arXiv subjects

Shahriar Iravanian

Publications and source records attributed to Shahriar Iravanian.

4 recordsLinked to original sources

arXiv2026

pyAmpliCol: fast tree-level matrix elements

Tree-level matrix elements sit at the heart of collider event generation, but their cost rises rapidly just where additional resolved radiation becomes most important. We present pyAmpliCol, a public generator aimed at extending the practical reach of colour-resolved matrix elements to higher multiplicities. It builds on the colour-ordered, off-shell-current strategy of AmpliCol, providing fast access to individual leading-colour flows while retaining the option of next-to-leading or exact full colour. The latter introduces its own high-multiplicity bottleneck: the dense interference of factorially many colour orderings. pyAmpliCol addresses this contraction with a fast Fourier transform on the symmetric group, preserving the exact colour result while avoiding the direct pairwise sum. Representative full-colour benchmarks demonstrate substantial speedups over the reference calculations considered here. It supports Standard Model processes and a broad class of UFO models, selected or summed helicities, vectorised evaluation and controllable numerical precision. It further provides colour- and spin-correlated Born quantities and tree-level colour connections at arbitrary perturbative order, opening the same framework to local subtraction and other higher-order applications. Together, these capabilities connect high-multiplicity matrix elements to their use in event generation, merging, parton showers and precision calculations within a single generator.

hep-ph↗

arXiv2024

Integer Polynomial Factorization by Recombination of Real Factors: Re-evaluating an Old Technique in Modern Era

Polynomial factorization over $ZZ$ is of great historical and practical importance. Currently, the standard technique is to factor the polynomial over finite fields first and then to lift to integers. Factorization over finite fields can be done in polynomial time using Berlekamp or Cantor-Zassenhaus algorithms. Lifting from the finite field to $ZZ$ requires a combinatorial algorithm. The van Hoeij algorithm casts the combinatorial problem as a knapsack-equivalent problem, which is then solved using lattice-reduction (the LLL algorithm) in polynomial time, which is implemented in many computer algebra systems (CAS). In this paper, we revisit the old idea of starting with factorization over $RR$ instead of a finite field, followed by recombination of the resulting linear and quadratic factors. We transform the problem into an integer subset sum problem, which is then solved using the Horowizt-Sinha algorithm. This algorithm can factor a random integer polynomial of degree $d$ in a time complexity of $O(2^(d slash 4))$. While the resulting algorithm is exponential, consistent with the integer subset sum problem being in NP, it has a few advantages. First, it is simple and easy to implement. Second, it is almost embarrassingly parallelizable. We demonstrate this by implementing the algorithm in a Graphic Processing Unit (GPU). The resulting code can factor a degree 100 polynomial is a few tenths of a second, comparable to some standard CAS. This shows that it is possible to use current hardware, especially massively parallel systems like GPU, to the benefit of symbolic algebra.

cs.SC↗

arXiv2022

Symbolic-Numeric Integration of Univariate Expressions based on Sparse Regression

Most computer algebra systems (CAS) support symbolic integration as core functionality. The majority of the integration packages use a combination of heuristic algebraic and rule-based (integration table) methods. In this paper, we present a hybrid (symbolic-numeric) methodology to calculate the indefinite integrals of univariate expressions. The primary motivation for this work is to add symbolic integration functionality to a modern CAS (the symbolic manipulation packages of SciML, the Scientific Machine Learning ecosystem of the Julia programming language), which is mainly designed toward numerical and machine learning applications and has a different set of features than traditional CAS. The symbolic part of our method is based on the combination of candidate terms generation (borrowed from the Homotopy operators theory) with rule-based expression transformations provided by the underlying CAS. The numeric part is based on sparse-regression, a component of Sparse Identification of Nonlinear Dynamics (SINDy) technique. We show that this system can solve a large variety of common integration problems using only a few dozen basic integration rules.

cs.SC↗

arXiv2020

Discovery of the Hidden State in Ionic Models Using a Domain-Specific Recurrent Neural Network

Ionic models, the set of ordinary differential equations (ODEs) describing the time evolution of the state of excitable cells, are the cornerstone of modeling in neuro- and cardiac electrophysiology. Modern ionic models can have tens of state variables and hundreds of tunable parameters. Fitting ionic models to experimental data, which usually covers only a limited subset of state variables, remains a challenging problem. In this paper, we describe a recurrent neural network architecture designed specifically to encode ionic models. The core of the model is a Gating Neural Network (GNN) layer, capturing the dynamics of classic (Hodgkin-Huxley) gating variables. The network is trained in two steps: first, it learns the theoretical model coded in a set of ODEs, and second, it is retrained on experimental data. The retrained network is interpretable, such that its results can be incorporated back into the model ODEs. We tested the GNN networks using simulated ventricular action potential signals and showed that it could deduce physiologically-feasible alterations of ionic currents. Such domain-specific neural networks can be employed in the exploratory phase of data assimilation before further fine-tuning using standard optimization techniques.

cs.LG↗