Page 201 - Foundations Of Differential Calculus
P. 201

184    9. On Differential Equations
        We must very carefully separate these kinds of equations into imaginary
        and real, especially in integral calculus, since it would be ridiculous to seek
        an integral for a differential equation, that is, a finite equation that it would
        satisfy, when it is clear that none exists.
        310. In the first place, then, it is clear that every differential equation
        with three variables in which only two differentials occur is imaginary and
        absurd. Let us consider an equation that contains the variable z but only
        the differentials dx and dy, the differential dz being completely absent.
        It is obvious that no function of x and y can be exhibited that can be
        substituted for z to produce an identical equation. Indeed, the differentials
        dx and dy can in no way be removed. In these cases there is absolutely
        no satisfactory finite equation, unless perhaps it is possible to assign a
        relationship between x and y that persists no matter what z might be. For
        example, in the equation
                             zdy − zdx = ydy − x dx,

        which is satisfied by y = x. It is easy to investigate the cases in which this
        happens by looking for a relationship between x and y, first when z =0,
        and then whether the same relationship persists when z has an arbitrary
        value.

        311. Nor is it the case that an equation is absurd only if it involves three
        variables and two differentials. It can be absurd even if all three differentials
        are present. In order to show this let P and Q be functions of only x and
        y and consider the equation

                                 dz = Pdx + Qdy.
        If this equation is not to be absurd, then z is some function of x and y whose
        differential is dz = pdx + qdy so that P = p and Q = q. However, we have
        demonstrated (paragraph 232) that pdx + qdy cannot be the differential
        of any function of x and y unless
                                    ∂p    ∂q
                                       =    .
                                    ∂y    ∂x
        Here the notation means, as we previously assumed, that ∂p/∂y is the
        differential of p with only y variable, divided by dy, and ∂q/∂x is the
        differential of q, with only x variable, divided by dx. Hence, the equation
        dz = Pdx + Qdy cannot be real unless
                                    ∂P    ∂Q
                                       =     .
                                    ∂y    ∂x

        312. Absolutely the same reasoning applies to the equation
                                dZ = Pdx + Qdy,
   196   197   198   199   200   201   202   203   204   205   206