Searcharxiv⌕ Search

arXiv subjects

Sourav Pal

Publications and source records attributed to Sourav Pal.

At least 73 records · Page 4Linked to original sources
arXiv2023

Magnetogenesis from Anisotropic Universe

The existence of large-scale anisotropy can not be ruled out by the cosmic microwave background (CMB) radiation. Over the years, several models have been proposed in the context of anisotropic inflation to account for CMB's cold spot and hemispheric asymmetry. However, any small-scale anisotropy, if exists during inflation, is not constrained due to its nonlinear evolution in the subsequent phase. This small-scale anisotropy during inflation can play a non-trivial role in giving rise to the cosmic magnetic field, which is the subject of our present study. Assuming a particular phenomenological form of an anisotropic inflationary universe, we have shown that it can generate a large-scale magnetic field at $1$-Mpc scale with a magnitude $\sim 4\times 10^{-20}~G$, within the observed bound. Because of the anisotropy, the conformal flatness property is lost, and the Maxwell field is generated even without explicit coupling. This immediately resolves the strong coupling problem in the standard magnetogenesis scenario. In addition, assuming very low conductivity during the reheating era, we can further observe the evolution of the electromagnetic field with the equation of state (EoS) $ω_{eff}$ and its effects on the present-day magnetic field.

astro-ph.CO↗
arXiv2023

The $2 \times 2$ block matrices associated with an annulus

A bounded Hilbert space operator $T$ for which the closure of the annulus \[ \mathbb{A}_r=\{z \ : \ r<|z|<1\} \subseteq \mathbb{C}, \qquad (0<r<1) \] is a spectral set is called an $\mathbb A_r$-contraction. A celebrated theorem due to Douglas, Muhly and Pearcy gives a necessary and sufficient condition such that a $2 \times 2$ block matrix of operators $ \begin{bmatrix} T_1 & X 0 & T_2 \end{bmatrix} $ is a contraction. We seek an answer to the same question in the setting of annulus, i.e., under what conditions $\widetilde{T}_Y=\begin{bmatrix} T_1 & Y 0 & T_2 \end{bmatrix} $ becomes an $\mathbb A_r$-contraction. For a pair of $\mathbb A_r$-contractions $T_1,T_2$ and an operator $X$ that commutes with $T_1,T_2$, here we find a necessary and sufficient condition such that each of the block matrices \[ T_X= \begin{bmatrix} T & X 0 & T \end{bmatrix} \,, \quad \widehat{T}_X=\begin{bmatrix} T_1 & X(T_1-T_2) 0 & T_2 \end{bmatrix} \] becomes an $\mathbb A_r$-contraction. Thus, the general block matrix $\widetilde{T}_Y$ is an $\mathbb A_r$-contraction if $T_1-T_2$ (as in $\widehat{T}_X$) is invertible.

math.FA↗
arXiv2023

Next-to-soft radiation from a different angle

Soft and collinear radiation in collider processes can be described in a universal way, that is independent of the underlying process. Recent years have seen a number of approaches for probing whether radiation beyond the leading soft approximation can also be systematically classified. In this paper, we study a formula that captures the leading next-to-soft QCD radiation affecting processes with both final- and initial-state partons, by shifting the momenta in the non-radiative squared amplitude. We first examine W+jet production, and show that a previously derived formula of this type indeed holds in the case in which massive colour singlet particles are present in the final state. Next, we develop a physical understanding of the momentum shifts, showing precisely how they disrupt the well-known angular ordering property of leading soft radiation.

hep-ph↗
arXiv2023

Dilation and Birkhoff-James orthogonality

We study the interaction between unitary $ρ$-dilations of a pair of Hilbert space operators and Birkhoff-James orthogonality. We prove that for two orthogonal operators $T,A$ if $\|T\|=ρ$, then $U_T \perp_B U_A$ for any unitary $ρ$-dilations $U_T$ of $T$ and $U_A$ of $A$ acting on a common space. We characterize $\varepsilon$-approximate Birkhoff-James orthogonality for complex Hilbert space operators. Then find a sharp bound on $\varepsilon$ such that $T \perp_B A$ implies that $U_T \perp_B^{\varepsilon} U_A$ for any unitary $ρ$-dilations $U_T, U_A$ of $T$ and $A$ respectively. The Schäffer unitary dilations of a pair of contractions $T,A$ are not orthogonal in general. We construct Schäffer-type unitary dilations for contractions which are pairwise orthogonal.

