Search

Complex dynamic systems are typically either modeled using expert knowledge in the form of differential equations or via data-driven universal approximation models such as artificial neural networks (ANN). While the first approach has advantages with respect to interpretability, transparency, data-efficiency, and extrapolation, the second approach is able to learn completely unknown functional relations from data and may result in models that can be evaluated more efficiently. To combine the complementary advantages, universal differential equations (UDE) have been suggested. They replace unknown terms in the differential equations with ANN. Such hybrid models allow to both encode prior domain knowledge, such as first principles, and to learn unknown mechanisms from data. Often, data for the training of UDE can only be obtained via costly experiments. We consider optimal experimental design (OED) for planning of experiments and generating data needed to train UDE. The number of weights in the embedded ANN usually leads to an overfitting of the regression problem. To make the OED problem tractable for optimization, we propose and compare dimension reduction methods that are based on lumping of weights and singular value decomposition of the Fisher information matrix (FIM), respectively. They result in lower-dimensional variational differential equations, which are easier to solve and yield regular FIM. Our numerical results showcase the advantages of OED for UDE, such as increased data-efficiency and better extrapolation properties.

We present a novel intelligent-system architecture called "Dynamic Net Architecture" (DNA) that relies on recurrence-stabilized networks and discuss it in application to vision. Our architecture models a (cerebral cortical) area wherein elementary feature neurons encode details of visual structures, and coherent nets of such neurons model holistic object structures. By interpreting smaller or larger coherent pieces of an area network as complex features, our model encodes hierarchical feature representations essentially different than artificial neural networks (ANNs). DNA models operate on a dynamic connectionism principle, wherein neural activations stemming from initial afferent signals undergo stabilization through a self-organizing mechanism facilitated by Hebbian plasticity alongside periodically tightening inhibition. In contrast to ANNs, which rely on feed-forward connections and backpropagation of error, we posit that this processing paradigm leads to highly robust representations, as by employing dynamic lateral connections, irrelevant details in neural activations are filtered out, freeing further processing steps from distracting noise and premature decisions. We empirically demonstrate the viability of the DNA by composing line fragments into longer lines and show that the construction of nets representing lines remains robust even with the introduction of up to $59\%$ noise at each spatial location. Furthermore, we demonstrate the model's capability to reconstruct anticipated features from partially obscured inputs and that it can generalize to patterns not observed during training. In this work, we limit the DNA to one cortical area and focus on its internals while providing insights into a standalone area's strengths and shortcomings. Additionally, we provide an outlook on how future work can implement invariant object recognition by combining multiple areas.

Accurate models for open quantum systems -- quantum states that have non-trivial interactions with their environment -- may aid in the advancement of a diverse array of fields, including quantum computation, informatics, and the prediction of static and dynamic molecular properties. In recent years, quantum algorithms have been leveraged for the computation of open quantum systems as the predicted quantum advantage of quantum devices over classical ones may allow previously inaccessible applications. Accomplishing this goal will require input and expertise from different research perspectives, as well as the training of a diverse quantum workforce, making a compilation of current quantum methods for treating open quantum systems both useful and timely. In this Review, we first provide a succinct summary of the fundamental theory of open quantum systems and then delve into a discussion on recent quantum algorithms. We conclude with a discussion of pertinent applications, demonstrating the applicability of this field to realistic chemical, biological, and material systems.

Living organisms exhibit remarkable adaptations across all scales, from molecules to ecosystems. We believe that many of these adaptations correspond to optimal solutions driven by evolution, training, and underlying physical and chemical laws and constraints. While some argue against such optimality principles due to their potential ambiguity, we propose generalized inverse optimal control to infer them directly from data. This novel approach incorporates multi-criteria optimality, nestedness of objective functions on different scales, the presence of active constraints, the possibility of switches of optimality principles during the observed time horizon, maximization of robustness, and minimization of time as important special cases, as well as uncertainties involved with the mathematical modeling of biological systems. This data-driven approach ensures that optimality principles are not merely theoretical constructs but are firmly rooted in experimental observations. Furthermore, the inferred principles can be used in forward optimal control to predict and manipulate biological systems, with possible applications in bio-medicine, biotechnology, and agriculture. As discussed and illustrated, the well-posed problem formulation and the inference are challenging and require a substantial interdisciplinary effort in the development of theory and robust numerical methods.

