on
Machine learning narrows search for additional particles in the Higgs boson family
What if the Higgs boson found in 2012 is not alone but is the only sibling we have encountered so far? Scientists at CERN discovered the particle that year, and it was a major discovery because it explained how other particles acquire mass. For a long time, scientists thought this was the final piece of the puzzle.
They have a framework called the Standard Model that describes the smallest particles in everything we see. This includes electrons in atoms and light particles called photons. However, this framework does not explain everything. It does not tell us about dark matter or why the universe has so much more matter than antimatter. It is like having a map that shows only half the world.
The discovery of the Higgs boson created new questions. Many physicists started wondering whether the Higgs we found is the only one of its kind. They began to ask whether there is a larger family of these particles hiding in the universe. If we find more members of this family, we might finally understand the parts of nature that the current framework misses.
A three-Higgs proposal
This idea is the focus of two recent studies by researchers from IIT Gandhinagar. They worked with experts from Harish-Chandra Research Institute (HRI) to explore a theory in which there are three types of these fields instead of just one.
The team at IIT Gandhinagar included Akshat Khanna, who has a Ph.D. in physics; Baradhwaj Coleppa, associate professor of physics; Nipun Batra, associate professor of computer science and engineering; and Gokul B. Krishna, who has a Ph.D. in physics. They published their findings and shared their work at an international conference, PASCOS, at Durham University.
Their first study, published in Physical Review D, used a machine learning method called active learning to sort through thousands of possibilities, laying the foundation.
The second study, published in The European Physical Journal, looked at how a future particle collider could detect these new family members. Together, the papers provide a better map for future science experiments.
How the fields divide the work
To understand their work, think of the Higgs field as a basic building block of the universe. In the standard framework, there is only one of these blocks. As the universe evolved, interactions with this field gave mass to all the particles that make up everything around us. This leaves behind one Higgs particle.
The new theory suggests there are three different building blocks. This means there would be several new types of particles. Some might have electric charges, and others might behave in ways we have never seen before.
Each block also carries a background value—the strength of the field even in empty space—and, with three blocks instead of one, the fields can mix with one another. The result is a much richer collection of Higgs particles, including charged Higgs bosons that do not exist in the Standard Model and are now being sought in experiments.
This specific model is often called a democratic theory because the three building blocks share the work of giving mass to different kinds of matter. One block takes care of particles like electrons. Another block handles a specific set of quarks, and the third block looks after another set.
If these building blocks all interacted freely with the same kinds of matter, they might cause strange reactions that have never been seen in experiments. By giving each block a specific job, the theory stays consistent with what we already know to be true.
Fitting the known Higgs in
One important idea in the study is deciding which of the new particles is the one we already found. Since we know the mass of the current Higgs boson, we have to fit it into the new family. It could be the lightest member, or it could be the middle member.
It could even be the heaviest member. These different arrangements change how the rest of the family looks. The researchers looked at each of these scenarios to see which ones make the most sense according to the latest scientific data.
Every set of parameters in the model must pass several strict tests before it is considered valid. The energy of the system must stay stable so it does not collapse. The predictions must also match data from past experiments at particle colliders.
Most importantly, the Higgs particle we already discovered must still look and act the same way it does in our observations. The new particles cannot interfere with laws of physics that we have already proven to be correct. This keeps the theory grounded in reality.
Coleppa explains, "A plethora of theoretical possibilities remain in which this model could be consistent with experimental observations. Our task is to test these possibilities in the most general and comprehensive way."
Teaching the search where to dig
Here is where the size of the problem becomes almost comical. Imagine a board game with 14 different dials, with every dial turned to a different setting. Flip just one dial on or off, and you already have 214, or more than 16,000 combinations.
But these dials are not simple on-off switches; each one can be set to any value within a range, like a volume knob rather than a light switch. To map out this game properly, the team generated 600 billion random combinations of mass and angle settings and tested each one against the rules.
Out of those 600 billion attempts, only a tiny sliver survived all the tests. Checking each one by traditional methods, one at a time, would be like trying every possible combination on a bank vault by hand.
Picture this entire 14-dial search space as a vast, foggy beach, with a few important coins buried somewhere in the sand. A metal detector that beeps everywhere would be useless; you would dig forever. The trick the researchers used, called active learning, is closer to a smarter detector that learns from where it has already dug. It starts small, digging in a handful of random spots and keeping track of which ones turn up coins.
After a few rounds, it stops digging at random altogether. It starts noticing a rough line in the sand: On one side, coins keep showing up; on the other, nothing. Right along that line is where it keeps getting surprised.
So that is where it digs next, again and again, each new hole sharpening its sense of exactly where that line runs, until it has a confident map of the whole boundary without ever touching most of the beach. As one final safety check, anything it flagged as promising was dug up by hand anyway, just to make sure it was not a false alarm.
