matrix product state representation for mixed states along with operations for their time evolution according to a Lindblad equation or discrete non-unitary gates.
What carries the argument
Matrix product state representation of mixed states with associated Lindblad and quantum-channel evolution routines.
If this is right
- Researchers can write simulations of dissipative quantum systems in a few lines of code.
- State-of-the-art algorithms such as DMRG and TDVP become available for mixed-state problems.
- Both pure and mixed states can be handled within the same framework and interface.
Where Pith is reading between the lines
- The library may make studies of decoherence in larger many-body systems more routine.
- It could serve as a base for adding support for spatially varying dissipation or measurement-based dynamics.
- Benchmarks on standard problems would clarify the largest system sizes reachable in practice.
Load-bearing premise
The ITensor library supplies reliable low-level tensor manipulation and state-of-the-art algorithms sufficient for efficient mixed-state simulations.
What would settle it
Running a small dissipative spin-chain simulation and comparing its observables or steady state against exact diagonalization results.
read the original abstract
We introduce TensorMixedStates, a Julia library built on top of ITensor which allows the simulation of quantum systems in the presence of dissipation using matrix product states (MPS). It offers three key features: i) it implements the MPS representation for mixed states along with associated operations, in particular the time evolution according to a Lindblad equation or discrete time evolution using non-unitary gates (quantum channels), ii) it is based on ITensor, which has proven its effectiveness and which gives access to efficient low-level tensor manipulation as well as state-of-the-art algorithms (like DMRG or TDVP), finally iii) it presents a user-friendly interface allowing users to write sophisticated simulations for pure and mixed quantum states in a few lines of code.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript introduces TensorMixedStates, a Julia library built on ITensor for simulating pure and mixed quantum states with matrix product states. It claims three main contributions: (i) MPS representations for mixed states together with operations including time evolution under Lindblad master equations or discrete quantum channels, (ii) reuse of ITensor’s low-level tensor routines and algorithms such as DMRG and TDVP, and (iii) a concise user-facing API that lets researchers write simulations in a few lines of code.
Significance. If the implementation is shown to be numerically stable and accurate, the library would provide a practical, open-source extension of the ITensor ecosystem to open quantum systems, lowering the barrier for mixed-state tensor-network studies of dissipative many-body dynamics. The absence of any benchmarks, convergence tests, or validation against exact results, however, prevents a quantitative assessment of its performance or reliability.
major comments (2)
- [Abstract and Feature Overview] The manuscript provides no numerical validation whatsoever for the central claim that the library correctly implements Lindblad evolution or quantum-channel dynamics. No small-system comparisons to exact diagonalization, no error metrics (e.g., trace distance or negativity), and no convergence plots with respect to bond dimension or time step appear in any section. This omission is load-bearing because the reliability of the mixed-state MPS truncation and the preservation of trace and positivity are not demonstrated.
- [Implementation] Section describing the mixed-state representation (purification versus superoperator mapping) does not specify how truncation is performed or how the resulting operators are projected back onto the physical subspace. Without this detail it is impossible to judge whether the claimed time-evolution routines remain stable or positivity-preserving under the bond-dimension cuts that are inevitable in practical MPS simulations.
minor comments (2)
- [Abstract] The abstract states that the library “presents a user-friendly interface,” yet the manuscript contains no concrete code examples or usage snippets that would allow a reader to reproduce even a minimal Lindblad simulation.
- [Introduction] Citation to the underlying ITensor library is appropriate, but additional references to standard mixed-state MPS literature (e.g., works on purification or vectorized Liouvillian MPS) would help situate the new package within the existing literature.
Simulated Author's Rebuttal
We thank the referee for their constructive and detailed comments. We agree that the original manuscript would benefit from explicit numerical validation and expanded implementation details, and we have revised the manuscript to address these points directly.
read point-by-point responses
-
Referee: [Abstract and Feature Overview] The manuscript provides no numerical validation whatsoever for the central claim that the library correctly implements Lindblad evolution or quantum-channel dynamics. No small-system comparisons to exact diagonalization, no error metrics (e.g., trace distance or negativity), and no convergence plots with respect to bond dimension or time step appear in any section. This omission is load-bearing because the reliability of the mixed-state MPS truncation and the preservation of trace and positivity are not demonstrated.
Authors: We acknowledge the validity of this observation. The initial submission prioritized the description of the library's API and integration with ITensor over extensive benchmarking. In the revised manuscript we have added a new section containing numerical validation: small-system Lindblad evolutions are compared against exact diagonalization, with quantitative error measures including trace distance and negativity; convergence with respect to bond dimension and time-step size is shown explicitly. These additions demonstrate that trace is preserved to machine precision and that positivity is maintained within controllable truncation error. revision: yes
-
Referee: [Implementation] Section describing the mixed-state representation (purification versus superoperator mapping) does not specify how truncation is performed or how the resulting operators are projected back onto the physical subspace. Without this detail it is impossible to judge whether the claimed time-evolution routines remain stable or positivity-preserving under the bond-dimension cuts that are inevitable in practical MPS simulations.
Authors: We thank the referee for highlighting this omission. The original text was intentionally concise but insufficiently explicit. The revised section now details the truncation procedure: for the purification representation we perform SVD truncation on the purified MPS and renormalize to restore unit trace; for the superoperator (vectorized) representation we apply standard MPS truncation to the operator tensor network followed by a projection step that enforces the trace-preserving condition on the resulting channel. We also discuss the approximate preservation of positivity under truncation and provide practical guidance on bond-dimension selection to keep unphysical eigenvalues below a user-specified tolerance. revision: yes
Circularity Check
No circularity: library implementation paper with no derivations
full rationale
This manuscript is a software library description introducing TensorMixedStates on top of ITensor for MPS-based mixed-state simulations, including Lindblad evolution and quantum channels. It contains no mathematical derivations, equations, predictions, fitted parameters, or uniqueness theorems. The three key features listed in the abstract are straightforward implementation claims that rely on the external ITensor library's established algorithms (DMRG, TDVP) without any reduction of outputs to inputs by construction or self-citation chains. No load-bearing steps exist that could be circular; the paper is self-contained as a tool announcement.
Axiom & Free-Parameter Ledger
Forward citations
Cited by 1 Pith paper
-
Decoherence in Waveguide Quantum Electrodynamics using Matrix Product States
A generalization of matrix product state techniques to density matrices enables efficient inclusion of Lindblad decoherence terms in waveguide QED simulations.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.