Imaginary Time Evolution Improves Quantum Many-Body Systems
Imaginary Time Evolution
IBM Research, the STFC Hartree Center, and the University of Edinburgh devised a sophisticated algorithm to prepare excited eigenstates in many-body quantum systems, furthering quantum physics and computational chemistry. The result in “Shift-Invert Imaginary Time Evolution for Many-Body Excited States” lets researchers target quantum energy levels without calculating every lower state. This could speed up medicine, solar cell, and electronic material development.
Fighting “Excited” Matter
Like a ball in a bowl, quantum physics' "ground state" is a system's lowest energy level. Ground states have been well-identified by researchers. Many critical events in physics and quantum chemistry occur in “excited states,” where atoms or electrons absorb energy and move to higher energy levels.
“The excited state manifold of quantum systems fundamentally shapes a wide range of chemical and physical phenomena,” researchers say. Excited states affect how a molecule reacts to light in a photovoltaic cell and chemical bonding.
Despite their importance, these states are difficult to recreate. As systems grow, computers cannot select through hundreds or thousands of undesirable lower-energy states to find a highly excited one. Many-body localization and eigenstate thermalization research are limited by our understanding of highly entangled excited states.
New Mathematical Mirror: Shift-Invert Trick
D. A. Millar and S. J. Thomson lead the research team to overcome this issue by ingeniously merging the shift-invert mechanism with imaginary time development.
The shift-invert approach transforms an excited state at the target energy (δ) into the ground state of an effective Hamiltonian by modifying the system's total energy operator (Hamiltonian). The authors' main advance is imaginary-time evolution with respect to the shift-inverted Hamiltonian. This lets researchers use imaginary-time evolution, a valid ground-state identification method, to cool the system directly to the goal state.
Avoiding Exponential Wall
The shift-invert approach was formerly considered too "costly" for big systems since it needed the direct inversion of a huge matrix, which scales exponentially with system size and increases computational effort fast as the number of variables increases. For this, the group created a variational self-consistency criterion. Instead of inverting the matrix, they evolve the quantum state in tiny, microscopic time steps to solve a simpler optimization issue at each iteration.
This new method replaces an impossible computation with manageable iterations. The approach was tested on disordered spin chains, a prominent model in condensed matter physics, and yielded low-variance states for systems up to 128 sites, a notable performance for mid-energy states.
Bridge to Quantum Computing
Its interoperability with near-term quantum technology makes the algorithm most promising. It works on conventional supercomputers nowadays.
“Highly entangled” states occur when a quantum system's components are so closely connected that a normal computer can't represent them. This “exponential wall” doesn't affect quantum computers.
The researchers recommend a hybrid approach: the program starts with a “loose approximation” of a standard computer state. When the system gets too entangled for classical hardware, a shallow quantum circuit is created as a “warm start” for imaginary time evolution on a quantum processor. We think that [the algorithm's] genuine long-term usefulness will be in building highly entangled states directly on quantum hardware,” the team says.
Industrial and Scientific Impact
Directly targeting excited states has huge implications. Molecular excitations can show drug developers how a molecule reacts to light or changes shape after absorbing energy. It could also properly define the "band gap" in materials science, the energy difference that determines whether a material is an insulator or a conductor, to make better semiconductors.
The approach is Hamiltonian-agnostic, thus no system knowledge is needed. The same algorithm works for complicated proteins and novel superconducting magnets.
Cooperative Work
A multi-institutional project was supported by EPSRC and Hartree National Center for Digital Innovation. The partnership includes IBM Research UK, Hartree Center, Southampton, Cambridge, Oxford, and Edinburgh professionals. As quantum hardware progresses, this technique gives researchers a roadmap for unleashing the excited-state manifold's full potential, bringing us closer to quantum-designed materials and treatments.