math.FA↗
arXiv2023

Distinguished varieties in the polydisc and dilation of commuting contractions

A distinguished variety in the polydisc $\mathbb D^n$ is an affine complex algebraic variety that intersects $\mathbb D^n$ and exits the domain through the $n$-torus $\mathbb T^n$ without intersecting any other part of the topological boundary of $\mathbb D^n$. We find two different characterizations for a distinguished variety in the polydisc $\mathbb D^n$ in terms of the Taylor joint spectrum of certain linear matrix-pencils and thus generalize the seminal work due to Agler and M\raise.45ex\hbox{c}Carthy [Acta Math., 2005] on distinguished varieties in $\mathbb D^2$. We show that a distinguished variety in $\mathbb D^n$ is a part of an affine algebraic curve which is a set-theoretic complete intersection. We also show that if $(T_1, \dots , T_n)$ is commuting tuple of Hilbert space contractions such that the defect space of $T=\prod_{i=1}^n T_i$ is finite dimensional, then $(T_1, \dots , T_n)$ admits a commuting unitary dilation $(U_1, \dots , U_n)$ with $U=\prod_{i=1}^n U_i$ being the minimal unitary dilation of $T$ if and only if some certain matrices associated with $(T_1, \dots , T_n)$ define a distinguished variety in $\mathbb D^n$.

math.AG↗
arXiv2023

Dilation of normal operators associated with an annulus

For $0<r<1$, let us consider the following annulus: \[ \mathbb A_r= \{ z\in \mathbb C\, : \, r<|z|<1 \}. \] A Hilbert space operator $T$ for which $\overline{\mathbb A}_r$ is a spectral set is called an $\mathbb A_r$-\textit{contraction}. Also, a normal operator $U$ whose spectrum lies on the boundary $\partial \mathbb A_r$ of $\mathbb A_r$ is called an $\mathbb A_r$-\textit{unitary}. We prove that any $m$ number of commuting normal $\mathbb A_r$-contractions $N_1, \dots , N_m$ can be simultaneously dilated to commuting $\mathbb A_r$-unitaries $U_1, \dots , U_m$. To construct such a dilation, we solve a Dirichlet problem for the polyannulus $\mathbb A_r^m$. Also, we show that any finitely many doubly commuting subnormal $\mathbb A_r$-contractions simultaneously dilate to commuting $\mathbb A_r$-unitaries. Finally, we show that such a simultaneous $\mathbb A_r$-unitary dilation holds for any finite number of doubly commuting $2 \times 2$ scalar $\mathbb A_r$-contractions.

math.FA↗
arXiv2023

Deciphering Colour Building Blocks of Massive Multiparton Amplitudes at 4-loops and beyond

The soft function in non-abelian gauge theories exponentiate, and their logarithms can be organised in terms of the collections of Feynman diagrams called Cwebs. The colour factors that appear in the logarithm are controlled by the web mixing matrices. Direct construction of the diagonal blocks of Cwebs using the new concepts of Normal ordering, basis Cweb and Fused-Web was recently carried out in~\cite{Agarwal:2022wyk}. In this article we establish correspondence between the boomerang webs introduced in ~\cite{Gardi:2021gzz} and non-boomerang Cwebs. We use this correspondence together with Uniqueness theorem and Fused web formalism introduced in ~\cite{Agarwal:2022wyk} to obtain the diagonal blocks of four general classes of Cwebs to all orders in perturbation theory which also cover all the four loop Boomerang Cwebs connecting four Wilson lines. We also fully construct the mixing matrix of a special Cweb to all orders in perturbation theory.

hep-ph↗
arXiv2022

Lifting $Q$-commuting and $Q$-intertwining operators

There are several proofs of the classical commutant lifting and intertwining lifting theorems in the literature. In this article, we present analogous proofs to a few $Q$-commuting lifting and $Q$-intertwining lifting theorems. We provide several proofs and show explicit constructions of Ando-type dilations for a pair of $Q$-commuting contractions $(T_1,T_2)$ when $Q$ is a bounded operator. We establish a few connections of these results with graph theory.

math.FA↗
arXiv2022

Probing Non-classicality of Primordial Gravitational Waves and Magnetic Field Through Quantum Poincare Sphere

