Why Matlantis

From Following to Leading: A New Era for Simulation

The unprecedented computational speed of Matlantis enables high-speed screening before experiments and visualization at the atomic level, expanding the role of simulation beyond just "theoretical verification of experimental results" to "providing guidelines for experiments."

Speed comparison with DFT calculations

close

Reliable and Practical Calculation Accuracy

Matlantis' potential is trained with over 59 million large-scale data sets that intentionally include unstable structures. This allows it to faithfully reproduce DFT calculations while being reliable enough for practical use.

Freedom from Computation Time Constraints

Compared to conventional DFT calculations, the number of atoms that can be handled and the simulation time are both dramatically improved. This allows us to analyze more realistic and complex phenomena, such as complex systems that could not be handled before and long-term dynamics.

Overcoming the Limitations of Computing Resources and Expanding the Range of Research Themes That Can be Undertaken

By dramatically shortening the time required for each research topic, the overall throughput of R&D will improve dramatically. This will create opportunities to use the power of computation to approach challenging topics that previously had to be lowered in priority.

General Machine Learning Potential (MLIP) That is Not Restricted by Element Type

Since its release, the software has been continuously updated and now supports 96 elements, covering all stable elements present in nature. This allows customers to perform simulations without being aware of restrictions on element types.

High-Precision Energy and Force Calculation to Predict Physical Phenomena

The energy and forces that are the basis of physical property calculations are faithfully reproduced with DFT accuracy. Based on this, various physical property values such as density, diffusion coefficient, viscosity, and thermal conductivity can be derived through molecular dynamics calculations.

A Single Model That Can be Applied Across Disciplines

A single AI model can handle complex systems that are important in real materials, such as bulk crystals, amorphous materials, surfaces, interfaces, and defects. This accelerates research and development across material fields with vastly different properties, such as catalysts, batteries, semiconductors, and polymers.

AI Models That Respond Immediately to Research Themes, Without the Need for Fine Tuning

There is no need for fine-tuning or additional training for each topic, which is required for general MLIP. Customers can simply apply the pre-trained, highly accurate PFP to their own research topic and start calculations immediately.

No Need to Set Up an Environment, Just Start Calculations from your Browser

Because it operates on a cloud-based system, there is no need to prepare high-performance hardware or install software. You can perform calculations immediately from your PC.

Actual Matlantis operation screen example

close

Freedom from Operations and Maintenance, Focus on Core Research

Matlantis handles all the management and updates of hardware and software, which requires specialized knowledge and a great deal of effort. This allows customers to focus on their core simulation work without having to worry about maintaining their computational environment.

Accelerating Collaborative Research through Smooth Data Sharing within an Organization

All calculation data is managed centrally on the cloud, making it easy to share information within an organization or across multiple locations, accelerating collaboration and improving the speed of research and development.

Flexible Workflow Construction Using Python

Matlantis provides a Jupyter Notebook environment that uses Python, a common language widely used in scientific and technical computing. This allows you to smoothly build your own workflow, such as flexible customization for your research topic and multifaceted data analysis combined with a wide range of public libraries.

Scalability of the Computing Environment and Integration with External Tools

There is no need for a dedicated high-performance GPU. Matlantis provides abundant cloud GPU resources that can be flexibly allocated to each user according to the computational load. In addition, SSH integration allows seamless connection to familiar PCs and external tools, streamlining the entire R&D flow.

Matlantis Provision Environment

close

LightPFP: Specialized Potentials for Larger-Scale Simulations

By using the results of PFP calculations as training data, the construction of specialized potentials, which usually takes several months, can be shortened to a few days. This makes it possible to perform simulations on a scale of hundreds of thousands of atoms and long-term molecular dynamics calculations, which would be difficult to perform using PFP alone, in a realistic amount of time while maintaining the same level of accuracy as PFP.

Continuous Performance Updates That Evolve with Users

We value customer feedback and continually improve our products, responding quickly to the ever-changing needs of research sites and providing the best possible research environment.

Implementation Support to Help Your Research Get Started Smoothly

Our expert team will provide thorough support, from proposing the optimal implementation plan based on your issues and goals, to training you on basic operations and answering your questions after you start using the system. We will help you until you are able to advance your research autonomously.

Global Technical Support Based on Knowledge of Materials Science and Computational Chemistry

Matlantis is supported by a global team of experts based in Tokyo and Cambridge, USA. With extensive knowledge covering a wide range of materials fields, we are able to provide technical advice on analysis approaches and interpretation of results, not just on how to use the tools.

Extensive Information Resources to Promote Self-Learning and Skill Development

Our extensive online resources include technical documentation, tutorials, and webinar archives, and our user community is always available to help you continually improve your skills.

Technical Co-Creation with Preferred Networks, a Leader in AI Development in Japan

Matlantis was born from the technology of Preferred Networks, a leader in AI development in Japan. By combining Preferred Networks' cutting-edge AI technology with its knowledge of materials science, it overcomes the limitations of the trade-off between accuracy and speed.

A Track Record of Being Chosen by Top Companies in a Variety of Industries

We have been chosen by many leading companies in a variety of industries to contribute to our customers' material development. Our partnerships with top companies around the world and numerous success stories are a testament to our technical capabilities and reliability.

Quickly Confirm Microscopic Behavior That Cannot Be Seen in Experiments

Matlantis meets the demand for "more accurate and faster understanding of what is happening in the microscopic world." With its overwhelming calculation speed and versatility, it enables visualization of phenomena with minimal preparation and calculation time. Being able to "see" things immediately brings new insights.