The result of swapping blind digging for this smarter search was dramatic: An analysis that used to take about four hours could now be done in around 10 minutes, with no real particles missed along the way.
Batra explains, "Not every data point is equally informative for training a machine learning model. Active learning addresses this by selecting a small but highly informative subset of data, enabling the model to learn more efficiently."
Three possible family trees
With the search done, the team had a clear map, and it turned out the family of three Higgs bosons can be arranged in three different ways, depending on whether the 125 GeV Higgs we already know is the youngest, middle or oldest sibling.
In the first arrangement, our known Higgs is the youngest, the lightest of the three. Here, the model is well-behaved. Its two heavier siblings are predicted to weigh somewhere between roughly 350 and 580 GeV, several times heavier than the Higgs we have already found.
Nature, it seems, is allowed to have bigger siblings hiding just out of reach of past experiments, but not absolutely any weight will do; they have to fall inside this specific window.
The most interesting twist comes from the second arrangement, where our known Higgs is the middle child. This setup quietly allows for something remarkable: a second, lighter Higgs boson, possibly as light as 82 GeV, that nobody has detected yet.
It is not that this particle would have been invisible; it is that it may interact with other particles differently enough to have hidden in plain sight within data we already have. That is the headline result of the whole study. A sibling of the Higgs boson, lighter than the one history books already credited with completing the Standard Model, remains entirely on the table.
The third arrangement is the one nature seems to veto. Here, our known Higgs would have to be the oldest, the heaviest of the three, which would mean two undiscovered Higgs bosons lighter than 125 GeV are floating around. On paper, this sounds just as exciting as the middle-child case. But when the team pushed this scenario through the full gauntlet of tests, it failed.
A cleaner machine for the hunt
The second paper looks at how we can actually find these hidden particles. The researchers focused on a proposed machine called the International Linear Collider. This machine would collide tiny particles to create a very clean environment for study. It is different from the machines we use today because it allows scientists to see the results of collisions much more clearly.
The team investigated several ways this future machine could produce the new Higgs family members in a laboratory setting.
The researchers found several promising ways to spot the new particles. When the machine collides particles, it can create pairs of the new Higgs family. These particles quickly break down into other things that we can see and measure. By counting these final pieces, scientists can separate the real signal from the background noise.
This is similar to listening for a specific voice in a crowded room. If you know what to listen for, you can find the information you need even when there is a lot of other sound.
The study shows that many of these paths could lead to a real discovery. Some channels are very likely to show results if the machine runs for a certain amount of time. The researchers looked at different types of signals, including those that produce specific patterns of quarks and leptons.
This provides a clear set of instructions for future experiments. It turns the theoretical map into a practical guide for the scientists who build and run these massive machines.
From theory to a test plan
The strength of this work is how it connects theory with experiment. The first study uses smart computers to find the best possibilities, and the second study tells us how to look for them. This is very important because theories with too many options can be hard to test.
The researchers at IIT Gandhinagar made their theory testable by narrowing down the search. As the team explains, combining physics with smart computer strategies allows us to search for the secrets of the universe much faster than before.
This research is about more than just one theory. It helps the global effort to see whether the Higgs boson is alone or has siblings. It also proves that smart computer tools can help scientists navigate very complex problems.
This method can be used in many other areas of science where there are too many possibilities to test. The immediate goal is to help find new particles, but the long-term goal is to improve how we do science in a world with so much information.
The story of the Higgs boson is really a story about how the universe is built. We started with one particle, but we might have a much bigger family. The studies at IIT Gandhinagar show that this idea is very much alive.
One lighter sibling might be hiding in our current data, while others might be waiting for future machines. By using smart computers to draw the map and collider studies to lead the way, we are getting closer to understanding the full architecture of our world.
Khanna adds, "The key to our work was using innovative ways like combining modern ML techniques with particle physics to tackle a formidable challenge. In the near future, I am eager to pursue this line of research even further, in search of evidence of new particles."
Publication details
Nipun Batra et al, Constraining the 3HDM parameter space using active learning, Physical Review D (2025). DOI: 10.1103/t5df-67wh
Baradhwaj Coleppa et al, ILC phenomenology of the Z3 symmetric type-Z three Higgs doublet model, The European Physical Journal C (2026). DOI: 10.1140/epjc/s10052-026-15413-9
Provided by Indian Institute of Technology Gandhinagar
Who's behind this story?
BA art history, MA material culture. Former museum editor, paramedic, and transplant coordinator. Editing for Science X since 2021. Full profile →
Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →
Citation: Machine learning narrows search for additional particles in the Higgs boson family (2026, July 23) retrieved 23 July 2026 from https://phys.org/news/2026-07-machine-narrows-additional-particles-higgs.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.