TEPID-ADAPT: Adaptive variational method for simultaneous preparation of low-temperature Gibbs and low-lying eigenstates
Bharath Sambasivam, Kyle Sherbert, Karunya Shirali, Nicholas J. Mayhall, Edwin Barnes, Sophia E. Economou · 2025
Read TEPID-ADAPT: Adaptive variational method for simultaneous preparation of low-temperature Gibbs and low-lying eigenstates by Bharath Sambasivam, Kyle Sherbert, Karunya Shirali, Nicholas J. Mayhall, Edwin Barnes, Sophia E. Economou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-14490.txt.
Introduction
Bharath Sambasivam and colleagues present a quantum algorithm for preparing thermal states in this two thousand twenty-five paper. A Gibbs state describes a system settled at a fixed temperature. Preparing such states on a quantum computer is key for studying matter, above all at low temperatures. Their method, TEPID-ADAPT, grows a trial circuit step by step. It also recovers the system's lowest energy states along the way.
A mixed-state trial form lets them compute entropy with no extra cost. The method can run with a small number of extra qubits, or with none at all. It works from a trimmed list of the system's energy levels. The paper serves quantum computing researchers who simulate physics. Better state preparation widens what near-term machines can study.
Catalog record
| CREATOR | Bharath Sambasivam, Kyle Sherbert, Karunya Shirali, Nicholas J. Mayhall, Edwin Barnes, Sophia E. Economou |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-14490
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
TEPID-ADAPT: Adaptive variational method for simultaneous preparation of low-temperature Gibbs and low-lying eigenstates Bharath Sambasivam, Kyle Sherbert, Karunya Shirali, Nicholas J. Mayhall, Edwin Barnes, Sophia E. Economou Abstract Preparing Gibbs states, which describe systems in equilibrium at finite temperature, is of great importance, particularly at low temperatures. In this work, we propose a new method– TEPID-ADAPT– that prepares the thermal Gibbs state of a given Hamiltonian at low temperatures using a variational method that is partially adaptive and uses a minimal number of ancillary qubits. We achieve this using a truncated, parametrized eigenspectrum of the Hamiltonian. Beyond preparing Gibbs states, this approach also straightforwardly gives us access to the truncated low-energy eigenspectrum of the Hamiltonian, making it also a method that prepares excited states simultaneously. As a result of this, we are also able to prepare thermal states at any lower temperature of the same Hamiltonian without further optimization. I Introduction The problem of state preparation is a crucial subroutine in several quantum algorithms, both near-term and fault-tolerant. Therefore, finding efficient ways to prepare quantum states is of great interest. The majority of currently available techniques focus on the preparation of pure states of closed systems, most of which tackle ground-state preparation. In a more realistic setting, quantum systems inevitably have a coupling to an environment. This makes the effective state of the system of interest mixed, described by a density matrix. As a result, addressing the problem of density matrix preparation is an important milestone in the simulation of real physical systems using quantum computers. Of particular interest are thermal Gibbs states, which describe the state of an open quantum system at thermal equilibrium with a bath held at finite temperature \rho_{G}\equiv\frac{e^{-\beta\,{H}}}{Z},\qquad Z=\Tr(e^{-\beta\,{H}}), (1) where H is the Hamiltonian of the system of interest, and \beta is the inverse temperature. The Gibbs state uniquely minimizes the free energy F=\langle{H}\rangle-T\,S, (2) where \langle{\,\bm{\cdot}\,}\rangle\equiv\Tr(\rho\,\boldsymbol{\cdot}) , and S=-\Tr(\rho\log\rho) is the von Neumann entropy. From the perspective of state preparation, the low-temperature regime is more interesting and challenging Watrous 2008 ; Aharonov et al. 2013 ; Bakshi et al. 2024 . In this regime, only the low-energy eigenstates of the Hamiltonian contribute significantly to the Gibbs state. Preparing excited states, particularly the low-lying eigenstates that make up the low-temperature Gibbs states of a Hamiltonian is a closely related task of great importance for applications in both physics and chemistry. It has applications including understanding dynamics of systems, measuring transition and decay rates Ibe et al. 2022 ; Ciavarella 2020 , and mass gaps of quantum field theories Chandani et al. 2024 . Existing methods include