For continuous decision spaces, nonlinear programs (NLPs) can be efficiently solved via sequential quadratic programming (SQP) and, more generally, sequential convex programming (SCP). These algorithms linearize only the nonlinear equality constraints and keep the outer convex structure of the problem intact. The aim of the presented sequential mixed-integer quadratic programming (MIQP) algorithm for mixed-integer nonlinear problems (MINLPs) is to extend the SQP/SCP methodology to MINLPs and leverage the availability of efficient MIQP solvers. The algorithm employs a three-step method in each iterate: First, the MINLP is linearized at a given iterate. Second, an MIQP with its feasible set restricted to a specific region around the current linearization point is formulated and solved. Third, the integer variables obtained from the MIQP solution are fixed, and only an NLP in the continuous variables is solved. The outcome of the third step is compared to previous iterates, and the best iterate so far is used as a linearization point in the next iterate. Crucially, the objective values and derivatives from all previous iterates are used to formulate the polyhedral region in the second step. The linear inequalities that define the region build on concepts from generalized Benders' decomposition for MINLPs. Although the presented MINLP algorithm is a heuristic method without any global optimality guarantee, it converges to the exact integer solution when applied to convex MINLP with a linear outer structure. The conducted numerical experiments demonstrate that the proposed algorithm is competitive with other open-source solvers for MINLP. Finally, we solve two mixed-integer optimal control problems (MIOCPs) transcribed into MINLPs via direct methods, showing that the presented algorithm can effectively deal with nonlinear equality constraints, a major hurdle for generic MINLP solvers., Comment: Andrea Ghezzi and Wim Van Roy contributed equally to this work. 34 pages, 11 figures, 3 tables

We present a novel, general, and unifying point of view on sparse approaches to polynomial optimization. Solving polynomial optimization problems to global optimality is a ubiquitous challenge in many areas of science and engineering. Historically, different approaches on how to solve nonconvex polynomial optimization problems based on convex relaxations have been developed in different scientific communities. Here, we introduce the concept of monomial patterns. A pattern determines what monomials are to be linked by convex constraints in a convex relaxation of a polynomial optimization problem. This concept helps to understand existing approaches from different schools of thought, to develop novel relaxation schemes, and to derive a flexible duality theory, which can be specialized to many concrete situations that have been considered in the literature. We unify different approaches to polynomial optimization including polyhedral approximations, dense semidefinite relaxations, SONC, SAGE, and TSSOS in a self-contained exposition. We also carry out computational experiments to demonstrate the practical advantages of a flexible usage of pattern-based sparse relaxations of polynomial optimization problems., Comment: This manuscript evolved from arxiv:1901.05675 but its presentation and contents are significantly different. The manuscript uses the same numerical results as in arxiv:1901.05675

The moment-sum of squares hierarchy by Lasserre has become an established technique for solving polynomial optimization problems. It provides a monotonically increasing series of tight bounds, but has well-known scalability limitations. For structured optimization problems, the term-sparsity SOS (TSSOS) approach scales much better due to block-diagonal matrices, obtained by completing the connected components of adjacency graphs. This block structure can be exploited by semidefinite programming solvers, for which the overall runtime then depends heavily on the size of the largest block. However, already the first step of the TSSOS hierarchy may result in large diagonal blocks. We suggest a new approach that refines TSSOS iterations using combinatorial optimization and results in block-diagonal matrices with reduced maximum block sizes. Numerical results on a benchmark library show the large potential for computational speedup for unconstrained and constrained polynomial optimization problems, while obtaining almost identical bounds in comparison to established methods.

In the early 20th century, Laurent Schwartz observed that we can identify functions that extend smoothly to the point at infinity of one-point compactifications of Euclidean spaces. We show a similar result for a different compactification of Euclidean spaces, namely, the real projective spaces.

