To simulate the time evolution of a quantum many-body system, researchers traditionally expand the wave function in a set of fixed basis functions, such as static atomic orbitals or a rigid spatial grid. However, as particles interact and move, the system explores vast regions of Hilbert space. To maintain accuracy, a static basis must be large enough to cover every configuration the system might visit, leading directly to the curse of dimensionality.
In single-configuration mean-field theories, like the Gross-Pitaevskii equation for Bose-Einstein condensates, the entire system is constrained to a single orbital. While computationally efficient, Streltsov et al. demonstrate in Reduced density matrices and coherence of trapped interacting bosons that such approximations fail to capture momentum-space correlations and the loss of coherence during fragmentation. When a system undergoes structural changes—such as a condensate splitting in a double-well potential—fixed-basis models underestimate quantum correlations because they cannot account for the structural evolution of spatial orbitals (Exact Quantum Dynamics of a Bosonic Josephson Junction). As Alberto Baiardi argues in Large-Scale Quantum Dynamics with Matrix Product States, relying on a fixed local basis fundamentally limits the flexibility of a simulation to adapt to significant structural changes.
The Core Mechanism: Time-Dependent Variational Subspaces
The core insight of multiconfigurational time-dependent methods is to make the basis functions themselves time-dependent and optimize them dynamically. Instead of choosing a massive, permanent set of basis functions, the simulation selects a much smaller set of orbitals and allows them to change shape, translate, and deform in real time. The many-body wave function is then constructed as a superposition of multiple configurations built from these moving orbitals.
To determine exactly how these orbitals should evolve, the method invokes the time-dependent variational principle. At each time step, the algorithm solves equations of motion that simultaneously update both the expansion coefficients of the configurations and the mathematical forms of the underlying orbitals. By forcing the basis functions to follow the physics, the simulation tracks the wave function within a compact, moving subspace. As shown by Streltsov et al. in Accurate multi-boson long-time dynamics in triple-well periodic traps, this time-adaptive approach reduces the required configuration space by several orders of magnitude compared to fixed-basis methods while maintaining strict accuracy for complex many-body phenomena.
Visualizing Fragmentation and Tunneling
This adaptive framework is uniquely suited for capturing "fragmentation"—a phenomenon where particles macroscopically occupy multiple quantum states simultaneously, breaking mean-field assumptions. In Universality of fragmentation in the Schrödinger dynamics of bosonic Josephson junctions, Streltsov et al. show that condensates in a double-well potential inevitably fragment over long times, regardless of how weak the interactions are or how many particles are in the system. A static single-orbital model cannot represent this breakdown. In contrast, multiconfigurational time-dependent many-body simulations allow the orbitals to split and deform variationally, capturing the intricate particle correlations that emerge (Quantum dynamics of attractive versus repulsive bosonic Josephson junctions: Bose-Hubbard and full-Hamiltonian results).
Similarly, when particles tunnel through a barrier into open space, the process is not a simple, unified flow. Lode et al. demonstrate in How an interacting many-body system tunnels through a potential barrier to open space that tunneling triggers a complex quantum interference of multiple single-particle processes, causing the system to lose coherence. By expanding the wave function across multiple time-varying orbitals, the simulation directly captures this emergent fragmentation. Mark A. Kasevich leverages this variational framework in Single-shot simulations of dynamic quantum many-body systems to simulate single-shot experimental images, revealing fluctuations and interference fringes that vanish in standard theoretical averages.
Deforming Nodal Surfaces and Tracking Electrons
The power of adaptive bases extends to fermions (like electrons), where the geometric constraints of the wave function are even more severe due to the Pauli exclusion principle. This principle creates "nodal surfaces"—regions where the wave function is exactly zero. In standard quantum simulations, holding these nodal surfaces in a fixed configuration introduces a severe systematic bias known as fixed-node error.
To circumvent this, researchers integrate time-dependent backflow transformations into the variational ansatz, allowing the nodal surfaces to deform during the simulation. In Ab-initio variational wave functions for the time-dependent many-electron Schrödinger equation, Nys et al. demonstrate that dynamic adjustment of the wave function's nodal surface prevents the accumulation of errors over time. This mechanism provides the mathematical flexibility required to model highly non-equilibrium processes, such as electron ionization under intense laser fields where electrons transition to continuum states (Time-dependent multiconfiguration self-consistent-field and time-dependent optimized coupled-cluster methods for intense laser-driven multielectron dynamics). By optimizing these boundaries on the fly, as also explored by Parker et al. in Accuracy of quantum Monte Carlo methods for point defects in solids and Bird et al. in Partial suppression of magnetism in the square lattice SU(3) Hubbard model, the wave function can accurately model phase transitions and defects without being biased by a rigid initial nodal structure.
Scaling to Complex Molecular Vibrations
When modeling molecular vibrations, large-amplitude motions typically require an exponentially large static basis set to capture nuclear displacements. By switching to an adaptive basis, the basis functions contract and expand alongside the molecular motion. Jensen et al. show in Efficient time-dependent vibrational coupled cluster computations with time-dependent basis sets at the two-mode coupling level: full and hybrid TDMVCC[2] that using time-dependent basis functions reduces the computational cost from exponential to cubic or quadratic scaling, enabling simulations of molecules with hundreds of vibrational modes.
While maintaining strict orthogonality among these moving basis functions provides numerical stability, Højlund et al. point out in Time-dependent coupled cluster with orthogonal adaptive basis functions: General formalism and application to the vibrational problem that it introduces subtle mathematical trade-offs in convergence behavior compared to non-orthogonal alternatives. Nonetheless, whether applied to vibrational coupled-cluster theory or density functional theory (Optimized local basis set for Kohn-Sham density functional theory), the underlying principle remains identical: optimizing a small set of basis functions on the fly allows the simulation to adapt to the specific chemical environment rather than relying on a rigid, pre-calculated space.
Key Foundational Work
Exact Quantum Dynamics of a Bosonic Josephson Junction (2009) — contributed 3 concept elements to this direction.
Reduced density matrices and coherence of trapped interacting bosons (2008) — contributed 1 concept elements to this direction.