The universe is believed to be originated from a quantum state. However, defining measurable quantities for the quantum properties in the present universe has gained interest recently. In this submission, we propose a quantum Poincare sphere as an observable quantity that can hint at the quantumness of primordial gravitational waves and large-scale magnetic fields. The Poincare sphere is defined in terms of quantum stokes operators associated with the polarization of those fields, which can be measured directly. We have further studied the effects of the initial non-BD vacuum on the power spectrum and squeezing parameter of the primordial gravitational waves and magnetic field. We have found that the initial non-BD vacuum increases the value of the squeezing parameter as expected at the end of inflation, which further enhances the possibility of measuring the quantumness of the fields under consideration. To support our results, we further explored the possible Bell violation test for a set of generalized pseudo spin operators defined in the polarization space of those fields.

gr-qc↗
arXiv2022

Multi Resolution Analysis (MRA) for Approximate Self-Attention

Transformers have emerged as a preferred model for many tasks in natural langugage processing and vision. Recent efforts on training and deploying Transformers more efficiently have identified many strategies to approximate the self-attention matrix, a key module in a Transformer architecture. Effective ideas include various prespecified sparsity patterns, low-rank basis expansions and combinations thereof. In this paper, we revisit classical Multiresolution Analysis (MRA) concepts such as Wavelets, whose potential value in this setting remains underexplored thus far. We show that simple approximations based on empirical feedback and design choices informed by modern hardware and implementation challenges, eventually yield a MRA-based approach for self-attention with an excellent performance profile across most criteria of interest. We undertake an extensive set of experiments and demonstrate that this multi-resolution scheme outperforms most efficient self-attention proposals and is favorable for both short and long sequences. Code is available at \url{https://github.com/mlpen/mra-attention}.

cs.LG↗
arXiv2022

Deep Unlearning via Randomized Conditionally Independent Hessians

Recent legislation has led to interest in machine unlearning, i.e., removing specific training samples from a predictive model as if they never existed in the training dataset. Unlearning may also be required due to corrupted/adversarial data or simply a user's updated privacy requirement. For models which require no training (k-NN), simply deleting the closest original sample can be effective. But this idea is inapplicable to models which learn richer representations. Recent ideas leveraging optimization-based updates scale poorly with the model dimension d, due to inverting the Hessian of the loss function. We use a variant of a new conditional independence coefficient, L-CODEC, to identify a subset of the model parameters with the most semantic overlap on an individual sample level. Our approach completely avoids the need to invert a (possibly) huge matrix. By utilizing a Markov blanket selection, we premise that L-CODEC is also suitable for deep unlearning, as well as other applications in vision. Compared to alternatives, L-CODEC makes approximate unlearning possible in settings that would otherwise be infeasible, including vision models used for face recognition, person re-identification and NLP models that may require unlearning samples identified for exclusion. Code can be found at https://github.com/vsingh-group/LCODEC-deep-unlearning/

cs.CV↗
arXiv2022

Building blocks of Cwebs in multiparton scattering amplitudes

The correlators of Wilson-line operators in non-abelian gauge theories are known to exponentiate, and their logarithms can be organised in terms of the collections of Feynman diagrams called Cwebs. The colour factors that appear in the logarithm correspond to completely connected diagrams and are determined by the web mixing matrices. In this article we introduce several new concepts: (a) Normal ordering of the diagrams of a Cweb, (b) Fused-Webs (c) Basis and Family of Cwebs. We use these ideas together with a Uniqueness theorem that we prove to arrive at an understanding of the diagonal blocks, and several null matrices that appear in the mixing matrices. We demonstrate using our formalism that, once the basis Cwebs present upto order $α_{s}^{n}$ are determined, the number of exponentiated colour factors for several classes of Cwebs starting at order $α_{s}^{n+1}$ can be predicted. We further provide complete results for the mixing matrices, to all orders in perturbation theory, for two special classes of Cwebs using our framework.

