USPEX@HOME

USPEX@HOME
USPEX@HOME logo
Project
Status	Active
Category	Materials science, Crystallography, Computational chemistry
Compute	CPU
Development
Developer	Artem R. Oganov and the USPEX Team
Author	Artem R. Oganov
Sponsor	Skolkovo Institute of Science and Technology
Maintainer	USPEX Team
Initial release	June 1, 2023  (3 years ago)
Software
Written in	C++, Fortran
Operating system	Windows, Linux, macOS
Metadata
Website	https://uspex-at-home.ru/prediction/

[[File:{{#setmainimage:Uspex.png}}|alt=logo image|center|frameless]]

USPEX@HOME is a volunteer computing project based on the BOINC platform. The project allows volunteers around the world to donate spare processing power from their computers to help researchers discover and predict new crystalline materials and compounds.

The project is based on the USPEX evolutionary crystal structure prediction method, a computational approach used in materials science and condensed matter physics to predict stable atomic structures under different physical conditions.

USPEX@HOME focuses on identifying novel materials with potentially useful properties, including superconductors, superhard materials, energy storage compounds, catalysts, and high pressure phases of matter.

Why USPEX@HOME?

Predicting the atomic arrangement of materials is one of the major challenges in computational materials science. The physical and chemical properties of a material are strongly determined by the arrangement of its atoms in three dimensional space. Determining the most stable structure for a chemical composition often requires searching an enormous number of possible atomic configurations.

For a crystal structure containing many atoms, the number of possible arrangements increases combinatorially. Researchers attempt to minimize the total free energy of the system:

$G=H-TS$

where:

• $G$ is the Gibbs free energy
• $H$ is enthalpy
• $T$ is temperature
• $S$ is entropy

Stable crystal structures generally correspond to local or global minima of the free energy surface.

Because these calculations are computationally intensive, distributed volunteer computing provides a practical way to perform large numbers of simulations simultaneously across thousands of participating computers.

Goal

The primary goal of USPEX@HOME is to accelerate the discovery of new materials through large scale computational simulations performed on volunteer computers. The project uses evolutionary algorithms and first principles calculations to search for stable and metastable crystal structures.

Researchers associated with the project investigate materials that may exhibit important technological properties, including:

• High temperature superconductivity
• Superhardness
• Improved battery performance
• Hydrogen rich compounds
• Novel semiconductors
• High pressure phases of matter

Many of these calculations involve quantum mechanical modeling techniques derived from density functional theory and related computational chemistry methods.

The project also serves as a public participation initiative in modern scientific research, allowing volunteers to contribute directly to materials discovery and computational physics research.

Methods

Brianite crystal structure (Alkemper-Fuess 1998) crystallographic standard alignment

USPEX@HOME uses the USPEX evolutionary algorithm framework together with BOINC distributed computing infrastructure.

The USPEX method applies concepts inspired by biological evolution to optimize crystal structures. Candidate structures are generated and refined over multiple generations using operations such as:

• Heredity
• Mutation
• Permutation
• Natural selection

The fitness of a structure is generally determined by minimizing total energy or enthalpy:

$H=E+PV$

where:

• $H$ is enthalpy
• $E$ is internal energy
• $P$ is pressure
• $V$ is volume

Structures with lower enthalpy are considered more stable under given pressure conditions.

The BOINC infrastructure enables these workloads to be distributed across thousands of volunteer systems. Each participant downloads work units, processes calculations locally, and uploads results back to the project servers.

Volunteer computing is particularly well suited to this type of research because crystal structure searches often require evaluating many independent candidate structures in parallel. This parallelism allows the project to scale efficiently across geographically distributed computers.

USPEX has been widely cited in computational materials science literature and has contributed to the prediction of numerous experimentally verified materials and high pressure compounds.

Volunteer Computing

USPEX@HOME is part of the broader ecosystem of BOINC based volunteer computing projects. Volunteers install the BOINC client software and attach to the project using an account on the project website.

The project primarily distributes CPU based tasks related to crystal structure optimization and materials simulations. Depending on the workload, tasks may vary significantly in runtime and computational complexity.

Like many BOINC projects, USPEX@HOME uses a credit system to reward volunteers for completed work units and to encourage community participation.

Scientific significance

BOINC project architecture

Crystal structure prediction has become increasingly important in modern materials science. Computational methods can reduce the need for expensive laboratory experiments by identifying promising candidate materials before synthesis.

Applications of predicted materials may include:

• Energy storage systems
• Aerospace engineering
• Electronics
• Catalysis
• Medical technologies
• High pressure physics

USPEX methods have been used in the prediction of new superconducting hydrides and other exotic materials under extreme conditions.

Project team / Sponsors

The project is developed and maintained by the USPEX research team led by Artem R. Oganov, a computational materials scientist known for his work in crystal structure prediction and high pressure chemistry.

Research associated with USPEX has involved collaborations with institutions including:

• Skolkovo Institute of Science and Technology
• Moscow Institute of Physics and Technology
• Russian Academy of Sciences

Statistics

As with other BOINC projects, USPEX@HOME publishes statistics regarding participating users, hosts, and computational throughput through its server status pages.

The project maintains international participation from volunteers contributing computational resources from personal computers and servers worldwide.

See also

• BOINC
• Volunteer computing
• Distributed computing
• Materials science
• Crystal structure prediction
• Density functional theory
• Computational chemistry

External links

• Official USPEX@HOME website
• Official USPEX Team website
• BOINC official website