Spatiotemporal control of laser intensity using differentiable programming Publication type: Paper

Kyle G Miller 1,∗ \orcid 0000-0003-4826-9001, Tomas E Gutierrez 1 \orcid 0009-0003-7966-719X, Archis S Joglekar 2,1,3 \orcid 0000-0003-3599-5629, Amanda Elliott 1 \orcid 0009-0006-0020-8591, Dustin H Froula 1 \orcid 0000-0001-6981-3956, and John P Palastro 1 \orcid 0000-0002-6721-1924 · 2025

Read Spatiotemporal control of laser intensity using differentiable programming Publication type: Paper by Kyle G Miller 1,∗ \orcid 0000-0003-4826-9001, Tomas E Gutierrez 1 \orcid 0009-0003-7966-719X, Archis S Joglekar 2,1,3 \orcid 0000-0003-3599-5629, Amanda Elliott 1 \orcid 0009-0006-0020-8591, Dustin H Froula 1 \orcid 0000-0001-6981-3956, and John P Palastro 1 \orcid 0000-0002-6721-1924 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-25673.txt.

Introduction

Kyle G Miller led five co-authors controlling laser intensity. Tomas E Gutierrez, Archis S Joglekar, and Amanda Elliott helped. Dustin H Froula and John P Palastro complete six. Their 2025 paper uses differentiable programming for pulses. Optical techniques produce pulses with custom form. Structured pulses usually follow forward design methods.

Inverse methods enable discovery of new pulse forms. Superior results may result from such discovery. Gradient-based tuning designs structured pulses from goals. Uniform longitudinal intensity over extended regions is first. A superluminal intensity peak traveling far is second. It keeps constant duration, spot size, and amplitude. A pulse ionizing gas into uniform plasma is third. Tuning full spatiotemporal form better results fifteenfold. Spatial or time tuning alone performed worse.

Catalog record

CREATORKyle G Miller 1,∗ \orcid 0000-0003-4826-9001, Tomas E Gutierrez 1 \orcid 0009-0003-7966-719X, Archis S Joglekar 2,1,3 \orcid 0000-0003-3599-5629, Amanda Elliott 1 \orcid 0009-0006-0020-8591, Dustin H Froula 1 \orcid 0000-0001-6981-3956, and John P Palastro 1 \orcid 0000-0002-6721-1924
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-25673

