| Partha
Solapurkar |
Why is an 'almost everywhere continuous' function Riemann integrable? |
||
| Arun
Iyer |
the way i see it is "that particular thing is because how riemann integrals are defined" (what follows are my ideas and may vary with the original text explanations) since we deal with intervals here and then reduce the length of interval so as to make that rectangular strip very very thin. so when we think of a discontinuous function (which means there is a break in the graph) , we see that reducing the length of that rectangular strip creates a problem for us at the point of discontinuity. hopefully, whatever i said made sense there. love arun |
||
| Partha
Solapurkar |
Is the reverse implication true?ie a bounded function is Riemann integrable iff it is almost everywhere continuous. -Partha |
||
| David
Loeffler |
I think that's true, but I can't remember the proof - if nobody else can recall it I might go and look it up. David |
||
|
Michael Doré
|
The proof is not too hard. Let be bounded and defined (say) on [0,1]. Suppose that is continuous a.e. Let . We know that for almost every there exists such that . Let be the set of such that works. We see that is increasing (that is is contained in for all ) and the union of has measure 1. Fix . Since measure is a continuous set function we have (here stands for the measure of the set ) so there exists such that . In other words for any and any we can find an such that the set of such that: (*) has measure at least . Now take a dissection of [0,1] into intervals of length . The sum of the lengths of intervals which have an with the property (*) certainly is at least - since their union contains the set of with property (*) which has measure . In each of these intervals there exists with the property (*) so varies by less than in each of these intervals. Now let's estimate the difference between the upper and lower sum. What about the contribution of the intervals which have an with (*)? Well varies by less than in each of these intervals and the sum of the lengths of these intervals cannot exceed 1 so the maximum contribution to the difference between the lower and upper sum is . What about the contribution of the intervals which don't have an with property (*)? Well the sum of the lengths of these intervals is at most and varies by no more than in these intervals. (Where , .) Hence the maximum contribution is . Hence the difference between the lower and upper sum is at most . To recap: given and we can find such that any dissection whose intervals have size has the property that difference between the lower and upper sums is . This can be made arbitrarily small by taking small and close to 1. So we're done one way. Suppose is not continuous a.e. Let be the set of such that: there doesn't exist with . The union of the is the set on which is discontinuous. Since their union does not have measure zero there exists an such that does not have measure zero. In particular there exists a such that cannot be covered by disjoint intervals whose lengths sum to . Dissect [0,1] into intervals of length . must intersect at least of these intervals (not counting intersections at endpoints). varies by at least on these intervals. So the difference between the upper and lower sums is , for any . If were Riemann integrable then the difference would tend to zero as which it doesn't. Hence is not Riemann integrable. So not continuous a.e. not Riemann integrable so we're done this way as well. |