Hello, I'm Tae-Geun Kim 👋
- Yonsei HEP-COSMO
- Department of Physics, Yonsei University
- Curriculum Vitae
- Blog
- High energy astrophysics, dark matter and cosmology
- Scientific computation
- Machine Learning / Deep Learning / Statistics
- Quantum Computing
- Rust numeric library for linear algebra, numerical analysis, statistics, and machine learning
- Provides customizable features for pure Rust, BLAS/LAPACK integration, plotting, and data handling
- Offers user-friendly syntax similar to R, NumPy, and MATLAB
- Supports functional programming, automatic differentiation, and various numerical algorithms
- Includes statistics, special functions, plotting, and DataFrame capabilities
- Compatible with mathematical structures and leverages Rust's performance and package management
- Pure Rust library for special functions with no dependencies
- Implements gamma, beta, and error functions
- Provides regularized and inverse versions of the functions
- Lightweight and efficient implementation
- Ideal for mathematical and scientific computing applications
- Based on algorithms from "Numerical Recipes" by Press and Vetterling
- Reinforcement Learning (RL) library in Rust
- Modular design with components for agents, environments, policies, and utilities
- Efficient and safe implementation leveraging Rust's performance and safety features
- Provides a framework for creating and managing diverse RL environments
- Supports customizable agent strategies and learning algorithms
- Includes implementations of Epsilon Greedy Policy, Value Iteration, and Q-Learning
More projects
Radient
- Rust library for automatic differentiation using computational graphs
- Implements forward and backward propagation for gradient computation
- Supports various mathematical operations, including exponential, logarithmic, power, and trigonometric functions
- Provides two options for gradient calculation:
-
gradient: Concise but relatively slower -
gradient_cached: Fast but slightly more verbose
-
- Includes examples demonstrating basic operations with symbols, gradient calculation, and a single-layer perceptron implementation
DeeLeMa
- Deep learning network for estimating mass and momenta in particle collisions at high-energy colliders
- Generates robust mass distributions with peaks at physical masses, even with combinatoric uncertainties and detector smearing effects
- Adaptable to different event topologies, particularly effective when corresponding kinematic symmetries are adopted
- Current version (v1.0.0) is constructed on the
$t\bar{t}$ -like antler event topology - Provides clear instructions for installation, training, and monitoring using Pip or Huak (recommended)
- Encourages citation of the associated research paper if DeeLeMa benefits users' research
-
Chang Min Hyun, Tae-Geun Kim, and Kyounghun Lee, Unsupervised sequence-to-sequence learning for automatic signal quality assessment in multi-channel electrical impedance-based hemodynamic monitoring, CMPB 108079, arXiv:2305.09368 (2023)
-
Kayoung Ban, Dong Woo Kang, Tae-Geun Kim, Seong Chan Park and Yeji Park, DeeLeMa : Missing information search with Deep Learning for Mass estimation, Phys. Rev. Research 5, 043186, arXiv:2212.12836 (2022)
-
Yongsoo Jho, Tae-Geun Kim, Jong-Chul Park, Seong Chan Park and Yeji Park, Axions from Primordial Black Holes, arXiv:2212.11977 (2022)
Github contributions
More specific
- Functional Analysis
- Differential Geometry
- Numerical Analysis
- Quantum Field Theory
- General Relativity
- Mathematical Physics
- Main Languague : Rust
- Sub Languages : C++, Julia, R, Python
- Frameworks or Libraries
- Numerical: peroxide, BLAS, LAPACK, numpy, scipy
- Visualization: matplotlib, vegas, ggplot2, plotly
- Web: Django, Vue, Firebase, Surge, Hugo
- Machine Learning: Scikit-Learn
- Deep Learning: PyTorch, Flux