Search

Extraction of structure, in particular of group symmetries, is increasingly crucial to understanding and building intelligent models. In particular, some information-theoretic models of parsimonious learning have been argued to induce invariance extraction. Here, we formalise these arguments from a group-theoretic perspective. We then extend them to the study of more general probabilistic symmetries, through compressions preserving well-studied geometric measures of complexity. More precisely, we formalise a trade-off between compression and preservation of the divergence from a given hierarchical model, yielding a novel generalisation of the Information Bottleneck framework. Through appropriate choices of hierarchical models, we fully characterise (in the discrete and full support case) channel invariance, channel equivariance and distribution invariance under permutation. Allowing imperfect divergence preservation then leads to principled definitions of "soft symmetries", where the "coarseness" corresponds to the degree of compression of the system. In simple synthetic experiments, we demonstrate that our method successively recovers, at increasingly compressed "resolutions", nested but increasingly perturbed equivariances, where new equivariances emerge at bifurcation points of the trade-off parameter. Our framework suggests a new path for the extraction of generalised probabilistic symmetries., Comment: 34 pages, 2 figures

The presence of symmetries imposes a stringent set of constraints on a system. This constrained structure allows intelligent agents interacting with such a system to drastically improve the efficiency of learning and generalization, through the internalisation of the system's symmetries into their information-processing. In parallel, principled models of complexity-constrained learning and behaviour make increasing use of information-theoretic methods. Here, we wish to marry these two perspectives and understand whether and in which form the information-theoretic lens can "see" the effect of symmetries of a system. For this purpose, we propose a novel variant of the Information Bottleneck principle, which has served as a productive basis for many principled studies of learning and information-constrained adaptive behaviour. We show (in the discrete case and under a specific technical assumption) that our approach formalises a certain duality between symmetry and information parsimony: namely, channel equivariances can be characterised by the optimal mutual information-preserving joint compression of the channel's input and output. This information-theoretic treatment furthermore suggests a principled notion of "soft" equivariance, whose "coarseness" is measured by the amount of input-output mutual information preserved by the corresponding optimal compression. This new notion offers a bridge between the field of bounded rationality and the study of symmetries in neural representations. The framework may also allow (exact and soft) equivariances to be automatically discovered., Comment: 23 pages, 0 figures

The mastery of skills such as playing tennis or balancing an inverted pendulum implies a very accurate control of movements to achieve the task goals. Traditional accounts of skilled action control that focus on either routinization or perceptual control make opposite predictions about the ways we achieve mastery. The notion of routinization emphasizes the decrease of the variance of our actions, whereas the notion of perceptual control emphasizes the decrease of the variance of the states we visit, but not of the actions we execute. Here, we studied how participants managed control tasks of varying levels of complexity, which consisted in controlling inverted pendulums of different lengths. We used information-theoretic measures to compare the predictions of alternative theoretic accounts that focus on routinization and perceptual control, respectively. Our results indicate that the successful performance of the control task strongly correlates with the decrease of state variability and the increase of action variability. As postulated by perceptual control theory, the mastery of skills consists in achieving stable control goals by flexible means., Comment: 15 pages, 7 figures

Traditionally, Euclidean geometry is treated by scientists as a priori and objective. However, when we take the position of an agent, the problem of selecting a best route should also factor in the abilities of the agent, its embodiment and particularly its cognitive effort. In this paper we consider geometry in terms of travel between states within a world by incorporating information processing costs with the appropriate spatial distances. This induces a geometry that increasingly differs from the original geometry of the given world as information costs become increasingly important. We visualise this "cognitive geometry" by projecting it onto 2- and 3-dimensional spaces showing distinct distortions reflecting the emergence of epistemic and information-saving strategies as well as pivot states. The analogies between traditional cost-based geometries and those induced by additional informational costs invite a generalisation of the notion of geodesics as cheapest routes towards the notion of infodesics. In this perspective, the concept of infodesics is inspired by the property of geodesics that, travelling from a given start location to a given goal location along a geodesic, not only the goal, but all points along the way are visited at optimal cost from the start., Comment: Includes supplementary material, 6 figures in the main document, 5 figures in the supplementary material. Replacing preprint with author accepted manuscript

