Technical advisor, R&D position and / or CPU optimization engineering.
My interests mainly resides on cryptography (cryptocurrency) and CPU optimization. I want to do a research about cryptography (cryptocurrency), develop / implement cryptographic technologies to build a unique service which has never been achieved in the world and / or optimize codes to speed up their execution time.
Researcher/Inventor/Developer in Physics and Information Science with a highly sophisticated background in Mathematics. Conducted technological research in physics and cryptocurrencies and developed many web services.
- Cryptocurrency brain wallet.
- Open-source blockexplorer.
- Hacking-proof cryptocurrency EC infrastructure.
- Cryptocurrency mining pool.
- Mobile top-up service (joint work with Ceres, Inc.).
- Cryptocurrency exchange.
- The world's first practical implementation of private information retrieval (PIR).
All services except explicitly noted as joint work are developed by my self (solo works).
The basics of thermodynamics and statistical mechanics. Theoretical and numerical analysis of the basics of thermal systems with long-range interactions and thermal quantum systems using GPUs and supercomputers.
Deep knowledge about cryptography (cryptocurrency), group theory, primarily tests, quantum theory, statistical mechanics, thermodynamics.
- Frameworks: Git, automake, CMake, npm, Jest, Nuxt.js (Vue.js).
- High Performance Computation (HPC): GPGPU (OpenCL), MPI and OpenMP. Optimization for specific devices including AMD GPU, NEC SX9 (vector computer) and FUJITSU FX10 (K -computer compatible).
- Languages: Japanese (native), English (daily conversation)
- Misc: Ubuntu Linux, vim, Nginx, Apache, Autotools, Git, Debianize (package creation), (My)SQL (MariaDB) and AVR MCU.
- University of Tokyo, Tokyo, Japan
- Ph.D (dropout), Graduate School of Arts and Sciences (Physics), April 2014
- M.S., Graduate School of Arts and Sciences (Physics), April 2012 - March 2014
- B.S., Department of Physics, Faculty of Science, April 2008 - March 2012
EllipticPIR is an implementation of private information retrieval (PIR). PIR is a cryptographic building block which enables users to fetch data from a server without revieling any information about their query. Despite many researches on PIRs in history, there were no implementation capable of serving a practical size of database. I invented a new PIR instanciation from the EC-ElGamal encryption and implemented it using C++ and OpenCL (server side) and C, Rust, TypeScript and WebAssembly+WebWorker (client-side). My implementation could reply to 100M database elements with 32 bytes length in ~2600ms with a single NVIDIA RTX3080.
Fressets EWM© is a cryptocurrency wallet with enterprise-level security and availability. I founded a company named "Fressets" in 2017 and created a cryptocurrency backend service which is highly secure and robust. EWM is designed to be used in cryptocurrency exchanges who accept customers' deposits and will keep millions of dollars. We successfully delivered EWM to two licensed exchanges in Japan.
CoinGift is an EC service for selling gift cards in exchange for cryptocurrency. Its backend provides a hacking-proof EC platform by separating private keys from receiving addresses (public keys) and storing private keys in a safe place (e.g. in a GPG-encrypted file stored in a cloud storage service) using a hierarchical deterministic (HD) wallet structure. This prevents attackers from being able to extract coins which customers sent to the service should the attackers acquire even if attackers could have full access to the web server.
chainseeker is an open-source blockchain explorer fully written in Rust. Block explorer is a (web) service users can examine blocks, transactions and addresses on the blockchain. My implementation could sync to the Bitcoin mainnet within 24 hours which has more than 500GB of data. I also implemented a web interface using Nuxt.js (Vue.js).
Analysis of systems with long-range interactions.
Long-range systems are known to be extremely hard to analyze by numerical means because of its complexity. In situations where computational intensity is very high, the GPGPU computation is very efficient. We have numerically verified the scaling law of such systems, which was originally speculation, and extended it for the case where there are two or more scaling factors.
Analysis of quantum many-particle system at finite temperature.
A new algorithm was invented recently that enablesd us to compute a larger system via traditional means. We extended this algorithm to many-particle systems and applied it to a well-known system using supercomputers. The results from these experiments validated the previous results that were derived in a probabilistic fashion.
- M.Hyuga, S.Sugiura, K.Sakai and A.Shimizu, Thermal Pure Quantum States of Many-Particle Systems, Phys. Rev. B 90, 121110 (2014)
- M.Hyuga and A.Shimizu, The Scaling Laws of Systems With Long-Range Interactions, Bussei-kenkyu vol.3, 2 (2014) (Japanese)
- M.Hyuga and T.Imai, The Frontier of Bitcoin Improvement Proposals (BIPs), Information Processing Society of Japan 2016 Vol.57 No.12, (2016) (Japanese)
- M.Hyuga and K.Sawada, Wallet System for Cryptocurrency and the Way to Sign Using It, J.P. Patent 2020-074559 (2020)