Hello SymPy Community, I am Sai Kumar, and I propose enhancing SymPy's code generation capabilities to efficiently compute Jacobians and Hessians for optimization and ODE integration. This project will bridge the gap between symbolic and numerical computing, enabling users to seamlessly integrate SymPy with SciPy's optimization and ODE solvers. To extend SymPy’s capabilities, I propose implementing the following improvements:
1 . Efficient Jacobian and Hessian Evaluation: - Develop functions like generate_minimize_derivative_funcs to generate numerical functions, Jacobians, and Hessians from symbolic expressions. - Optimize performance by reusing common subexpressions and leveraging SymPy's symbolic differentiation. 2 . ODE Integration Support: - Implement generate_ode_derivative_funcs to generate numerical functions and Jacobians for ODE systems. - Add sparsity pattern detection for large-scale ODE systems to improve computational efficiency. 3 . Seamless Integration with SciPy: - Ensure compatibility with SciPy's optimization (minimize) and ODE (solve_ivp) solvers. - Provide clear documentation and examples for users to easily integrate SymPy with SciPy. 4 . Performance Optimization: - Benchmark the new functionality against existing tools like pyodesys and symopt. - Optimize code generation to minimize computational overhead for large-scale problems. These enhancements will make SymPy more versatile and efficient for optimization and ODE integration, benefiting researchers, engineers, and students working on complex numerical problems. I would love to discuss this further and receive feedback from the community on how best to approach these improvements. Looking forward to your insights! Best regards, Sai Kumar New chat -- You received this message because you are subscribed to the Google Groups "sympy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/sympy/fd7b6f58-8a34-4ba1-9fae-7bc18cb37734n%40googlegroups.com.