We study Chevalley-Eilenberg cohomology of physically relevant Lie superalgebras related to supersymmetric theories, providing explicit expressions for their cocycles in terms of their Maurer-Cartan forms. We then include integral forms in the picture by defining a notion of integral forms related to a Lie superalgebra. We develop a suitable generalization of Chevalley-Eilenberg cohomology extended to integral forms and we prove that it is isomorphic to the ordinary Chevalley-Eilenberg cohomology of the Lie superalgebra. Next we study equivariant Chevalley-Eilenberg cohomology for coset superspaces, which plays a crucial role in supergravity and superstring models. Again, we treat explicitly several examples, providing cocycles' expressions and revealing a characteristic infinite dimensional cohomology., Comment: Added some references

We study the cohomology of the complexes of differential, integral and pseudo forms on odd symplectic manifolds taking the wedge product with the symplectic form as differential. We show that the cohomology classes are in correspondence with inequivalent Lagrangian submanifolds and that they all define semidensities on them. Further, we introduce new operators that move from one Lagragian submanifold to another and we investigate their relation with the so-called picture changing operators for the de Rham differential. Finally, we prove the isomorphism between the cohomology of the de Rham differential and the cohomology of BV Laplacian in the extended framework of differential, integral and pseudo forms., Comment: 23 pages

Inspired by the analogy between different types of differential forms on supermanifolds and string fields in superstring theory, we construct new multilinear non-associative products of forms which yield an $A_\infty$-algebra., Comment: 29 pages, no figures

Inspired by superstring field theory, we study differential, integral, and inverse forms and their mutual relations on a supermanifold from a sheaf-theoretical point of view. In particular, the formal distributional properties of integral forms are recovered in this scenario in a geometrical way. Further, we show how inverse forms extend the ordinary de Rham complex on a supermanifold, thus providing a mathematical foundation of the Large Hilbert Space used in superstrings. Last, we briefly discuss how the Hodge diamond of a supermanifold looks like, and we explicitly compute it for super Riemann surfaces., Comment: LaTeX, 39pp, no figures

We use the techniques of integral forms to analyse the easiest example of two dimensional sigma models on a supermanifold. We write the action as an integral of a top integral form over a D=2 supermanifold and we show how to interpolate between different superspace actions. Then, we consider curved supermanifolds and we show that the definitions used for flat supermanifold can also be used for curved supermanifolds. We prove it by first considering the case of a curved rigid supermanifold and then the case of a generic curved supermanifold described by a single superfield $E$., Comment: 18 pages. Contribution to the Special Issue of Universe "Super Geometry for Super Strings"

We reconstruct the action of $N=1, D=4$ Wess-Zumino and $N=1, 2, D=4$ super-Yang-Mills theories, using integral top forms on the supermanifold ${\cal M}^{(4|4)}$. Choosing different Picture Changing Operators, we show the equivalence of their rheonomic and superspace actions. The corresponding supergeometry and integration theory are discussed in detail. This formalism is an efficient tool for building supersymmetric models in a geometrical framework., Comment: 40 pages, Latex2e

We reformulate Super Quantum Mechanics in the context of integral forms. This framework allows to interpolate between different actions for the same theory, connected by different choices of Picture Changing Operators (PCO). In this way we retrieve component and superspace actions, and prove their equivalence. The PCO are closed integral forms, and can be interpreted as super Poincar\'e duals of bosonic submanifolds embedded into a supermanifold.. We use them to construct Lagrangians that are top integral forms, and therefore can be integrated on the whole supermanifold. The $D=1, ~N=1$ and the $D=1,~ N=2$ cases are studied, in a flat and in a curved supermanifold. In this formalism we also consider coupling with gauge fields, Hilbert space of quantum states and observables., Comment: 41 pages, no figures. Use birkjour.cls. Minor misprints, moved appendix A and B in the main text. Version to be published in Annales H. Poincar\'e

