Scalar computational primitives with perturbative phase interferometry Journal: opticajournal Publication type: Research Article

Christopher R. Schwarze, 1, *, Anthony D. Manni, 1 David S. Simon, 1,2 and Alexander V. Sergienko 1,3 · 2025

Read Scalar computational primitives with perturbative phase interferometry Journal: opticajournal Publication type: Research Article by Christopher R. Schwarze, 1, *, Anthony D. Manni, 1 David S. Simon, 1,2 and Alexander V. Sergienko 1,3 (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-2510-25006.txt.

Introduction

Christopher R Schwarze, Anthony D Manni, David S Simon, and Alexander V Sergienko do small math tasks with weak shifts in the phase of plain light. In their rig, glass paths are set in a tweaked line form. A task is done not in a fixed state, but by a move from one state to the next.

That move taps the odd phase map of the rig, so plain linear glass can yield bent math such as splits and powers. In and out are both phase shifts off a set base point, and the out shift is read as a shift in light power. The math is near, not exact, and gets more true as the in shifts shrink. Small inputs buy more true outs.

Catalog record

CREATORChristopher R. Schwarze, 1, *, Anthony D. Manni, 1 David S. Simon, 1,2 and Alexander V. Sergienko 1,3
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE74 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-25006

Browse the full library · About the Tower of Babel

Full text

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

Scalar computational primitives with perturbative phase interferometry Journal: opticajournal Publication type: Research Article Christopher R. Schwarze, 1, *, Anthony D. Manni, 1 David S. Simon, 1,2 and Alexander V. Sergienko 1,3 Abstract We describe how weak phase modulations applied to classical coherent light in specially modified linear interferometers can be used to perform primitive computational tasks. Instead of encoding operations within a fixed unitary state, the operations are enacted by moving from one state to another. This harnesses the particular phase parametrization of an interferometer, allowing entirely linear optics to produce nonlinear operations such as division and powers. This is due to the nonlinear structure of the underlying phase parametrizations. The realized operations are approximate but can be made more accurate by decreasing the size of the input perturbations. For each operation, the inputs and outputs are changes in phase relative to a fixed bias point. The output phase is ultimately read out as a change in optical power. 1 Introduction Alongside modern efforts in traditional quantum computing with photonic qubits, much attention is being directed toward optical computing with classical light sources. Broadly speaking, this analog alternative is thought to be promising due to its potential to increase the speed, bandwidth, and energy efficiency of specific computing tasks relative to digital electronic computers [ 1 ] . A review of modern approaches is given by Ref. [ 2 ] . Certainly light possesses some concrete characteristics that appear amenable to building useful computing systems, such as the interference of amplitudes, the many orthogonal information-carrying modes which can be accessed, and the ability to carry information great distances in minimum time. Nonetheless a number of challenges need to be addressed, such as the programming interface, limitations on realizable operations, accumulation of errors, and information bottlenecks that could drastically hinder the overall compute throughput. A particular class of computations of widespread interest is matrix-vector multiplication [ 3 , 4 , 5 , 6 , 7 , 8 ] in part due to its heavy and ever-increasing role in contemporary approaches to machine learning [ 9 , 10 , 11 , 12 , 13 , 14 ] . Certain optical techniques for matrix-vector multiplication include spatial-light modulation [ 2 , 15 , 16 ] , photonic meshes [ 17 , 18 ] , reprogrammable multimode waveguides [ 19 ] , and photoelectric multiplication [ 20 ] . In many of these approaches, the n\times n matrix transformation U_{0} is obtained by tuning some m internal phases \Phi=(\phi_{1},\phi_{2},\dots,\phi_{m}) of an interferometer to realize the scattering action U(\Phi) . This automatically provides a basis for programmability, but introduces two related complications. The first is that the programming can be nontrivial due to the indirect encoding between the elements of \Phi and the elements of U(\Phi) ; algorithms are often required to load a desired matrix U_{0} using an iterative procedure that uses feedback in the form of measurements of U(\Phi) . The procedure ends when U(\Phi) is sufficiently close to U . Once the phases \Phi are fixed, the computation itself is fast. But when the target U_{0} needs to be changed, the system needs to be placed at a new bias point in \Phi , and the associated reconfiguration time can be costly. U_{0}\approx U(\Phi) is a single point in the parameter space of U spanned by \Phi . Errors in the construction of U and/or selection of a desired \Phi lead to deviations from the obtained matrix U(\Phi) and the desired matrix U_{0} , which in effect produces a different computation than what was targeted. The indirect encoding is also responsible for an undesirable accumulation of errors, as a single \phi_{j} can affect many elements of U_{0} . Another aspect of positioning the system at just one point in the parameter space spanned by U(\Phi) is that it discards the device’s particular parametrization of the phases \Phi in that space, which is due to the graphical structure of the interferometer. For a general network of scatterers forming another aggregate, phase-tunable scatterer, the parametrization is highly dependent on both the properties of the constituent scatterers and their interconnect topology. For a fixed wavelength and polarization, the local geometry of some neighborhood of scattering states in a given interferometer’s parameter space may be understood in terms of a local expansion around some point \Phi_{0} . The expansion coefficients dictate physical quantities such as phase responsivity and the frequency dispersion coefficients. Unlike the individual values of the state amplitudes at a given state (of a fixed frequency and polarization), these quantities are tied to the particular parametrization used. Different wavelengths and/or polarizations will have their own associated parameter spaces which in general can be quite different in character; broadband and non-polarizing devices are designed to drive these differing modal parameter spaces toward coalescence. Certain phase parametrizations have desirable properties which can only be harnessed as a resource for computation if multiple states in a neighborhood of the state space are involved in the same computational procedure. A more elaborate description of this will be provided in the next section. In general, the considered phase modulation technique could be electrical, thermal, optical (via self-phase modulation or cross phase modulation) etc. Crucially, instead of designing computations enacted at a particular point of parameter space, we propose a scheme which enacts computation by a change in state from one point to another in some compact subset of the parameter space. Then, allowing the magnitude of the change to be a continuously varying parameter, the various subsets of the parameter space considered become the grounds of different operations. The proposed optical computing scheme is conceptually compared to the traditional, fixed-state encoding in Fig. 1 . With our proposed encoding, we will illustrate how the four arithmetic primitives can be obtained from the same basic interferometric configuration. The scheme uses entirely linear optics yet is able to produce nonlinear operations such as division and powers due to the nonlinear nature of the parametrizations considered. Although this approach is presently used to produce a variety of scalar operations, it can be extended to matrix and vector operations, which will be left for a future work. Figure 1: A comparison of interferometric encoding schemes. The traditional approach (top) assigns information values to individual points in the parameter space. This ties information to the optical state (i.e. photon number distribution among various spatial modes) but discards the geometric information in the parameter space. The proposed approach (bottom) ties the information to a change in the optical state induced by a change in one or more phases. In coupled-phase interferometers, changing one phase can deform the curve which is defined by spanning another phase. In this encoding approach, the nature of the deformation can now be used to enact certain information processing operations. That is, \Delta P represents a computed output for a given set of inputs (\Delta\phi_{1},\Delta\phi_{2}) . In particular, \Delta P is intended to directly read out the change in phase \Delta\gamma(\Delta\phi_{1},\Delta\phi_{2}) , so all computations are entirely encoded in the optical phase. Examples derived in the text include addition and subtraction (Sec. 3.1), multiplication (Sec. 3.2), nonlinear operations such as squaring or scalar inversion (Sec. 3.3), and cascaded application of these

Read the complete text