Air pollution generates substantial health damages and economic costs worldwide. Pollution exposure varies greatly, both between countries and within them. However, the degree of air quality inequality and its' trajectory over time have not been quantified at a global level. Here I use economic inequality indices to measure global inequality in exposure to ambient fine particles with 2.5 microns or less in diameter (PM2.5). I find high and rising levels of global air quality inequality. The global PM2.5 Gini Index increased from 0.32 in 2000 to 0.36 in 2020, exceeding levels of income inequality in many countries. Air quality inequality is mostly driven by differences between countries and less so by variation within them, as decomposition analysis shows. A large share of people facing the highest levels of PM2.5 exposure are concentrated in only a few countries. The findings suggest that research and policy efforts that focus only on differences within countries are overlooking an important global dimension of environmental justice.

Quantum computers are promising tools for simulating many-body quantum systems due to their potential scaling advantage over classical computers. While significant effort has been expended on many-fermion systems, here we simulate a model entangled many-boson system with the contracted quantum eigensolver (CQE). We generalize the CQE to many-boson systems by encoding the bosonic wavefunction on qubits. The CQE provides a compact ansatz for the bosonic wave function whose gradient is proportional to the residual of a contracted Schr\"odinger equation. We apply the CQE to a bosonic system, where $N$ quantum harmonic oscillators are coupled through a pairwise quadratic repulsion. The model is relevant to the study of coupled vibrations in molecular systems on quantum devices. Results demonstrate the potential efficiency of the CQE in simulating bosonic processes such as molecular vibrations with good accuracy and convergence even in the presence of noise.

An outstanding challenge in chemical computation is the many-electron problem where computational methodologies scale prohibitively with system size. The energy of any molecule can be expressed as a weighted sum of the energies of two-electron wave functions that are computable from only a two-electron calculation. Despite the physical elegance of this extended ``aufbau'' principle, the determination of the distribution of weights -- geminal occupations -- for general molecular systems has remained elusive. Here we introduce a new paradigm for electronic structure where approximate geminal-occupation distributions are ``learned'' via a convolutional neural network. We show that the neural network learns the $N$-representability conditions, constraints on the distribution for it to represent an $N$-electron system. By training on hydrocarbon isomers with only 2-7 carbon atoms, we are able to predict the energies for isomers of octane as well as hydrocarbons with 8-15 carbons. The present work demonstrates that machine learning can be used to reduce the many-electron problem to an effective two-electron problem, opening new opportunities for accurately predicting electronic structure.

Fermion-exciton condensation in which both fermion-pair (i.e., superconductivity) and exciton condensations occur simultaneously in a single coherent quantum state has recently been conjectured to exist. Here, we capture the fermion-exciton condensation through a model Hamiltonian that can recreate the physics of this new class of highly correlated condensation phenomena. We demonstrate that the Hamiltonian generates the large-eigenvalue signatures of fermion-pair and exciton condensations for a series of states with increasing particle numbers. The results confirm that the dual-condensate wave function arises from the entanglement of fermion-pair and exciton wave functions, which we previously predicted in the thermodynamic limit. This model Hamiltonian -- generalizing well-known model Hamiltonians for either superconductivity or exciton condensation -- can explore a wide variety of condensation behavior. It provides significant insights into the required forces for generating a fermion-exciton condensate, which will likely be invaluable for realizing such condensations in realistic materials with applications from superconductors to excitonic materials.

Fermion-exciton condensates (FECs) -- computationally and theoretically predicted states that simultaneously exhibit the character of superconducting states and exciton condensates -- are novel quantum states whose properties may involve a hybridization of superconductivity and the dissipationless flow of energy. Here, we exploit prior investigations of superconducting states and exciton condensates on quantum devices to identify a tuneable quantum state preparation entangling the wave functions of the individual condensate states. Utilizing this state preparation, we prepare a variety of FEC states on quantum computers -- realizing strongly correlated FEC states on current, noisy intermediate-scale quantum devices -- and verify the presence of the dual condensate via postmeasurement analysis. This confirmation of the previously predicted condensate state on quantum devices as well as the form of its wave function motivates further theoretical and experimental exploration of the properties, applications, and stability of FECs.