We propose a flexible design for the identification of optimal dose combinations in dual-agent dose-finding clinical trials. The design is called AAA, standing for three adaptations: adaptive model selection, adaptive dose insertion, and adaptive cohort divi- sion. The adaptations highlight the need and opportunity for innovation for dual-agent dose finding, and are supported by the numerical results presented in the proposed simulation studies. To our knowledge, this is the first design that allows for all three adaptations at the same time. We find that AAA improves the statistical inference, enhances the chance of finding the optimal dose combinations, and shortens the trial duration. A clinical trial is being planned to apply the AAA design.

In our previous work, a reduced order model (ROM) for a stochastic system was made, where noisy data was projected onto principal component analysis (PCA)-derived basis vectors to obtain an accurate reconstruction of the noise-free data. That work used techniques designed for deterministic data, PCA was used for the basis function generation and $L_2$ projection was used to create the reconstructions. In this work, probabilistic approaches are used. The probabilistic PCA (PPCA) is used to generate the basis, which then allows the noise in the training data to be estimated. PPCA has also been improved so that the derived basis vectors are orthonormal and the variance of the basis expansion coefficients over the training data set can be estimated. The standard approach assumes a unit variance for these coefficients. Based on the results of the PPCA, model selection criteria are applied to automatically choose the dimension of the ROM. In our previous work, a heuristic approach was used to pick the dimension. Lastly, a new statistical approach is used for the projection step where the variance information obtained from the improved PPCA is used as a prior to improve the projection. This gives improved accuracy over $L_2$ projection when the projected data is noisy. In addition, the noise statistics for the projected data are not assumed to be the same as that of the training data, but are estimated in the projection process. The entire approach gives a fully stochastic method for computing a ROM from noisy training data, determining ideal model selection, and projecting noisy test data, thus enabling accurate predictions of noise-free data from data that is dominated by noise., Comment: 21 pages, 7 figures

By using integral forms we derive the superspace action of D=3, N=1 supergravity as an integral on a supermanifold. The construction is based on target space picture changing operators, here playing the role of Poincare' duals to the lower-dimensional spacetime surfaces embedded into the supermanifold. We show how the group geometrical action based on the group manifold approach interpolates between the superspace and the component supergravity actions, thus providing another proof of their equivalence., Comment: 22 pages

We present few types of integral transforms and integral representations that are very useful for extending to supergeometry many familiar concepts of differential geometry. Among them we discuss the construction of the super Hodge dual, the integral representation of picture changing operators of string theories and the construction of the super-Liouville form of a symplectic supermanifold., Comment: A section of the present paper was already presented in a previous preprint arXiv:1511.05105. This is the complete, extended version of that work. 21 pages. No figures. Names of authors corrected in the metadata. Minor corrections, version accepted for publication Lett. Math. Phys

We present a new construction for the Hodge operator for differential manifolds based on a Fourier (Berezin)-integral representation. We find a simple formula for the Hodge dual of the wedge product of differential forms, using the (Berezin)-convolution. The present analysis is easily extended to supergeometry and to non-commutative geometry., Comment: 10 pages, some misprint, changed factor in the last formula, added acknowledgement

We discuss the cohomology of superforms and integral forms from a new perspective based on a recently proposed Hodge dual operator. We show how the superspace constraints (a.k.a. rheonomic parametrisation) are translated from the space of superforms $\Omega^{(p|0)}$ to the space of integral forms $\Omega^{(p|m)}$ where $0 \leq p \leq n$, $n$ is the bosonic dimension of the supermanifold and $m$ its fermionic dimension. We dwell on the relation between supermanifolds with non-trivial curvature and Ramond-Ramond fields, for which the Laplace-Beltrami differential, constructed with our Hodge dual, is an essential ingredient. We discuss the definition of Picture Lowering and Picture Raising Operators (acting on the space of superforms and on the space of integral forms) and their relation with the cohomology. We construct non-abelian curvatures for gauge connections in the space $\Omega^{(1|m)}$ and finally discuss Hodge dual fields within the present framework., Comment: 35 pages

