The program could be much improved if it was rewritten from scratch.
We could write it directly for numerical computations. Then we would not have to use formal series, or to ensure that we get Kac tables with rational (as opposed to real) conformal dimensions. We would
not use SymPy at all, and avoid having to convert SymPy expressions into numbers.
We have several separate classes for computing four-point functions in various models, but one class might be enough. We have a wrapper class Data which does not much and could conceivably be fused with the class which would compute four-point functions.
Julia can apparently do splines, latex displays, integration of complex functions. Apparently it manages complex numbers better than Python. It would be nice if it really led to faster calculations.
The program could be much improved if it was rewritten from scratch.
We could write it directly for numerical computations. Then we would not have to use formal series, or to ensure that we get Kac tables with rational (as opposed to real) conformal dimensions. We would
not use SymPy at all, and avoid having to convert SymPy expressions into numbers.
We have several separate classes for computing four-point functions in various models, but one class might be enough. We have a wrapper class Data which does not much and could conceivably be fused with the class which would compute four-point functions.
Julia can apparently do splines, latex displays, integration of complex functions. Apparently it manages complex numbers better than Python. It would be nice if it really led to faster calculations.