Show description

M. , defined in Exercise 6 of Sec. 2. *25. Let/be measurable with respect to J*\ and Z be contained in a null set. Define (/ onZ% J \κ οηΖ, where K is a constant. Then/is measurable with respect to & provided that (Ω, F, &) is complete. Show that the conclusion may be false otherwise. 1 Random variable. Expectation. Independence General definitions Let the probability space (Ω, J*;^) be given. ^ 1 = ( — oo, +00) the (finite) real line, «^* = [ — oo, +00] the extended real line, &x = the Euclidean Borel field on St1, @* = the extended Borel field.

Example 2. or not): Let <%η be the «-dimensional cube (immaterial whether it is closed <%" = { ( * ! , . . , xn) : 0 < *, < 1 ; 1 < j < n}. η The trace of (β , &η, mn), where &n is the «-dimensional Euclidean space, &n and mn the usual Borel field and measure on the cube, is a probability space. m. mn on °lln is a product measure having the property analogous to (8). ,Xn)) Then {Xh 1 < j the n coordinate dent variables" The two usages =fj(Xj). 's. In particular if/}(*/) = * j , we obtain variables in the cube.

I i Having thus defined the measure for all open sets, we find that its values for all closed sets are thereby also determined by property (vi) of a probability measure. In particular, its value for each singleton {a} is determined to be F(a) — F(a — ) , which is nothing but the jump of F at a. Now we also know its value on all countable sets, and so on—all this provided that no contradiction is ever forced on us so far as we have gone. F. US1. The next step will be the G6 sets and the Fa sets, and there already the picture is not so clear.