We construct the Hodge dual for supermanifolds by means of the Grassmannian Fourier transform of superforms. In the case of supermanifolds it is known that the superforms are not sufficient to construct a consistent integration theory and that the integral forms are needed. They are distribution-like forms which can be integrated on supermanifolds as a top form can be integrated on a conventional manifold. In our construction of the Hodge dual of superforms they arise naturally. The compatibility between Hodge duality and supersymmetry is exploited and applied to several examples. We define the irreducible representations of supersymmetry in terms of integral and superforms in a new way which can be easily generalised to several models in different dimensions. The construction of supersymmetric actions based on the Hodge duality is presented and new supersymmetric actions with higher derivative terms are found. These terms are required by the invertibility of the Hodge operator., Comment: LateX2e, 51 pages. Corrected some further misprints

Integral forms provide a natural and powerful tool for the construction of supergravity actions. They are generalizations of usual differential forms and are needed for a consistent theory of integration on supermanifolds. The group geometrical approach to supergravity and its variational principle are reformulated and clarified in this language. Central in our analysis is the Poincare' dual of a bosonic manifold embedded into a supermanifold. Finally, using integral forms we provide a proof of Gates' so-called "Ectoplasmic Integration Theorem", relating superfield actions to component actions., Comment: 36 pages, LaTex, corrected some misprints to match the published version in NPB

Dirac fermions have a central role in high energy physics but it is well known that they emerge also as quasiparticles in several condensed matter systems supporting topological order. We present a general method for deriving the topological effective actions of (3+1) massless Dirac fermions living on general backgrounds and coupled with vector and axial-vector gauge fields. The first step of our strategy is standard (in the Hermitian case) and consists in connecting the determinants of Dirac operators with the corresponding analytical indices through the zeta-function regularization. Then, we introduce a suitable splitting of the heat kernel that naturally selects the purely topological part of the determinant (i.e. the topological effective action). This topological effective action is expressed in terms of gauge fields using the Atiyah-Singer index theorem which computes the analytical index in topological terms. The main new result of this paper is to provide a consistent extension of this method to the non Hermitian case where a well-defined determinant does not exist. Quantum systems supporting relativistic fermions can thus be topologically classified on the basis of their response to the presence of (external or emergent) gauge fields through the corresponding topological effective field theories., Comment: 7 pages, published version

We present a study on the integral forms and their Cech/de Rham cohomology. We analyze the problem from a general perspective of sheaf theory and we explore examples in superprojective manifolds. Integral forms are fundamental in the theory of integration in supermanifolds. One can define the integral forms introducing a new sheaf containing, among other objects, the new basic forms delta(dtheta) where the symbol delta has the usual formal properties of Dirac's delta distribution and acts on functions and forms as a Dirac measure. They satisfy in addition some new relations on the sheaf. It turns out that the enlarged sheaf of integral and "ordinary" superforms contains also forms of "negative degree" and, moreover, due to the additional relations introduced, its cohomology is, in a non trivial way, different from the usual superform cohomology., Comment: 20 pages, LaTeX, we expanded the introduction, we add a complete analysis of the cohomology and we derive a new duality between cohomology groups

We first review the definition of superprojective spaces from the functor-of-points perspective. We derive the relation between superprojective spaces and supercosets in the framework of the theory of sheaves. As an application of the geometry of superprojective spaces, we extend Donaldson's definition of balanced manifolds to supermanifolds and we derive the new conditions of a balanced supermanifold. We apply the construction to superpoints viewed as submanifolds of superprojective spaces. We conclude with a list of open issues and interesting problems that can be addressed in the present context., Comment: 24 pages, Latex, no figures