Oxidation reactions on aluminum surfaces

close

Easily Verify the Conditions You Want to Try. A New Cycle That Combines Calculation and Experiment

"What will happen under these conditions?" Condition changes that would normally require additional experiments can now be easily tested computationally. This realizes a new style of research in which experimental chemists and computational scientists can discuss in real time while looking at the same calculation results, and quickly decide on the next experimental direction.

Ultra-High-Speed Screening to Narrow Down Candidates before Experimentation

Matlantis's unprecedented computational speed allows DFT-level material screening, which previously took months, to be completed in a matter of hours or days. This allows you to predict the "hit or miss" of the conditions you want to try before the experiment, and create an efficient experimental plan that narrows down to only promising candidates.

Comprehensive Materials Research That Expands Researchers' Knowledge

Matlantis starts from the knowledge of researchers and comprehensively examines a vast search space that humans alone cannot fully explore, maximizing the possibility of encountering unexpected materials and mechanisms (serendipity), leading to breakthrough discoveries.

Zero-Setup Environment: Start Instantly from Your Browser

Matlantis provides a high-performance computing environment on the cloud. No dedicated PC or complicated setup is required. Simply log in from your web browser and you can immediately start atomic-level simulations.

Learning Resources and Expert Technical Support to Support Independent Research

We provide a wide range of tutorials and technical materials so that even those who are not experts in computational chemistry can use it with confidence. In addition, experts with knowledge of materials science are available to provide technical consultations on analysis approaches and interpretation of results, helping customers improve their skills and produce results.

Published in Nature Catalysis: Mechanism of Catalyst for Methanol Synthesis from CO₂ Elucidated

In collaboration with experimental results, atomic and electronic level simulations helped elucidate the mechanism of a high-performance CO2 reduction catalyst and identify the key factors for improving its performance. The combination of experiments and computational chemistry has paved the way for the rational design of better catalysts.

Published in J. Am. Ceram. Soc.: Reproducing Complex Glass Structures with Experimental Level Precision

The complex microstructure of glass, which was difficult to reproduce using classical force fields, has been reproduced with a high degree of accuracy consistent with experimental data. The atomic connections and special ring structures that are the source of the physical properties have been accurately captured, dramatically improving the accuracy of predicting the physical properties of materials.

Published in Macromolecules: Visualization of the Behavior of Polymer Membranes That Affect Fuel Cell Performance

Why do performances differ depending on the material? The mechanism behind the performance differences observed in experiments has been elucidated through large-scale simulations. It has been suggested that the difference in proton conductivity of the polymer electrolyte membranes (PEMs) used in fuel cells is due to the shape of the molecular chains.

Accelerating Materials Research through Large-Scale "Virtual Experiments"

Matlantis boasts unprecedented computational speed, shortening the material search process that previously took several months to less than a day. By quickly conducting thousands or tens of thousands of "virtual experiments" before physical experiments, it is possible to quickly identify promising themes and accelerate the research and development cycle itself.

Streamlining the Development Process by "Preempting" Experiments

By narrowing down promising candidates through virtual experiments before physical experiments begin, the number of "failed experiments" is dramatically reduced, shortening development time and making the development process much more efficient.

Increase Investment Efficiency and Contribute to Business Value

The acceleration and efficiency of R&D through virtual experiments reduces uncertainty in R&D and increases the predictability of results, promoting strategic management of innovation and contributing to improved investment efficiency and, ultimately, increased business value.

Adopted by over 100 Companies and Research Institutes Leading a Wide Range of Industries

Matlantis is used by over 100 leading companies, universities, and public research institutes in a wide range of industries, from chemical and materials manufacturers to energy, semiconductors, and automotive. Its use is spreading around the world, including North America, Europe, and the Asia-Pacific region, and over 60 academic papers using Matlantis have been published.

Global usage

close

Confidence in Future Potential, Evidenced by Collaboration with Top Partners

Many companies promoting cutting-edge R&D are using Matlantis as an important tool for next-generation material development. Technical collaboration with top industry partners, including NVIDIA, the world leader in GPU computing, is also proof of this trust. Introducing cutting-edge tools sends a strong message to both inside and outside the company about the company's commitment to innovation.

A World-Class AI Model That Makes the Impossible Possible

At the core of Matlantis is a proprietary AI model that predicts the behavior of a wide range of atoms. By combining cutting-edge AI technology with chemical knowledge, Matlantis is the world's first company to put into practical use the concept of treating all substances with a single model, a concept that was previously considered "extremely difficult to achieve."

Practical Value Proven by over 100 Implementations

Since the launch of commercial services in 2021, the platform has been used in material development at over 100 companies and research institutions, and by incorporating practical feedback from customers, it continues to evolve into a platform that produces truly valuable results.

Advanced Technology Recognized by World-Renowned Authorities in Materials Science

Professor Ju Li of the Massachusetts Institute of Technology (MIT), a world authority in the field of materials science, is a user himself and also participates in the project as a technical advisor, incorporating his cutting-edge knowledge to determine the direction of development.

Eliminates the Need for Expensive IT Investments and Improves the Agility of Research and Development

There is no need for large initial investments in high-performance computers. Matlantis always provides the latest computing environment in the cloud. This frees companies from the risks and costs of owning IT assets, and gives them the agility to quickly begin research and development in response to market changes.

Reliable Security and Data Management in Compliance with International Standards

Matlantis protects customer data with a strict management system that complies with ISO certification, an international standard for information security. By providing an independent and secure environment for each customer, we minimize the risk of intellectual property leakage and ensure business continuity.