Toward selective quantum advantage in hadronic tomography:explicit cases from Compton form factors, GPDs, TMDs, and GTMDs

This paper argues that quantum advantage in hadronic tomography should be pursued selectively for specific observables like CFFs, GPDs, TMDs, and GTMDs by leveraging quantum algorithms for real-time dynamics, direct matrix element evaluation, and hybrid inference to overcome the limitations of classical Euclidean calculations and sparse experimental data.

The Gist

Imagine you are trying to take a high-resolution 3D photo of a tiny, invisible marble (a proton) that is constantly vibrating and made of even smaller, zipping particles (quarks and gluons). This is the goal of hadronic tomography: creating a detailed map of the inside of matter.

For decades, scientists have tried to do this using powerful classical supercomputers. But the paper you shared argues that for specific types of maps, classical computers are like trying to paint a moving car by only looking at its shadow. They are stuck in the dark.

The authors, Fernando and Keller, propose that quantum computers are the perfect tool for this specific job, not because they are "faster" at everything, but because they speak the same language as the particles themselves.

Here is the breakdown of their argument using simple analogies:

1. The Problem: The "Shadow" vs. The "Real Thing"

Classical computers usually try to solve these problems by working backward from a "shadow" (mathematical data from a frozen moment in time).

• The Analogy: Imagine trying to figure out the exact shape of a spinning top just by looking at a blurry photograph of its shadow on the wall. It's an "ill-posed" problem. There are infinite shapes that could cast that same shadow. It's like trying to guess the recipe of a cake just by tasting the crumbs.
• The Quantum Advantage: Quantum computers don't need to look at the shadow. They can simulate the spinning top while it is spinning. They work in "real-time" and "light-speed" (light-front) just like the particles do. They don't have to guess; they can watch the movie directly.

2. The Four "Special Maps" (The Targets)

The paper says we shouldn't try to use quantum computers for everything. Instead, we should focus on four specific types of maps where quantum computers shine:

• Compton Form Factors (CFFs) & Generalized Parton Distributions (GPDs):

  • What they are: These are like 3D X-rays showing where quarks are and how fast they are moving inside the proton.
  • The Quantum Hook: Extracting these from data is like trying to solve a puzzle where half the pieces are missing and the picture is blurry. The authors show that Quantum Neural Networks (a type of AI running on quantum hardware) are better at filling in the missing gaps and finding the hidden patterns in this noisy data than classical AI.

• TMDs (Transverse Momentum-Dependent Distributions):

  • What they are: These maps show not just where the particles are, but how they are swirling sideways (transverse momentum).
  • The Quantum Hook: This adds a whole new layer of complexity (like adding a third dimension to a 2D drawing). Quantum computers are naturally good at handling these extra layers of "swirl" and direction without getting tangled up.

• GTMDs (Generalized TMDs):

  • What they are: The "Mother of all maps." These combine position, momentum, and spin into a single, incredibly complex picture.
  • The Quantum Hook: These maps are so complex they are best described by "quantum math" (Hilbert spaces). Trying to calculate them on a classical computer is like trying to describe a symphony using only a single piano key. A quantum computer can play the whole symphony at once.

• Real-Time Response:

  • What they are: Watching how a proton reacts right now when hit by a particle, rather than just calculating its static weight.
  • The Quantum Hook: Classical computers struggle with "real-time" physics because it involves complex math that often leads to "sign problems" (where positive and negative numbers cancel each other out, leaving zero useful information). Quantum computers handle these cancellations naturally because they work with waves that interfere, just like the particles do.

