Open
Description
I'm trying to implement routines to solve systems of nonlinear equations. However, when I numerically evaluate objects created with the Lambdify function, the evaluation returns NaN or inf, depending on the case. However, when I evaluate the function using myfunction.subs(), it evaluates correctly. It is worth noting that in my implementation, Lambdify is essential since I evaluate the same function thousands of times in the process, and using .subs() is extremely slow.
It possible to use a backend that improves numerical precision? Is there any configuration that can force Lambdify to operate with higher precision numbers in its internal calculations?
Metadata
Metadata
Assignees
Labels
No labels