Code Underlying: Latent Moment Beamforming (LAMB) for Precipitation using Fast-Scanning Phased Array Weather Radars and Hamiltonian Monte Carlo (HMC)

This repository contains MATLAB code implementing a Latent Moment Beamforming (LAMB) technique for retrieving Doppler moments of precipitation using fast-scanning phased array weather radars. The estimation is performed using Hamiltonian Monte Carlo (HMC) to enable physically consistent inference under uncertainty.<br>Key Features:Latent Moment Model: Represents the Doppler spectrum in terms of interpretable latent variables (mean μ, spread σ, and strength M) per beam direction.Hamiltonian Monte Carlo (HMC): Performs Bayesian inference using a leapfrog-integrated sampling framework for efficient exploration of the posterior.Gradient-based Likelihood: Includes analytical gradients to accelerate HMC convergence.Customizable Radar Parameters: Easily adapt the code to different array geometries, scan strategies, or velocity resolutions.<br>

本代码仓库包含用于实现潜矩波束成形（Latent Moment Beamforming, LAMB）技术的MATLAB代码，该技术可借助快速扫描相控阵天气雷达获取降水的多普勒矩量。该估计过程采用哈密顿蒙特卡洛（Hamiltonian Monte Carlo, HMC）执行，以在不确定性条件下实现物理一致性推断。
核心特性：
1. 潜矩模型：针对每个波束方向，以可解释的隐变量（均值μ、展宽σ以及强度M）表征多普勒频谱。
2. 哈密顿蒙特卡洛（HMC）：采用蛙跳积分采样框架执行贝叶斯推断，实现后验分布的高效探索。
3. 基于梯度的似然函数：内置解析梯度以加速HMC的收敛速度。
4. 可自定义雷达参数：可轻松将代码适配至不同的阵列构型、扫描策略或速度分辨率。