hep-ph↗
arXiv2022

Common reducing subspaces and decompositions of contractions

A commuting triple of Hilbert space operators $(A,B,P)$, for which the closed tetrablock $\overline{\mathbb E}$ is a spectral set, is called a \textit{tetrablock-contraction} or simply an $\mathbb E$-\textit{contraction}, where \[ \mathbb E=\{(x_1,x_2,x_3)\in \mathbb C^3:\, 1-x_1z-x_2w+x_3zw \neq 0 \quad \text{ whenever } \; |z|\leq 1, \; \; |w|\leq 1 \} \subset \mathbb C^3, \] is a polynomially convex domain which is naturally associated with the $μ$-synthesis problem. By applications of the theory of $\mathbb E$-contractions, we obtain several results on decompositions of contractions.

math.FA↗
arXiv2022

Triangular Tetrablock-contractions, factorization of contractions, dilation and subvarieties

A commuting triple of Hilbert space operators $(A,B,P)$, for which the closed tetrablock $\bar{\mathbb E}$ is a spectral set, is called a \textit{tetrablock-contraction} or simply an $\mathbb E$-\textit{contraction}, where \[ \mathbb E=\{(a_{11},a_{22}, \det A):\, A=[a_{ij}]\in \mathcal M_2(\mathbb C), \; \|A\| <1 \} \subset \mathbb C^3 \] is a polynomially convex domain which is naturally associated with the $μ$-synthesis problem. We introduce triangular $\mathbb E$-contractions and prove that every pure triangular $\mathbb E$-contraction dilates to a pure triangular $\mathbb E$-isometry. We construct a functional model for a pure triangular $\mathbb E$-isometry and apply that model to find a new proof for the famous Berger-Coburn-Lebow Model Theorem for commuting isometries. Next we give an alternative proof to the more generalized version of Berger-Coburn-Lebow Model, namely the factorization of a pure contraction due to Das, Sarkar and Sarkar (\textit{Adv. Math.} 322 (2017), 186 -- 200). We find a necessary and sufficient condition for the existence of $\mathbb E$-unitary dilation of an $\mathbb E$-contraction $(A,B,P)$ on the smallest dilation space and show that it is equivalent to the existence of a distinguished variety in $\mathbb E$ when the defect space $D_{P^*}$ is finite dimensional.

math.FA↗
arXiv2022

A Nagy-Foias program for a c.n.u. $Γ_n$-contraction

A tuple of commuting Hilbert space operators $(S_1, \dots, S_{n-1}, P)$ having the closed symmetrized polydisc \[ Γ_n = \left\{ \left(\sum_{i=1}^{n}z_i, \sum\limits_{1\leq i<j\leq n} z_iz_j, \cdots, \prod_{i=1}^{n}z_i\right) : |z_i|\leq 1\,, \; \; \; 1\leq i \leq n-1 \right\} \] as a spectral set is called a $Γ_n$-contraction. From the literature we have that a point $(s_1, \dots , s_{n-1},p)$ in $Γ_n$ can be represented as $s_i=c_i+pc_{n-i}$ for some $(c_1, \dots, c_{n-1}) \in Γ_{n-1}$. We construct a minimal $Γ_n$-isometric dilation for a particular class of c.n.u. $Γ_n$-contractions $(S_1, \cdots, S_{n-1},P)$ and obtain a functional model for them. With the help of this model we express each $S_i$ as $S_i=C_i+PC_{n-i}$, which is an operator theoretic analogue of the scalar result. We also produce an abstract model for a different class of c.n.u. $Γ_n$-contractions satisfying $S_i^*P=PS_i^*$ for each $i$. By exhibiting a counter example we show that such abstract model may not exist if we drop the hypothesis that $S_i^*P=PS_i^*$. We apply this abstract model to achieve a complete unitary invariant for such c.n.u. $Γ_n$-contractions. Additionally, we present different necessary conditions for dilation and a sufficient condition under which a commuting tuple $(S_1, \dots , S_{n-1},P)$ becomes a $Γ_n$-contraction. The entire program goes parallel to the operator theoretic program developed by Sz.-Nagy and Foias for a c.n.u. contraction.

math.FA↗
arXiv2022

On $q$-commuting co-extensions and $q$-commutant lifting

Consider a nonzero contraction $T$ and a bounded operator $X$ satisfying $TX=qXT$ for a complex number $q$. There are some interesting results in the literature on $q$-commuting dilation and $q$-commutant lifting of such pair $(T,X)$ when $|q|=1$. Here we improve a few of them to the class of scalars $q$ satisfying $|q|\leq \dfrac{1}{\|T\|}$.

math.FA↗