Exciton condensation, a Bose-Einstein condensation of excitons into a single quantum state, has recently been achieved in low-dimensional materials including twin layers of graphene and van der Waals heterostructures. Here we examine computationally the beginnings of exciton condensation in a double layer comprised of coronene, a seven-benzene-ring patch of graphene. As a function of interlayer separation, we compute the exciton population in a single coherent quantum state, showing that the population peaks around 1.8 at distances near 2 \AA. Visualization reveals interlayer excitons at the separation distance of the condensate. We determine the exciton population as a function of the twist angle between the two coronene layers to reveal the magic angles at which the condensation peaks. As with previous recent calculations showing some exciton condensation in hexacene double layers and benzene stacks, the present two-electron reduced-density-matrix calculations with coronene provide computational evidence for the ability to realize exciton condensation in molecular-scale analogs of extended systems like the graphene double layer.

The Bose-Einstein condensation of excitons into a single quantum state is known as exciton condensation. Exciton condensation, which potentially supports the frictionless flow of energy, has recently been realized in graphene bilayers and van der Waals heterostructures. Here we show that exciton condensates can be predicted from a combination of reduced density matrix theory and cumulant theory. We show that exciton condensation occurs if and only if there exists a large eigenvalue in the cumulant part of the particle-hole reduced density matrix. In the thermodynamic limit we show that the large eigenvalue is bounded from above by the number of excitons. In contrast to the eigenvalues of the particle-hole matrix, the large eigenvalue of the cumulant matrix has the advantage of providing a size-extensive measure of the extent of condensation. Here we apply this signature to predict exciton condensation in both the Lipkin model and molecular stacks of benzene. The computational signature has applications to the prediction of exciton condensation in both molecules and materials.

Quantum states of a novel Bose-Einstein condensate, in which both fermion-pair and exciton condensations are simultaneously present, have recently been realized theoretically in a model Hamiltonian system. Here we identify quantum phase transitions in that model between fermion-pair and exciton condensations based on a geometric analysis of the convex set of ground-state 2-particle reduced density matrices (2-RDMs). The 2-RDM set provides a finite representation of the infinite parameter space of Hamiltonians that readily reveals a fermion-pair condensate phase and two distinct exciton condensate phases, as well as the emergence of first- and second-order phase transitions as the particle number of the system is increased. The set, furthermore, shows that the fermion-exciton condensate (FEC) lies along the second-order phase transition between the exciton and fermion-pair condensate phases. The detailed information about the exciton and fermion-pair phases, the forces behind these phase, as well as their associated transitions provides additional insight into the formation of the FEC condensate, which we anticipate will prove useful in its experimental realization.

The 3UCubed project is a 3U CubeSat being jointly developed by the University of New Hampshire, Sonoma State University, and Howard University as a part of the NASA Interstellar Mapping and Acceleration Probe, IMAP, student collaboration. This project comprises of a multidisciplinary team of undergraduate students from all three universities. The mission goal of the 3UCubed is to understand how Earths polar upper atmosphere the thermosphere in Earths auroral regions, responds to particle precipitation and solar wind forcing, and internal magnetospheric processes. 3UCubed includes two instruments with rocket heritage to achieve the science mission: an ultraviolet photomultiplier tube, UVPMT, and an electron retarding potential analyzer ERPA. The spacecraft bus consists of the following subsystems: Attitude Determination and Control, Command and Data Handling, Power, Communication, Structural, and Thermal. Currently, the project is in the post-PDR stage, starting to build and test engineering models to develop a FlatSat prior to critical design review in 2023. The goal is to launch at least one 3U CubeSat to collect science data close to the anticipated peak of Solar Cycle 25 around July 2025. Our mother mission, IMAP, is also projected to launch in 2025, which will let us jointly analyze the science data of the main mission, providing the solar wind measurements and inputs to the magnetosphere with that of 3UCubed, providing the response of Earths cusp to these inputs.