Or, if the given integral is ƒ (+) dy da, then if we re a place x and y by aέ and bŋ respectively, and substitute for έ and n in terms of u and v, the integral becomes 216.] Although the limits of the first integration of a multiple integral are functions of one or more of the variables, in reference to which subsequent integrations are to be effected, yet it is possible, and frequently convenient, so to change the variables, that the limits of the first integration may be constant. The mode of effecting this transformation is explained in Art. 93. Let us take the case of a double integral; and suppose it be and observing that t=1, when y = y1; and t = 0, when y = y; we have x1 1 wherein the limits of the first integration are constant. Thus, for example, if I = ((1+x2) § e-(a2x2+b2y2) dy dx; 0 (1 + x2 + y2) 1 (62) let y = (1+x2)1t; so that t=1, when y = (1+x2); and t = 0, when y = 0; then SECTION 3.-The Differentiation of a Definite Multiple Integral with respect to a Variable Parameter. 217.] If a parameter, capable of variation, is contained in the element-function of a definite multiple integral, and also in some or all of the limits of integration, the integral is a function of that parameter; so that the value of the integral will change, if the parameter varies. Let us suppose the parameter continuously to vary, within such limits however that the element-function of the integral does not become infinite or discontinuous; then the value of the integral will also continuously vary; and it is the increment of the integral, due to the infinitesimal variation of the parameter, which it is our object to determine in the present section. In Art. 96, we have investigated the change of value of a definite single integral, due to the infinitesimal variation of a parameter, of which the limits as well as the element-function are functions; and we have shewn that if a is the variable parameter, and D denotes the total differential, Now this will be the foundation of our subsequent investigations. Let us first consider a definite double integral; viz. and where x, xo, yn, Yo, as also o, are all functions of a; a being a parameter capable of continuous variation. We have to determine the change of value in u due to an infinitesimal variation of a. In (64) let F(a, c) be replaced by "F (a, x, y) dy; that is, by and in the first term of the right-hand member of (67), inasmuch dx as does not contain y, putting it under the sign of the y-inda tegration for symmetry of expression, we have which gives the total variation of the definite double integral. Again, if the similar variation of a definite triple integral is required, let a in (68) be replaced by Sadz, where the new a is F(a, x, y, z); then by an exactly similar an exactly similar process it may be shown Yn ZA dx da ΤΟ · [** ["* [ a dz]" dy dx + [* ["* f* da dz dy dx. (69) + da. 20 Το 30 20 da Now the law of formation of the several terms in the second member of this equation is sufficiently obvious, so that the variation of a definite multiple integral may be easily expressed at length; and the general law may be proved inductively; viz. by assuming the truth of it for a multiple integral of the (n−1)th order, and deducing from that the same law for the multiple of the nth order. This process is so easy that it is unnecessary to express it at length. 218.] Definite multiple integrals also, when one integration has been effected, and the limits of that integration are functions of a variable parameter, are subject to variation by reason of the change of that variable parameter. The consequent variable of the reduced integral may be derived from the theorem given in (64). The following are examples of the process. In (64) let r (a, x) be replaced by [a]"; then D da. da dy *; dx (70) Yn da dy ]* dr. (71) S**[a]"dx = [[a Again, let a be replaced by s ̈ da da da dy ΤΟ a dz, so that dz [nda]+ must be replaced by [2] consequently [ady] + ZO S 20 dzi dzi do dy Again, for another example, from (70), we have Уо dx ૨ ૧ ૨ da da) dy]**. (73) The variation of all other similar integrals may be determined in a similar manner, the theorems contained in (64) and in (70) being of general application; and it is unnecessary to exemplify them further at present. We shall hereafter have occasion to apply the preceding process. |