Searcharxiv⌕ Search

arXiv subjects

Stephen Mann

Publications and source records attributed to Stephen Mann.

7 recordsLinked to original sources
arXiv2025

Nonlinear Noise2Noise for Efficient Monte Carlo Denoiser Training

The Noise2Noise method allows for training machine learning-based denoisers with pairs of input and target images where both the input and target can be noisy. This removes the need for training with clean target images, which can be difficult to obtain. However, Noise2Noise training has a major limitation: nonlinear functions applied to the noisy targets will skew the results. This bias occurs because the nonlinearity makes the expected value of the noisy targets different from the clean target image. Since nonlinear functions are common in image processing, avoiding them limits the types of preprocessing that can be performed on the noisy targets. Our main insight is that certain nonlinear functions can be applied to the noisy targets without adding significant bias to the results. We develop a theoretical framework for analyzing the effects of these nonlinearities, and describe a class of nonlinear functions with minimal bias. We demonstrate our method on the denoising of high dynamic range (HDR) images produced by Monte Carlo rendering. Noise2Noise training can have trouble with HDR images, where the training process is overwhelmed by outliers and performs poorly. We consider a commonly used method of addressing these training issues: applying a nonlinear tone mapping function to the model output and target images to reduce their dynamic range. This method was previously thought to be incompatible with Noise2Noise training because of the nonlinearities involved. We show that certain combinations of loss functions and tone mapping functions can reduce the effect of outliers while introducing minimal bias. We apply our method to an existing machine learning-based Monte Carlo denoiser, where the original implementation was trained with high-sample count reference images. Our results approach those of the original implementation, but are produced using only noisy training data.

cs.CV↗
arXiv2023

Molecular tuning of DNA framework-programmed silicification by cationic silica cluster attachment

The organizational complexity of biominerals has long fascinated scientists seeking to understand biological programming and implement new developments in biomimetic materials chemistry. Nonclassical crystallization pathways have been observed and analyzed in typical crystalline biominerals, involving the controlled attachment and reconfiguration of nanoparticles and clusters on organic templates. However, the understanding of templated amorphous silica mineralization remains limited, hindering the rational design of complex silica-based materials. Here, we present a systematic study on the stabilization of self-capping cationic silica cluster (CSC) and their assembly dynamics using DNA nanostructures as programmable attachment templates. By tuning the composition and structure of CSC, we demonstrate high-fidelity silicification at single-cluster resolution, revealing a process of adaptive templating involving cooperative adjustments of both the DNA framework and cluster morphology. Our results provide a unified model of silicification by cluster attachment and pave the way towards the molecular tuning of pre- and post-nucleation stages of sol-gel reactions. Overall, our findings provide new insights for the design of silica-based materials with controlled organization and functionality, bridging the gap between biomineralization principles and the rational design of biomimetic material.

physics.chem-ph↗
arXiv2022

Faithfulness of generalised Verma modules for Iwasawa algebras

We prove faithfulness of infinite-dimensional generalised Verma modules for Iwasawa algebras corresponding to split simple Lie algebras with a Chevalley basis. We use this to prove faithfulness of all infinite-dimensional highest-weight modules in the case of type $A_{2}$. In this case we also show that all prime ideals of the corresponding Iwasawa algebras are annihilators of finite-dimensional simple modules.

math.RT↗
arXiv2019

On the Clifford Algebraic Description of the Geometry of a 3D Euclidean Space

We discuss how transformations in a three dimensional euclidean space can be described in terms of the Clifford algebra $\mathcal{C}\ell_{3,3}$ of the quadratic space $\mathbb{R}^{3,3}$. We show that this algebra describes in a unified way the operations of reflection, rotations (circular and hyperbolic), translation, shear and non-uniform scale. Moreover, using the concept of Hodge duality, we define an operation called cotranslation, and show that the operation of perspective projection can be written in this Clifford algebra as a composition of the translation and cotranslation operations. We also show that the operation of pseudo-perspective can be implemented using the cotranslation operation. An important point is that the expression for the operations of reflection and rotation in $\mathcal{C}\ell_{3,3}$ preserve the subspaces that can be associated with the algebras $\mathcal{C}\ell_{3,0}$ and $\mathcal{C}\ell_{0,3}$, so that reflection and rotation can be expressed in terms of $\mathcal{C}\ell_{3,0}$ or $\mathcal{C}\ell_{0,3}$, as well-known. However, all other operations mix those subspaces in such a way that they need to be expressed in terms of the full Clifford algebra $\mathcal{C}\ell_{3,3}$. An essential aspect of our formulation is the representation of points in terms of objects called paravectors. Paravectors have been used previously to represents points in terms of an algebra closely related to the Clifford algebra $\mathcal{C}\ell_{3,3}$. We compare these different approaches.

math.GM↗
arXiv2018

Paravectors and the Geometry of 3D Euclidean Space

We introduce the concept of paravectors to describe the geometry of points in a three dimensional space. After defining a suitable product of paravectors, we introduce the concepts of biparavectors and triparavectors to describe line segments and plane fragments in this space. A key point in this product of paravectors is the notion of the orientation of a point, in such a way that biparavectors representing line segments are the result of the product of points with opposite orientations. Incidence relations can also be formulated in terms of the product of paravectors. To study the transformations of points, lines, and planes, we introduce an algebra of transformations that is analogous to the algebra of creation and annihilation operators in quantum theory. The paravectors, biparavectors and triparavectors are mapped into this algebra and their transformations are studied; we show that this formalism describes in an unified way the operations of reflection, rotations (circular and hyperbolic), translation, shear and non-uniform scale transformation. Using the concept of Hodge duality, we define a new operation called cotranslation, and show that the operation of perspective projection can be written as a composition of the translation and cotranslation operations. We also show that the operation of pseudo-perspective can be implemented using the cotranslation operation.

math.GM↗
arXiv2014

Multimodal Plasmonics in Fused Colloidal Networks

Harnessing the optical properties of noble metals down to the nanometer-scale is a key step towards fast and low-dissipative information processing. At the 10-nm length scale, metal crystallinity and patterning as well as probing of surface plasmon (SP) properties must be controlled with a challenging high level of precision. Here, we demonstrate that ultimate lateral confinement and delocalization of SP modes are simultaneously achieved in extended self-assembled networks comprising linear chains of partially fused gold nanoparticles. The spectral and spatial distributions of the SP modes associated with the colloidal superstructures are evidenced by performing monochromated electron energy loss spectroscopy with a nanometer-sized electron probe. We prepare the metallic bead strings by electron beam-induced interparticle fusion of nanoparticle networks. The fused superstructures retain the native morphology and crystallinity but develop very low energy SP modes that are capable of supporting long range and spectrally tunable propagation in nanoscale waveguides.

physics.optics↗