AI Insights

Machine Learning is Surprisingly Good at Simulating the Universe

Published

2 months ago

July 3, 2025

At the RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences (iTHEMS) in Japan, the showdown between artificial intelligence and supercomputers has begun. It was here that Riken researchers, along with an international team of colleagues, used machine learning to enhance a simulation of galaxy evolution. The results were compared to direct numerical simulations, like those typically run on supercomputers, and AI won this round! In addition, this approach could shed light on the origins of the Milky Way and the elements essential to life as we know it.

The research was led by Keiya Hirashima, a Postdoctoral Researcher at iTHEMS and the Flatiron Institute’s Center for Computational Astrophysics. He was joined by colleagues from the Max Planck Institute for Astrophysics (MPA), the Research Center for the Early Universe at the University of Tokyo, the Center for Planetary Science (CPS) at Kobe University, New York University, Princeton University, the Tohoku University of Community Service and Science, and the Japanese machine learning company Preferred Networks, Inc. (PFN).

The simulation tackled a key issue when it comes to galaxy formation, which is the role played by supernovae. Since opportunities to study these events are few and far between, scientists must rely on numerical simulations based on data gathered by telescopes and other observation methods. These simulations are remarkably complex since they must account for cosmological forces and possess high temporal resolution so major events are not missed. This includes supernovae, which evolve from core collapse to remnant in a few months to a few thousand years, orders of magnitude smaller than what typical simulations can achieve.

In ordinary numerical simulations, supernovae occur on timescales about 1000 times smaller than what supercomputers can achieve. Moreover, simulations capable of this level of temporal resolution take 1-2 years to complete and are restricted to relatively small galaxies. To overcome this bottleneck, the team incorporated AI into a simulation based on ASURA code, which combines N-body and Smoothed Particle Hydrodynamics (SPH) methods to simulate galaxy formation. They also included the Framework for Developing Particle Simulator (FBPS) code to simulate chemical processes, and a machine learning (ML) model developed by Preferred Networks Inc.

This yielded what the team describes as the ASURA-FBPS-ML model, which allowed them to match the output of a previously modeled dwarf galaxy but got the result much more quickly. As Hirashima said in a RIKEN press release:

When we use our AI model, the simulation is about four times faster than a standard numerical simulation,” says Hirashima. “This corresponds to a reduction of several months to half a year’s worth of computation time. Critically, our AI-assisted simulation was able to reproduce the dynamics important for capturing galaxy evolution and matter cycles, including star formation and galaxy outflows.

To train their AI, the researchers fed it data from 300 simulations of an isolated supernova in a molecular cloud one million times the mass of our Sun. This produced a model capable of predicting the density, temperature, and 3D velocities of gas particles during the initial phase of supernova shell expansion, which typically lasts for 100,000 years after core collapse occurs. Compared to the kinds of direct numerical simulations performed by supercomputers, the new model yielded similar galactic structures and a star formation history within one quarter of the computing time.

This research illustrates the potential of incorporating AI into cosmological simulations, including models of how the entire Universe evolved since the Big Bang (ca. 14 billion years ago). “[O]ur AI-assisted framework will allow high-resolution star-by-star simulations of heavy galaxies, such as the Milky Way, with the goal of predicting the origin of the Solar System and the elements essential for the birth of life,” said Hirashima. Currently, the lab is using the ASURA-FBPS-ML model to run simulations of galaxies as large as the Milky Way, which could also lead to new theories about the origins of life in our galaxy.

The paper describing their findings appeared in The Astrophysical Journal.

Further Reading: RIKEN

Source link

AI Insights

Artificial intelligence helps break barriers for Hispanic homeownership

Published

2 hours ago

September 4, 2025

The Editors

For many Hispanics the road to homeownership is filled with obstacles, including loan officers who don’t speak Spanish or aren’t familiar with buyers who may not fit the boxes of a traditional mortgage applicant.

Some mortgage experts are turning to artificial intelligence to bridge the gap. They want AI to help loan officers find the best lender for a potential homeowner’s specific situation, while explaining the process clearly and navigating residency, visa or income requirements.

This new use of a bilingual AI has the potential to better serve homebuyers in Hispanic and other underrepresented communities. And it’s launching as federal housing agencies have begun to switch to English-only services, part of President Donald Trump’s push to make it the official language of the United States. His executive order in August called the change a way to “reinforce shared national values, and create a more cohesive and efficient society.”

The number of limited-English households tripled over the past four decades, according to the Urban Institute, a nonprofit research organization based in Washington, D.C. The institute says these households struggle to navigate the mortgage process, making it difficult for them to own a home, which is a key factor in building generational wealth.

The nonprofit Hispanic Organization of Mortgage Experts launched an AI platform built on ChatGPT last week, which lets loan officers and mortgage professionals quickly search the requirements of more than 150 lenders, instead of having to contact them individually.

The system, called Wholesale Search, uses an internal database that gives customized options for each buyer. HOME also offers a training program for loan officers called Home Certified with self-paced classes on topics like income and credit analysis, compliance rules and intercultural communication.

Cubie Hernandez, the organization’s chief technology and learning officer, said the goal is to help families have confidence during the mortgage process while pushing the industry to modernize. “Education is the gateway to opportunity,” he said.

HOME founder Rogelio Goertzen said the platform is designed to handle complicated cases like borrowers without a Social Security number, having little to no credit history, or being in the U.S. on a visa.

Loan officer Danny Velazquez of GFL Capital said the platform has changed his work. Before, he had to contact 70 lenders one by one, wait for answers and sometimes learn later that they wouldn’t accept the buyer’s situation.

The AI tool lets him see requirements in one place, narrow the list and streamline the application. “I am just able to make the process faster and get them the house,” Velazquez said.

One of Velazquez’s recent clients was Heriberto Blanco-Joya, 38, who bought his first home this year in Las Vegas. Spanish is Blanco-Joya’s first language, so he and his wife expected the process to be confusing.

Velazquez told him exactly what paperwork he needed, explained whether his credit score was enough to buy a home, and answered questions quickly.

“He provided me all the information I needed to buy,” Blanco-Joya said. “The process was pleasant and simple.”

From their first meeting to closing day took about six weeks.

Mortgage experts and the platform’s creators acknowledge that artificial intelligence creates new risks. Families rely on accurate answers about loans, immigration status and credit requirements. If AI gives wrong information, the consequences could be serious.

Goertzen, the CEO of HOME, said his organization works to reduce errors by having the AI pull information directly from lenders and loan officers. The platform’s database is updated whenever new loan products appear, and users can flag any problems to the developers.

“When there are things that are incorrect, we are constantly correcting it,” Goertzen said. “AI is a great tool, but it doesn’t replace that human element of professionalism, and that is why we are constantly tweaking and making sure it is correct.”

Jay Rodriguez, a mortgage broker at Arbor Financial Group, said figuring out the nuances of different investors’ requirements can mean the difference between turning a family away and getting them approved.

Rodriguez said HOME’s AI platform is especially helpful for training new loan officers and for coaching teams on how to better serve their communities.

Better Home & Finance Holding Company, an AI-powered mortgage lender, has created an AI platform called Tinman. It helps loan officers find lenders for borrowers who have non-traditional income or documents, which is common among small business owners.

They also built a voice-based assistant called Betsy that manages more than 127,000 borrower interactions each month. A Spanish-language version is in development.

“Financial literacy can be challenging for Hispanic borrowers or borrowers in other underserved populations,” Pierce said. “Tools like Betsy can interact and engage with customers in a way that feels supportive and not judgmental.”

Source link