weiming-hu / linearsystemsolvers

This repository is created for Penn State course CSE 597 Implementation of Parallel Computer Codes

Hi there 👋

My name is Weiming Hu. I'm a researcher in ⚡ Renewable Energy ⚡ and ☁️ Weather Forecasting ☁️ .

I'm also an enthusiast for free software and the open-source movement.

• 🚧 I'm working on 🚧

  • Developing R and Python packages for the scientific community
  • Becoming a Bangle.JS contributor.

• 💬 Ask me about 💬

  • R, Python, C++, Supercomputers

• 📫 How to reach me 📫

  • My personal website

linearsystemsolvers's Issues

Iterative methods only work for diagonally dominant matrices

Describe the bug
Both the Jacobi and the Gauss-Seidel iterative methods only work for diagonally dominant matrices. Although being diagonally dominant is one of the assumptions for the iterative methods, is there still an iterative solution to them?

To Reproduce
Steps to reproduce the behavior:

1. git clone the repository
2. Compile the program

mkdir build && cd build
cmake ..
make

3. Go to the binary output folder

cd ../output/bin

4. Run the successful test with Jacobi method for 100 iterations

./iterativeSolver J ../../data/A_5.csv ../../data/b_5.csv 100 1 2

5. Print out the correct result

cat ../../data/x_5.csv # they should be the same with the above

6. Run the failing test with Jacobi method for 100 iterations

./iterativeSolver J ../../data/A_failed.csv ../../data/b_failed.csv 100 1 2
# the residual is diverging

7. Print out the correct result

cat ../../data/x_failed.csv
# they are different from the above

Expected behavior
Jacobi method should be converging but it is not. The residual gradually diverges rather than becoming a tremendously large number suddenly suggesting that this is less of a numerical precision problem, but more of a systematic bug issue.

Desktop (please complete the following information):

• OS: Ubuntu 18.04.1 LTS (Bionic Beaver)
• Browser Chrome