3. The Three Types of "Advantage"

The authors break down how quantum computers help into three categories:

1. Algorithmic Advantage: The math says that for these specific tasks, quantum computers are theoretically unbeatable. It's like having a key that fits a lock that a classical computer simply cannot pick.
2. Computational Advantage: It's about efficiency. Instead of taking a long, winding road to calculate a result (classical), the quantum computer takes a shortcut because the calculation matches the hardware's natural way of working.
3. Representational Advantage (The "Hybrid" Approach): This is the most practical near-term idea. Imagine a hybrid team:
   • The Quantum Computer acts as the "Physics Expert." It generates the raw, complex "shape" of the proton based on the laws of physics.
   • The Classical Computer acts as the "Data Analyst." It takes that shape and fits it to the messy, noisy data from real-world experiments (accounting for detector errors, etc.).
   • Why this works: The quantum part handles the hard physics, and the classical part handles the messy real-world data. Together, they solve the puzzle better than either could alone.

4. Why We Need Real Machines (Not Just Simulations)

The paper emphasizes that we can't just simulate these quantum computers on our laptops. We need to run them on real quantum hardware.

• The Analogy: You can simulate a car crash on a computer, but you can't simulate the friction of the tires or the heat of the engine perfectly. Similarly, real quantum computers have "noise" and errors. To know if this technology actually works for physics, we have to test it on the real, noisy machines to see if the signal survives the noise.

The Bottom Line

The paper argues that we shouldn't try to replace all of physics with quantum computers. Instead, we should use them as a specialized tool for "hadronic tomography" (mapping the inside of protons).

By focusing on these specific, complex maps (CFFs, GPDs, TMDs, GTMDs), and using a hybrid approach where quantum computers handle the complex physics and classical computers handle the data fitting, we can finally see inside the proton in a way that was previously impossible. It's not about magic; it's about using the right tool for the right job.

Technical Summary

1. Problem Statement

The paper addresses the challenge of identifying specific, high-value applications for quantum computing within Hadronic Physics (Quantum Chromodynamics or QCD).

• The Core Issue: There is a prevailing misconception that quantum computers should replace classical methods for all QCD calculations. However, classical Lattice QCD already performs exceptionally well for static observables like stable hadron masses.
• The Gap: Many critical hadronic observables—specifically those involving hadronic tomography (imaging the internal 3D structure of hadrons)—are defined by light-front, off-forward, or real-time correlation functions. Extracting these from Euclidean (imaginary-time) lattice calculations or sparse experimental data often constitutes an ill-posed inverse problem.
• The Goal: The authors argue for a "selective" quantum advantage, focusing on observables where the formal definition is "quantum-native" (e.g., real-time dynamics, interference, sign problems) rather than blanket claims of superiority.

2. Methodology

The authors employ a multi-layered analytical framework to categorize and evaluate potential quantum advantages:

• Categorization of Advantage: They distinguish three distinct types of advantage:
  1. Algorithmic: Based on complexity theory (e.g., BQP-completeness of scattering tasks, hardness of the fermion sign problem).
  2. Computational: Based on the ability to measure matrix elements or correlators directly in real-time/light-front frames, bypassing unstable Euclidean reconstruction.
  3. Representational (Inference): Based on the ability of quantum models (e.g., Quantum Deep Neural Networks) to better capture the latent geometry of structured inverse problems compared to classical ansatzes.
• Target Observables: The analysis focuses on four specific classes of observables central to modern hadron imaging:
  • Compton Form Factors (CFFs): Extracted from Deeply Virtual Compton Scattering (DVCS).
  • Generalized Parton Distributions (GPDs): Interpolating between PDFs and form factors.
  • Transverse Momentum-dependent Distributions (TMDs): Including transverse momentum and process-dependent Wilson lines.
  • Generalized TMDs (GTMDs): The most differential correlators, retaining both transverse momentum and off-forward momentum transfer (related to Wigner distributions).
• Hybrid Architecture Proposal: For near-term applications, the authors propose a hybrid workflow where a quantum simulator or circuit provides a physics prior (generating states, correlators, or latent manifolds constrained by QCD symmetries), while a classical neural network handles detector effects, nuisance parameters, and final inference.

