數值軟體開發(英文授課)

This course discusses the art to build numerical software, i.e., computer programs applying numerical methods for solving mathematical or physical problems. We will be using the combination of Python and C++ and related tools (e.g., bash, git, make, etc.) to learn the modern development processes. By completing this course, students will acquire the fundamental skills for developing modern numerical software.

This is a graduate or senior level course open to students who have taken engineering mathematics or equivalence. Working knowledge of Linux and Unix-like is required. Prior knowledge to numerical methods is recommended. The instructor uses English in the lectures and discussions.

無備註

The instructor will use English. It is OK for students to use Mandarin in the class, but English is preferred. Computer program for homework and project should be developed against the latest Ubuntu LTS system. Course notes: https://github.com/yungyuc/nsd .

* There is usually homework given after a lecture to exercise the lectured materials, and it usually requires students to write computer programs in Python and/or C++. * Mid-term examination will be conducted to assess students' understandings to the analytical materials. * Term project will be used to assess students' overall coding skills. Presentation is required. Failure to present results in 0 point for this part. * Grading: homework 30%, mid-term exam: 30%, term project: 40%.

• Python and numpy

  organize Python modules, use numpy for array-oriented code, use tools for numerical analysis

• C++ and computer architecture

  fundamental data types, object-oriented programming, standard template library

• fundamental engineering practices

  automation, version control and regression, work that cannot be automated

• memory management

  stack and heap, Linux and POSIX API, C++ memory manager, object and memory statistics

• matrix operations

  POD arrays and majoring, matrix-vector and matrix-matrix operations, linear algebra

• cache optimization

  cache for performance, stride analysis, tiling

• SIMD: vector processing

  Single instruction and multiple data (SIMD) with modern hardware, compiler options to generate SIMD assemblies, SIMD-friendly memory layout

• modern C++ I: ownership and meta-programming

  pointers and ownership, revisit shared pointer, template meta-programming and compile-time computing

• modern C++ II: more than templates

  copy elision / RVO, move semantics, variadic template and perfect forwarding, anonymous function, closure, lambda expression

• xtensor: arrays in C++

  xtensor, carry arrays between Python and C++, speed up array-based Python code using C++

• pybind11: binding between Python and C++

  wrapping API, Python objects, reference counting, Python containers

• cpython API: operate Python from C

  built-in types, memory manager, threading and global interpreter lock

• profiling

  avoid memory allocation and data copy, home brew profiling, assembly inspection

• array-oriented design

  design interface with arrays, struct arrays between Python and C++, conversion between dynamic and static semantics

Textbook: None References: * Computer Systems: A Programmer's Perspective: https://csapp.cs.cmu.edu/ * Python documentation: https://docs.python.org/3/ * Cppreference: https://en.cppreference.com/ * Effective Modern C++, Scott Meyer, O'Reilly, 2014 * Source code: cpython, numpy, xtensor, and pybind11

地點

By appointment

時間

N/A

聯絡方式

Send email to yyc at solvcon.net, with "[nsd]" prefix in the subject line.