We review the extraordinary fertility and proliferation in mathematics and physics of the concept of a surface with constant and negative Gaussian curvature. In his outstanding 1868 paper Beltrami discussed how non-Euclidean geometry is actually realized and displayed in a disk on the plane. This metric is intrinsically defined and definite but only if indefinite metrics are introduced it is possible to fully understand the structure of pseudospheres. In a 3D flat space R^3 the fundamental quadric is introduced, with the same signature as the metric of R^3; this leads to 3 kinds of surfaces with constant Gaussian curvature: the sphere, the single-sheet hyperboloid and the two-sheet hyperboloid; the last one is shown to be isomorphic to Beltrami's disk. The spacetime corresponding to the single-sheet case is de Sitter cosmological model, which, due to its symmetry, has an important role in cosmology. When two of the three fundamental quadrics are combined, a simple, yet deep, solution of Einstein-Maxwell equations corresponding to a uniform electromagnetic field is obtained with many applications in mathematical physics. One of them is a "no go" theorem: when one tries to frame a Riemannian four-dimensional manifold in a Kahlerian structure, it is found that, while this is generically possible with a definite signature, in spacetime only BR fulfills the requirement. The BR metric plays an important role in the exploration of new principles in fundamental physical theories; we briefly mention some examples related to the horizon of a black hole and the dilaton in string theory., Comment: 30 pages, 4 figures, to appear in the Proceedings of the meeting "Un grande matematico dell '800: omaggio a Eugenio Beltrami" - Milano (14-15 October 2004) published by the Istituto Lombardo di Scienze e Lettere

We study de Rham cohomology for various differential calculi on finite groups G up to order 8. These include the permutation group S_3, the dihedral group D_4 and the quaternion group Q. Poincare' duality holds in every case, and under some assumptions (essentially the existence of a top form) we find that it must hold in general. A short review of the bicovariant (noncommutative) differential calculus on finite G is given for selfconsistency. Exterior derivative, exterior product, metric, Hodge dual, connections, torsion, curvature, and biinvariant integration can be defined algebraically. A projector decomposition of the braiding operator is found, and used in constructing the projector on the space of 2-forms. By means of the braiding operator and the metric a knot invariant is defined for any finite group., Comment: LaTeX, 25 pages, 4 figures. Added higher order exterior basis and volume forms of quaternion and dihedral groups, corrected sign in eq. (2.37) and (2.40), corrected misprints

Dramatic growth in nuclear energy would be required for nuclear power to provide a significant part of the carbon-free energy the world is likely to need in the 21st century, or a major part in meeting other energy challenges. This would require increased support from governments, utilities, and publics around the world. Achieving that support is likely to require improved economics and major progress toward resolving issues of nuclear safety, proliferation-resistance, and nuclear waste management. This is likely to require both research, development, and demonstration (RD&D) of improved technologies and new policy approaches. To gather information on the RD&D needs for the future of nuclear energy, the future cost and performance of nuclear technologies, and on the major barriers to large-scale deployment of nuclear energy, a team of researchers at Harvard University and the Fondazione Eni Enrico Mattei (FEEM) conducted two coordinated surveys of nuclear experts. The surveys asked experts how much they would recommend that their governments spend on nuclear energy RD&D; what progress in cost and performance might be expected by 2030 if those recommendations were followed; and what other factors might constrain or promote future nuclear energy growth. Leading experts from the United States (U.S.) and the European Union (E.U.) participated in this expert elicitation surveys during the summer and fall of 2010. In April 2011, the FEEM and Harvard teams held a workshop in Venice, Italy with a subset of the participating E.U. and U.S. experts to present and discuss the results of the elicitations, in an effort to understand where there is consensus and where the most important disputes and uncertainties lie. Given the Fukushima nuclear accident in Japan, the meeting opened with a discussion of the significance of that event for the future of nuclear power, and of the main lessons learned.