Hi Shivam,
Given the issue you're facing with NaN values in your matrices (hne, hnw, hns, etc.) and zeroes in your other matrices (term1, term2, term3, etc.), it seems there might be a problem with either the initialization of your variables or the calculations that involve divisions or square roots, which can easily lead to NaN values if not handled properly.
- Initialization Check: Ensure all variables are initialized properly before they are used in any operations that could generate NaN values. For instance, dividing by zero or taking the square root of a negative number can result in NaN.
- Boundary Conditions: When implementing numerical methods, boundary conditions are critical. Incorrect handling of boundary conditions can lead to NaN values. Ensure that your boundary condition logic (e.g., when im == 0 or ip == Nx+1) does not lead to invalid array indexing or division by zero.
- Division by Zero: Check all your division operations to ensure you're not dividing by zero. For example, when calculating gradients or other terms, adding a small number (like eps in MATLAB) to the denominator can prevent division by zero errors.
- Square Roots and Logarithms: If your equations involve square roots or logarithms, ensure the arguments to these functions are always valid (e.g., non-negative for square roots).
Debugging Zero Matrices
If your matrices are ending up with zero values, it might be due to:
- Initial Conditions: Verify that the initial conditions (initial values of phi, T, etc.) are set correctly and are not trivial (e.g., not all zeros if your system's evolution depends on non-zero initial states).
- Overwriting Data: Ensure you're not accidentally overwriting your data with zeros due to incorrect loop logic or variable assignments.
- Equation Implementation: Double-check the implementation of your equations. There might be a mistake in translating the equations to code, especially in complex term calculations. Make sure all terms are calculated as intended according to your equations.
General Suggestions
- Add Debugging Statements: Temporarily insert disp or fprintf statements in your code to print intermediate variables and check where they might be going wrong. This can help you identify the exact point in your code where NaN or unexpected zero values start appearing.
- Simplify: Temporarily simplify your model by reducing it to a simpler case (e.g., fewer equations, simplified geometry) that you can verify manually. Once the simpler case works as expected, gradually add complexity back.
- Plotting: Use plots to visualize the evolution of your variables (phi, T, etc.) at each timestep. This can sometimes give you insights into what might be going wrong.
I hope it helps