3. Key Contributions

A. Formal Justification for Selective Advantage

The paper rigorously argues that hadronic tomography observables are natural quantum targets because:

• Real-Time Dynamics: They are defined by Minkowskian (real-time) correlators, which are difficult to reconstruct from Euclidean data due to the "inverse problem" instability.
• Sign Problems: Many of these observables involve fermion sign problems or open-system evolution (e.g., small-$x$ evolution), which are NP-hard for classical Monte Carlo methods but potentially tractable for quantum simulation.
• High Dimensionality: GTMDs and TMDs involve high-dimensional phase-space structures (position and momentum) that map naturally to Hilbert space descriptions.

B. Specific Case Studies

1. CFF Extraction (Inference Level):
   • The extraction of CFFs from DVCS data is a sparse, noisy, complex-valued regression problem.
   • The authors cite recent studies showing that Quantum Deep Neural Networks (QDNNs) outperform classical networks in these regimes, particularly when data is sparse or noisy, by better matching the latent geometry of the problem.
2. Direct Matrix Element Computation (Computational Level):
   • Instead of fitting reconstructions, quantum devices can directly compute light-front bi-local operators (e.g., $\langle P | \bar{\psi} \gamma^+ W \psi | P \rangle$).
   • This avoids the need for analytic continuation from Euclidean time.
3. Timelike Response and Scattering (Algorithmic Level):
   • The paper highlights the feasibility of simulating timelike response functions (e.g., Hadronic Vacuum Polarization) and scattering amplitudes directly, which are inaccessible to standard lattice QCD.

C. Benchmark Criteria and Hardware Milestones

The authors propose a strict set of criteria for claiming "quantum advantage" in this field, moving beyond toy models:

• Native Workflow: The observable must be real-time, light-front, or open-system.
• Symmetry Awareness: Encodings must preserve gauge invariance and charge sectors natively.
• End-to-End Cost: Benchmarks must include state preparation, compilation, noise mitigation, and readout costs, not just idealized gate counts.
• Physics Payoff: The result must offer improved posterior constraints or direct access to previously indirect correlators.

They summarize recent hardware milestones (2024–2025) that validate this path, including:

• Preparation of interacting mesonic wave packets on Quantinuum H1-1.
• Hadron dynamics in the Schwinger model using 112 qubits on IBM Heron.
• The first hardware extraction of a Parton Distribution Function (PDF) via light-cone correlators.
• Observation of scattering in lattice gauge theories on IBM hardware.

4. Results

• Validation of Selectivity: The analysis confirms that while stable hadron masses are not good near-term targets for quantum advantage, tomography observables (CFFs, GPDs, TMDs, GTMDs) are highly promising.
• Hybrid Viability: The paper demonstrates that hybrid quantum-classical workflows are the most realistic near-term path. Specifically, using quantum circuits to generate physics-constrained priors for classical inference engines yields better performance in noisy, sparse data regimes than purely classical methods.
• Hardware Progress: Recent experiments on IBM and Quantinuum hardware have successfully executed the full operational pipeline for simple hadronic observables (PDFs, scattering), proving that the "state preparation + real-time evolution + measurement" loop is scientifically viable, even if current error rates are high.

5. Significance

• Paradigm Shift: The paper shifts the narrative from "quantum computers will solve all of QCD" to "quantum computers are essential for specific, structurally difficult classes of hadronic observables."
• Roadmap for Hadronic Tomography: It provides a concrete roadmap for the "Hadron Imaging Program," identifying exactly which observables (GTMDs, Wigner distributions) will benefit most from quantum resources due to their phase-space and off-forward nature.
• Scientific Necessity of Real Devices: It argues strongly that classical simulators are insufficient for validating these workflows because they cannot capture the operational realities of noise, leakage, and state preparation costs that define the scientific utility of a quantum device.
• Future Outlook: The work sets the stage for the next generation of experiments at facilities like the Electron-Ion Collider (EIC), suggesting that quantum-enhanced analysis and simulation will be critical for interpreting high-precision tomography data.

In summary, the paper establishes that hadronic tomography is the "cleanest" arena in high-energy physics to demonstrate selective quantum advantage, provided the community adopts rigorous, observable-specific benchmarks and hybrid architectures that leverage the unique strengths of quantum hardware for real-time and high-dimensional problems.