MIXMAX · Project

Ultra-Fast Random Number Generator Software for High-Performance Simulations

digitalPrototypeTRL 4Thin data (2/5)

inp.demokritos.gr

Imagine you need to flip a coin millions of times per second for a computer simulation — but the coin has to be perfectly fair every single time. That's what random number generators do, and most existing ones start repeating patterns or showing biases when pushed hard enough. A team of mathematicians and physicists built a new generator called MIXMAX, rooted in deep chaos theory, that produces random numbers faster and with better quality than standard tools. It was designed for massive physics experiments at CERN but works anywhere Monte Carlo simulations are used.

By the numbers

consortium partners

countries involved (AM, CH, CN, DK, EL)

EUR 252,000

EU contribution

dimensionality levels (N=256, 509, 1260)

industry partner ratio in consortium

The business problem

What needed solving

Companies running large-scale Monte Carlo simulations — in finance, drug discovery, materials science, or security — depend on random number generators that are fast, statistically flawless, and scalable across parallel computing clusters. Standard generators can introduce subtle biases or correlations that corrupt simulation results, leading to mispriced risk, flawed molecular models, or weak encryption.

The solution

What was built

The team built the MIXMAX generator: a ready-to-use pseudo random number generator program with tunable parameters at dimensionalities N=256, 509, and 1260, operating over a Galois field of order p=2^61-1, packaged with a user-friendly interface and online manual. It was implemented into CERN's distributed computing software for LHC experiments.

Audience

Who needs this

Quantitative finance firms running Monte Carlo risk simulationsCybersecurity companies building encryption and key-generation systemsPharmaceutical companies using molecular dynamics simulations for drug discoveryCloud computing providers offering simulation-as-a-service platformsNational laboratories and research computing centers

Business applications

Who can put this to work

Financial Services

enterprise

Target: Quantitative trading firms and risk management departments at banks

If you are a financial firm running Monte Carlo simulations for portfolio risk, derivatives pricing, or stress testing — this project developed the MIXMAX generator with tunable dimensionality up to N=1260 that produces higher-quality random sequences. Poor randomness in financial models can lead to mispriced risk. This generator was specifically built for massive parallel computing environments, meaning it scales across your existing infrastructure.

Cybersecurity

any

Target: Companies developing encryption software or hardware security modules

If you are a cybersecurity company that depends on strong pseudo-random number generation for cryptographic keys and protocols — this project created a mathematically proven generator based on K-systems with a Galois field order of p=2^61-1. Weak randomness is the root cause of many security breaches. The MIXMAX generator's mathematical foundation in ergodic theory provides a stronger theoretical guarantee of unpredictability than conventional generators.

Pharmaceutical & Materials Science

mid-size

Target: Drug discovery firms and materials simulation companies

If you are a pharma or materials company running molecular dynamics or quantum chemistry simulations — this project delivered a ready-to-use generator with an online manual and tunable parameters at dimensionalities of N=256, 509, and 1260. Simulation accuracy depends directly on random number quality. The MIXMAX generator was built for exactly these kinds of large-scale scientific computations across distributed computing setups.

Frequently asked

Quick answers

What would it cost to adopt this generator?

The MIXMAX generator was developed under an MSCA-RISE project with EUR 252,000 EU funding across 7 academic and research partners. As an academic output, the software is likely available as open-source or through academic licensing. Contact the coordinator at NCSR Demokritos for specific licensing terms.

Can this scale to industrial-grade computing environments?

Yes — the generator was explicitly designed for concurrent and distributed software, with implementation targeted at CERN's computing infrastructure for LHC experiments. It supports tunable dimensionality up to N=1260 and operates over a Galois field of order p=2^61-1, making it suitable for massive parallel simulations.

What is the IP situation and how can we license this?

The project was funded under MSCA-RISE (staff exchange program) with 7 partners across 5 countries. IP is likely held by the consortium, primarily academic institutions. Based on available project data, no commercial licensing structure was mentioned — you would need to negotiate directly with NCSR Demokritos in Greece.

How does this compare to existing random number generators?

The objective states MIXMAX is based on Kolmogorov-Anosov K-systems from ergodic theory, which the team claims demonstrates excellent statistical properties. The generator was specifically developed as a next-generation replacement for standard generators used in particle physics and quantum chemistry simulations.

Is there technical support or documentation available?

The deliverable explicitly includes an online manual and a user-friendly environment. The generator was delivered with tunable internal parameters at three dimensionality levels (N=256, 509, 1260). However, ongoing support would depend on the academic team's availability since there are no commercial partners in the consortium.

What industries has this actually been tested in?

Based on available project data, the generator was implemented and tested within CERN's high-energy physics computing environment for LHC experiments. Additional testing included large-scale simulations in quantum gravity and quantum field theory. No commercial or industrial pilot testing outside of research is documented.

Consortium

Who built it

The MIXMAX consortium of 7 partners across 5 countries (Armenia, Switzerland, China, Denmark, Greece) is entirely academic — 4 universities and 3 research organizations with zero industry partners and zero SMEs. This is a 100% research-driven project funded through MSCA-RISE, which is a staff exchange programme, not a technology commercialization scheme. The EUR 252,000 budget is modest and typical for mobility grants. For a business looking to adopt this technology, the absence of any commercial partner means there is no existing pathway to market — you would be the first to attempt commercialization, which is both an opportunity and a risk.

NATIONAL CENTER FOR SCIENTIFIC RESEARCH "DEMOKRITOS"Coordinator · EL
KOBENHAVNS UNIVERSITETparticipant · DK
ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNEpartner · CH
NANJING UNIVERSITYpartner · CN
ORGANISATION EUROPEENNE POUR LA RECHERCHE NUCLEAIREparticipant · CH
NATIONAL ACADEMY OF SCIENCES OF THE REPUBLIC OF ARMENIApartner · AM

How to reach the team

NCSR Demokritos, Greece — use Google AI search to find the principal investigator's contact details

Order a report on this consortium See NATIONAL CENTER FOR SCIENTIFIC RESEARCH "DEMOKRITOS" profile →

Next steps

Talk to the team behind this work.

SciTransfer can help you evaluate the MIXMAX generator for your specific simulation needs and arrange a technical discussion with the research team.

Order a report on this topic More in Digital & ICT