quantum phase estimation (QPE) Kitaev 1995 , variational quantum algorithms Higgott et al. 2019 ; Nakanishi et al. 2019 ; Wen et al. 2021 ; Xie et al. 2022 ; Sherbert and Nardelli 2022 ; Gocho et al. 2023 ; Xu et al. 2023 ; Grimsley and Evangelista 2025 ; Chandani et al. 2024 , and subspace methods McClean et al. 2017 ; Motta et al. 2019 ; Parrish and McMahon 2019 ; Stair et al. 2020 ; Asthana et al. 2023 ; Cianci et al. 2024 . Ref. Xu et al. 2023 uses a purification scheme to prepare an equal mixture of states to target low-lying states simultaneously. Preparing Gibbs states is important for applications such as quantum simulation Childs et al. 2018 , quantum chaos García-Mata et al. 2023 , combinatorial optimization problems Kirkpatrick et al. 1983 ; Somma et al. 2008 , and quantum machine learning Kieferová and Wiebe 2017 ; Biamonte et al. 2017 . Existing methods to study Gibbs states on quantum computers include Lindblad simulation and sampling Poulin and Wocjan 2009 ; Temme et al. 2011 ; Chowdhury and Somma 2017 ; Chen et al. 2023a ; Chen et al. 2023b ; Eassa et al. 2024 ; Bergamaschi et al. 2024 ; Rajakumar and Watson 2024 ; Brunner et al. 2024 , imaginary time evolution Motta et al. 2019 ; Sun et al. 2021 ; Kamakari et al. 2022 ; Getelina et al. 2023 , and state purification. Out of these, state purification is the most intuitive and amenable to near-term applications. The broad idea is to purify the mixed state of the effective open system by enlarging the Hilbert space using ancillary qubits. Unitary operations can then be used to find a pure state on the extended system that prepares the Gibbs state on the system register, after tracing out the ancillary qubits. Known variational methods can be used to find the purified state. A particular purification, the thermofield double (TFD) state, is interesting in its own right from the perspective of holography Maldacena 2003 ; Maldacena and Susskind 2013 . Preparing the TFD state, by definition, requires doubling the number of system qubits for purification Cottrell et al. 2019 ; Wu and Hsieh 2019 ; Sagastizabal et al. 2021 . Variationally preparing the Gibbs state this way can have the advantage of mapping to a problem of finding the ground state of a Hamiltonian, albeit in a larger Hilbert space Cottrell et al. 2019 . However, the cost of doubling the system size is excessive, especially if the goal is to prepare Gibbs states at low-temperatures. Figure 1: A diagrammatic workflow for TEPID-ADAPT. The white shapes indicate the inputs– the Hamiltonian H , the inverse temperature \bar{\beta} , an m -dimensional computational subspace, and an operator pool for ADAPT. The back circuit, labeled U_{m}(\vec{\mu}) is the static part of the ansatz, which prepares \rho_{m} on the system register. The pink circuit parametrized by \vec{\theta} is the adaptively generated portion that maps the computational subspace to the truncated eigenspace of the Hamiltonian. The two solid green arrows (from left-right) respectively indicate the convergence of the VQE subroutine and the pool gradients vanishing. The primary output of TEPID-ADAPT is the target Gibbs state at \rho_{G}(\bar{\beta}) . We can also prepare the low-lying eigenstates of the Hamiltonian in the truncated subspace \{\ket{\psi_k}\}_{k=1}^{m} and the Gibbs states at lower temperatures \rho_{G}(\beta>\bar{\beta}) . Variational methods that prepare Gibbs states using other purifications face an obstacle at the level of the cost function. The free energy Eq. ( 2 ) is difficult to measure even on quantum hardware because measuring the von Neumann entropy of an arbitrary density matrix requires at least partial state tomography. To get around this difficulty, Ref. Wang et al. 2021 uses a truncated Taylor series of the free energy as the cost function, while Ref. Warren et al. 2022 uses a different cost function that measures proximity to a truncated version of the target Gibbs state. In Ref. Consiglio et al. 2024 , the free energy without any truncation is used as a cost function. This is enabled by a clever modular construction of the parametrized quantum circuit that first prepares a diagonal mixed state on the system qubits, followed by operations that leave the von Neumann entropy invariant. However, the use of a hardware-efficient ansatz for the first block of the circuit and the fact that more ancilla qubits than necessary are used introduce ad hoc elements to the algorithm, additional measurements, and a classical post processing overhead. In this work, we introduce an algorithm that addresses all these issues