I really want to have a generalised differentiation method for functions whose output is a vector. This is really important in practice. For exemple: f(v) = h*v b(v, w) = f(v) * cos(w) + d - sqrt(d**2 - f(v)**2 * sin(w)**2) F(x, y, theta, v, w, a) = (x + b(v,w) * cos(theta), y + b(v,w) * sin(theta), theta + arcsin(sin(w) * f(v)/d), v + h*a)
I can do first order differentiation naturally and keep the type of symbolic expression: jacobian(F, (x,y,theta,v,w,a)) which the output is still a symbolic expression. But I cannot do F.hessian() since F is not a scalar valued function. I cannot do this G(x, y, theta, v, w, a) = (F[i].hessian() for i in range(4)) TypeError: unable to convert <generator object <genexpr> at 0x6fdc93dab40> to a symbolic expression because the parser this these parentheses are a generator. Of course I can do this G(x, y, theta, v, w, a) = [F[i].hessian() for i in range(4)] But it will lose all the good properties of symbolic expressions which are differentiable: jacobian(G,(x, y, theta, v, w, a)) will give all zeros since the output is a list. Last of all, I can however solve this problem by doing some ugly transformations and combine with np functions. But it will really be a big improvement if I can just press a button like diff(F,2) and even 3! -- You received this message because you are subscribed to the Google Groups "sage-support" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-support+unsubscr...@googlegroups.com. To post to this group, send email to sage-support@googlegroups.com. Visit this group at https://groups.google.com/group/sage-support. For more options, visit https://groups.google.com/d/optout.
