Add vectorized variant for functions of complex variable, especially exp

未关闭
#3,138 4 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
35/100
Issue 类型
功能
描述清晰度
基本清楚
活跃度
停滞
技术栈
cpp
领域
backend

调研方向

该 issue 没有指明文件或测试。先从复值基本函数指南和现有的向量化 pow 行为开始,然后追踪复标量函数如何与向量或矩阵 autodiff 交互。完成内容应涵盖所请求的 exp 向量化行为,并澄清 log 和三角函数更广泛的一致性问题。

由索引模型根据 Issue 内容生成。

描述

good first issue

Currently, according to the guide https://mc-stan.org/docs/functions-reference/complex-valued_basic_functions.html it seems that basic arithmetic operators and pow are the only functions that accept complex vectors.

I am modelling some periodicities in data using Fourier series, and I find it the most compact (and computationally efficient) to express what I do in terms of complex numbers. However, it's not efficient to run a function over a matrix using a loop when there's autodiff.

The most striking case is the exp function, which is ubiquitous in complex analysis and I want to be able to run it elementwise over a vector or a matrix.

I figured out a possible workaround is to use pow(exp(1), z) which does work on vectorized z, but still I believe that exp is more fundamental than pow and it should be implemented, because the current state is confusing at best.

Of course then there are all the other functions which accept a real or complex scalar and accept vectorized real, but don't accept vectorized complex, which is inconsistent. This includes log and the trigonometric functions.

主要语言
C++
星标
845
派生
225
平均合并
4 天 21 小时
30 天内合并 PR
13

环境准备

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

stan-dev/math 的其他 Issue

查看 stan-dev/math 的全部 Issue

相似的 Issue

更多 C++ Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。