phantom power
(1000+ posts)
Send PM |
Profile |
Ignore
|
Fri Jan-18-08 12:19 PM
Response to Reply #8 |
9. the variance of a random variable is different than its range. |
|
And I'm not sure what you're saying about digital communications. It's possible to compensate for any error rate, but your transmission rate goes down as a consequence.
None of that really bears on the fact that if you sum uncorrelated random variables, the variance of your sum goes up. As caraher points out, your mean also sums, so it's possible to compensate for that increased variance, but at the cost of additional redundancy: put another way, your peak output becomes much higher than it needs to be, so that the probability of dropping below your desired output remains sufficiently low.
|