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Spatiotemporal control of laser intensity using differentiable programming Publication type: Paper Kyle G Miller 1,∗ \orcid 0000-0003-4826-9001, Tomas E Gutierrez 1 \orcid 0009-0003-7966-719X, Archis S Joglekar 2,1,3 \orcid 0000-0003-3599-5629, Amanda Elliott 1 \orcid 0009-0006-0020-8591, Dustin H Froula 1 \orcid 0000-0001-6981-3956, and John P Palastro 1 \orcid 0000-0002-6721-1924 Abstract Optical techniques for spatiotemporal control can produce laser pulses with custom amplitude, phase, or polarization structure. In nonlinear optics and plasma physics, the use of structured pulses typically follows a forward design approach, in which the efficacy of a known structure is analyzed for a particular application. Inverse approaches, in contrast, enable the discovery of new structures with the potential for superior performance. Here, an implementation of the unidirectional pulse propagation equation that supports automatic differentiation is combined with gradient-based optimization to design structured pulses with features that are advantageous for a range of nonlinear optical and plasma-based applications: (1) a longitudinally uniform intensity over an extended region, (2) a superluminal intensity peak that travels many Rayleigh ranges with constant duration, spot size, and amplitude, and (3) a laser pulse that ionizes a gas to form a uniform column of plasma. In the final case, optimizing the full spatiotemporal structure improves the performance by a factor of 15 compared to optimizing only spatial or only temporal structure, highlighting the advantage of spatiotemporal control. keywords automatic differentiation, gradient-based optimization, nonlinear optics 1 Introduction Laser pulses structured in space, time, or coupled space–time provide additional degrees of freedom for optimizing laser-based applications. The spatial structure can be two-dimensional, as in pulses with constant orbital angular momentum (OAM) [ 1 , 2 ] , or three-dimensional, as in pulses with tailored trajectories [ 3 ] or spatially varying OAM [ 4 , 5 ] . Common examples of temporal structure include pulse shape, chirp, or higher-order temporal and spectral phase. Coupled space–time structure offers the greatest flexibility and enables custom evolution of the amplitude, phase, and polarization [ 6 , 7 , 8 , 9 ] . A prominent realization is the “flying focus,” which features a peak intensity that moves independently of the group velocity over many Rayleigh ranges [ 10 , 11 , 12 , 13 , 14 ] . While known structures like these present new opportunities to enhance applications [ 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ] , they are somewhat limited in their adaptability. The optimal structure for a particular application may combine existing structures or take an entirely new form. The optimization of laser-based applications with structured pulses has typically followed a forward design approach, in which a known structure is evaluated for its ability to improve an outcome such as efficiency or yield. The flying focus, for instance, has found utility in plasma and nonlinear optical processes that require velocity matching or extended interaction lengths [ 11 , 19 , 20 , 21 , 22 ] . Inverse design, by contrast, begins with a desired outcome and seeks a structure that maximizes or minimizes that outcome. This approach can yield finely tuned optimizations of existing structures or reveal undiscovered structures. High-dimensional inverse design is particularly well suited to machine learning techniques for gradient-based optimization, where optimal input parameters are obtained by minimizing a loss function that is chosen to produce a desired outcome. Advances in software tools and specialized computing architectures have made gradient-based optimization increasingly accessible to the scientific community. As an example, the JAX [ 23 ] library extends the NumPy [ 24 ] scientific library in Python by adding support for automatic differentiation (AD) and acceleration on graphical processing units (GPUs), features that can substantially speed up gradient-based optimization. This is especially true for inverse design problems that involve many inputs and few outputs, where computing finite-difference gradients is expensive [ 25 , 26 ] . Reverse-mode AD, in particular, efficiently computes gradients with respect to many input parameters: gradients are calculated at each step during forward evaluation and then accumulated in a reverse pass of the program—one pass for each output [ 27 , 26 ] . This mode of AD accelerates inverse design in scientific software while adding minimal computational and programming overhead. These advantages have motivated the adoption of AD across diverse fields, including optics [ 28 , 29 , 30 , 31 ] , astronomy [ 32 , 33 ] , and plasma physics [ 34 , 35 , 36 , 37 ] . Here, we present a differentiable implementation of the unidirectional pulse propagation equation (UPPE) [ 38 , 39 , 40 ] that enables inverse design optimization of near-field spatiotemporal structure for desired features in the far field. The optimization is performed for three examples with utility across a range of nonlinear optics and plasma physics applications. In each example, the parameters of the laser pulse in the near field are allowed to vary using a gradient descent algorithm. The propagation of the pulse in the far field is then simulated either in vacuum or in a nonlinear medium, and the gradients are computed using AD within the JAX framework. The three examples build in complexity: (1) production of a static, constant intensity over an extended focal region, (2) generation of an intensity peak that travels at a prescribed velocity while maintaining a constant spot size, duration, and amplitude, and (3) formation of an extended, uniform region of photoionized plasma. The first example uses pure spatial optimization, while the latter two leverage full spatiotemporal optimization. In the final example, the desired features can only be realized with coupled space–time structure; neither spatial nor temporal structuring alone is sufficient to achieve the desired outcome. 2 Methods 2.1 Near-field parameterization Consider a laser pulse incident on an optical assembly in the near field that can freely manipulate the spatial and temporal structure of the amplitude and phase. The goal is to optimize the structure to produce desired features in the far field. The performance of the structure can be evaluated in terms of a metric, or loss function \mathcal{L} , that approaches a minimum as the desired features are obtained. This constitutes an inverse design problem that can be addressed with a gradient descent algorithm, where the pulse parameters are iteratively adjusted according to derivatives of the loss function with respect to those parameters (Fig. 1 ). The solution requires a framework for parameterizing the amplitude and phase that is both general enough to represent a wide variety of spatiotemporal structures and specific enough to ensure effective tuning by the optimization algorithm. Figure 1: Diagram of the optimization loop in the differentiable program SUPER-JAX. The spatiospectral structure of the near-field electric field \hat{E}e^{i\hat{\phi}} is iteratively tuned via gradients of a loss function \mathcal{L} to find an optimized pulse that produces the desired far-field behavior. Far-field propagation is simulated with the unidirectional pulse propagation equation, and gradients of \mathcal{L} are calculated with automatic differentiation for each tunable parameter \hat{x} . At the exit of the optical assembly, the pulse propagates in the positive \vec{z} direction and is assumed to be cylindrically symmetric. In this configuration, the most general way to describe the structure of the amplitude and phase is to specify their values

Read the complete text