09.12.2012 Views

Concrete mathematics : a foundation for computer science

Concrete mathematics : a foundation for computer science

Concrete mathematics : a foundation for computer science

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

9.3 0 MANIPULATION 441<br />

(We are expanding everything out until we get a relative error of O(nP3),<br />

because the relative error of a product is the sum of the relative errors of the<br />

individual factors. All of the O(nP3) terms will coalesce.)<br />

In order to expand (1 - nP’)n, we first compute ln(1 - nP’ ) and then<br />

<strong>for</strong>m the exponential, enln(‘Pnm’l:<br />

(1 - nP’)n = exp(nln(1 -n-l))<br />

= exp(n(-nP’ - in-’ - in3 + O(nP4)))<br />

= exp(-1 - in-’ - in2 + O(nP3))<br />

= exp(-1) . exp(-in-‘) . exp(-$n2) . exp(O(nP3))<br />

= exp(-1) . (1 - in-’ + in2 + O(nP3))<br />

. (1 - in2 + O(nP4)) . (1 + O(nP3))<br />

= e-l (1 - in-’ - $ne2 + O(nP3)) .<br />

Here we use the notation expz instead of e’, since it allows us to work with<br />

a complicated exponent on the main line of the <strong>for</strong>mula instead of in the<br />

superscript position. We must expand ln(1 -n’) with absolute error O(ne4)<br />

in order to end with a relative error of O(nP3), because the logarithm is being<br />

multiplied by n.<br />

The right-hand side of (9.41) has now been reduced to fi times<br />

n+‘/e” times a product of several factors:<br />

(1 - in-’ - AnP2 + O(nP3))<br />

. (1 + n-l -t nP2 + O(nP3))<br />

. (1 - in-’ - &nP2 + O(nP3))<br />

. (1 + an-’ + (a + b)nP2 + O(nP3)) .<br />

Multiplying these out and absorbing all asymptotic terms into one O(n-3)<br />

yields<br />

l+an’+(a$-b-&)nP2+O(nP3).<br />

Hmmm; we were hoping to get 1 + an’ + bn2 + O(nP3), since that’s what<br />

we need to match the right-hand side of (9.40). Has something gone awry?<br />

No, everything is fine; Table 438 tells us that a = A, hence a + b - & = b.<br />

This perturbation argument doesn’t prove the validity of Stirling’s approximation,<br />

but it does prove something: It proves that <strong>for</strong>mula (9.40) cannot<br />

be valid unless a = A. If we had replaced the O(nA3) in (9.40) by<br />

cne3 + O(nP4) and carried out our calculations to a relative error of O(nP4),<br />

we could have deduced that b = A. (This is not the easiest way to determine<br />

the values of a and b, but it works.)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!