A very simple implementation of symbolic differentiation.
from symdiff import Variable, diff, grad
x = Variable("x")
y = Variable("y")
def f(x, y):
return x - y ** x
expr_df_dx = str(diff(f(x, y), x))
expr_df_dy = str(diff(f(x, y), y))
print(expr_df_dx)
print(expr_df_dy)
df_dx = grad(f, 0)
df_dy = grad(f, 1)
print(df_dx(1.0, 2.0), df_dy(1.0, 2.0))output:
(1 - ((y ** x) * ((1 * log(y)) + ((0 * x) / y))))
(0 - ((y ** x) * ((0 * log(y)) + ((1 * x) / y))))
-0.3862943611198906 -1.0