Solver slow when using BigFloat
shakisparki opened this issue · comments
Shakir commented
I found out that in scenarios where I need more machine precision in order to meet tolerance values, and I use the BigFloat datatype, the solver becomes much much slower (takes longer for each iteration) especially as the number of unknowns increases.
Antoine Levitt commented
BigFloats are slow, yes. Not much nlsolve can do there.
Patrick Kofod Mogensen commented
Try ArbFloat, double-double approaches or similar. I can write a tutorial. I don’t know for a fact that all extended precision types work, but if they don’t, I’ll consider it a bug.
Shakir commented
OK, I will look into using either. Thanks