Analytic parametrisations of the scale dependence of proton parton distributions
This paper presents an analytic parametrization of proton parton distribution functions (PDFs) derived from HERAPDF2.0 using machine-assisted fitting, which accurately reproduces integrated distributions and cross sections while significantly reducing Monte Carlo computation time compared to traditional grid interpolation.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer
Inside every proton, the fundamental building block of atomic nuclei, there is a chaotic, seething world of smaller particles called quarks and gluons. These particles do not sit still; they zip around, carrying the proton's momentum and constantly interacting with one another. To understand how protons behave when smashed together at the highest energies, such as in the massive particle accelerators used by physicists, scientists rely on a map known as the parton distribution function. This map tells them the probability of finding a specific type of particle carrying a certain fraction of the proton's speed at a given moment. However, this map is not static. As the energy of the collision changes, the distribution of these particles shifts and evolves. Calculating this evolution is a complex mathematical task that usually requires computers to look up pre-calculated tables and guess the values in between, a process that is accurate but slow.
A researcher has now proposed a different way to handle this problem, one that replaces the slow look-up tables with a direct, mathematical description. By treating the parameters that define the particle distribution as variables that change smoothly with energy, they created a new, analytic formula. This approach allows computers to calculate the state of the proton's interior instantly, without needing to pause and interpolate between data points. The result is a method that is not only faster but also maintains a high degree of accuracy, offering a more efficient tool for simulating the collisions that reveal the deepest secrets of matter.
In the world of high-energy physics, predicting the outcome of a collision between two protons is like trying to forecast the weather in a storm that never stops. The collision itself is governed by the laws of quantum mechanics, but the ingredients—the quarks and gluons inside the protons—are messy and difficult to pin down. Scientists use a framework called collinear factorisation to separate the messy, non-perturbative behavior of the proton's interior from the clean, calculable physics of the collision. The messy part is described by the parton distribution functions, which are determined by fitting experimental data at a specific reference energy. Once these functions are known at that starting point, they must be evolved to higher energies using a set of rules known as the Dokshitzer–Gribov–Lipatov–Altarelli–Parisi equations. In practice, modern physics groups provide these evolved distributions as vast grids of numbers. When a physicist runs a simulation to predict what happens in a collision, the computer must constantly stop to look up values on these grids and estimate the numbers in between. While this method is reliable, the constant stopping and guessing adds up, making the simulations take a long time to complete.
David F. Rentería-Estrada, working at the Instituto de Física Corpuscular in Spain, set out to see if there was a way to describe this evolution without relying on the grids. The goal was to find a single, continuous mathematical expression that could describe how the proton's internal map changes as the energy scale rises, from the lowest energies where the data begins up to one thousand gigaelectronvolts. Instead of treating the shape of the distribution as a fixed set of numbers that change only at specific points, the researcher proposed that the very parameters defining the shape should themselves be functions of the energy. This means that the formula describing the particles would not just be a snapshot, but a movie that plays smoothly as the energy increases.
To test this idea, the researcher took the standard, highly trusted set of data known as HERAPDF2.0 and sampled it at thousands of random points across a wide range of energies and particle momentum fractions. Using machine-assisted fitting techniques, they worked to find the best mathematical forms that could describe how the parameters of the distribution changed with energy. For most particles, a standard formula worked well, but for the gluons—the particles that carry the strong force holding the proton together—a more flexible approach was needed. The researcher introduced a special term that could switch on or off depending on the momentum fraction, allowing the formula to capture the complex behavior of gluons at very low momentum without distorting the description at higher momenta. This resulted in a new set of approximations, which the author calls ML-PDFs.
The results of this new approach were impressive. When the researcher compared the values generated by their new analytic formulas against the original grid-based data, they found that the agreement was remarkably close. For most types of particles, the difference in the total amount of momentum carried was less than one percent. Even for the strange quark, which showed the largest deviation, the error remained around 1.5 percent even at the highest energies tested. The new formulas also respected the fundamental conservation laws of physics. The total number of valence quarks and the total momentum carried by all particles inside the proton were preserved to within a few percent, a crucial check that confirmed the formulas were physically sensible.
The true test, however, was not just in the numbers themselves but in how they affected real-world predictions. The researcher used their new formulas to calculate the rates of two specific types of particle collisions: the production of a single pion and the production of a photon paired with a pion. In the first case, the predictions from the new formulas differed from the standard grid-based predictions by less than two percent. In the second, more complex case, the difference stayed within five percent for the most common collision energies. These deviations were small enough to be considered compatible with the existing uncertainties in the data, meaning the new method did not introduce any significant errors into the physics predictions.
Perhaps the most immediate benefit of this work was the speed. Because the new method does not require the computer to stop and look up values in a table, the simulations ran significantly faster. For the single pion production, the total time required to complete the calculation dropped by about eleven percent. For the more complex photon-pion production, the time savings were even more dramatic, with the run time reduced by roughly thirty percent. This gain is not a matter of a few seconds here and there; in the world of high-energy physics, where simulations can take days or weeks to run, such a reduction is a substantial improvement in efficiency.
The study concludes that this analytic approach is a successful proof of concept. It demonstrates that the complex, scale-dependent behavior of proton partons can be captured by a direct mathematical formula, offering a faster alternative to the traditional grid interpolation without sacrificing accuracy. While this method is not intended to replace the extraction of new data or to solve the evolution equations from first principles, it provides a powerful tool for speeding up the calculations that underpin our understanding of the subatomic world. The researcher suggests that this same strategy could be applied to other sets of data and even to the description of how particles break apart, opening the door to more compact and efficient representations of the fundamental forces that shape our universe.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.