We consider the 2-variable scalar function f(x)=exp(x1)(x^2_1 +x^2_2)^3 generate the computational graph of this expression...

 

Transcribed Image Text:

We consider the 2-variable scalar function f(x)=exp(x1)(x^2_1 +x^2_2)^3 • generate the computational graph of this expression (graph starts with c1 = x1, c2 = x2, and ends with f = c7) • use forward mode differentiation to compute Vf (0, 2). Show your intermediate values on the graph. • use reverse mode differentiation to compute the same gradient. Show your intermediate values (values of the adjoint variables cī = af/aci). • use the backward() method of pytorch to verify your answers. • comment briefly on the computational effectiveness of forward vs reverse modes of computing these derivatives. We consider the 2-variable scalar function f(x)=exp(x1)(x^2_1 +x^2_2)^3 • generate the computational graph of this expression (graph starts with c1 = x1, c2 = x2, and ends with f = c7) • use forward mode differentiation to compute Vf (0, 2). Show your intermediate values on the graph. • use reverse mode differentiation to compute the same gradient. Show your intermediate values (values of the adjoint variables cī = af/aci). • use the backward() method of pytorch to verify your answers. • comment briefly on the computational effectiveness of forward vs reverse modes of computing these derivatives.

Answer rating: 100% (QA)

Lets break down the steps to compute the gradient of the given function fxexpx1x12x223 at the point 0 2 using both forward mode differentiation and re... View the full answer