In the LeibnizโNewton calculus, one learns the di?erentiation and integration of deterministic functions. A basic theorem in di?erentiation is the chain rule, which gives the derivative of a composite of two di?erentiable functions. The chain rule, when written in an inde?nite integral form, yields the method of substitution. In advanced calculus, the RiemannโStieltjes integral is de?ned through the same procedure of โpartition-evaluation-summation-limitโ as in the Riemann integral. In dealing with random functions such as functions of a Brownian motion, the chain rule for the LeibnizโNewton calculus breaks down. A Brownian motionmovessorapidlyandirregularlythatalmostallofitssamplepathsare nowhere di?erentiable. Thus we cannot di?erentiate functions of a Brownian motion in the same way as in the LeibnizโNewton calculus. In 1944 Kiyosi Itห o published the celebrated paper โStochastic Integralโ in the Proceedings of the Imperial Academy (Tokyo). It was the beginning of the Itห o calculus, the counterpart of the LeibnizโNewton calculus for random functions. In this six-page paper, Itห o introduced the stochastic integral and a formula, known since then as Itห oโs formula. The Itห o formula is the chain rule for the Itหocalculus.Butitcannotbe expressed as in the LeibnizโNewton calculus in terms of derivatives, since a Brownian motion path is nowhere di?erentiable. The Itห o formula can be interpreted only in the integral form. Moreover, there is an additional term in the formula, called the Itห o correction term, resulting from the nonzero quadratic variation of a Brownian motion.
ืขืกืงืื ืืืฉืงืขืืช