Damn, it's too roundabout here. There is a sentence in the textbook I use-"Partial derivative is a special case of directional derivative". . . . . . NND, I've thought about it again. Is this correct? The parameter t in the directional derivative (tcosa, tcosb) may indeed be positive. Does it mean the length in a certain direction? Thus: Z = radical sign (x 2+y 2): (1) The derivative of x corresponds to the directional derivatives of "two" directions-x0+direction and X0- direction. In the process of finding the derivative of x according to the definition, that limit expression is 65438 in X0+ and X0- respectively. (2) The limit values "1,-1" when defining the x derivative here actually correspond to the directional derivatives in two directions-if the direction is (tcosa, tcosb)-corresponding to a = 0, b = 0 and a = 6544. B = 0 (geometrically, the two sides of the cone 1 and-1 correspond to the directional derivatives of "two directions" respectively), then it can be concluded that if and only if x has a derivative, it is a special case of directional derivative-otherwise, there may be cases where the directional derivative exists and the x derivative does not exist. -Is this understanding correct? Who can answer the fucking question? I was really moved by the emperor []