06.06.2013 Views

STOCHASTIC

STOCHASTIC

STOCHASTIC

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

CHOOSING INVESTMENT PORTFOLIOS<br />

Hence Xf(x l ) + (1 -A)/(x 2 ) >f{Xx l + (1 -A)* 2 }.<br />

Thus/is strictly convex unless a and b are proportional. But this requires<br />

that there exist constants qx and q2 not both zero such that q^a = q2b or<br />

that<br />

qlXSixi l = q2{\-X)Sixi 2 => qiW = 92(1-A)xj 2<br />

(since St > 0)<br />

or that x, 1 and x, 2 are proportional since qx' = q^X and q2' = q2(l—X) are<br />

both not zero. However, the condition that £"=i x, 1 and £?=1 x; 2 = 1 means<br />

that x 1 and x 2 cannot be proportional unless x, 1 = x; 2 for all i, which is a<br />

contradiction.<br />

Remark f is convex but not strictly convex on M = {x \ x ^ 0} because /<br />

is linear on every ray that contains the origin.<br />

Lemma 2 q> is a concave function of /? > /?t, where pL > 0 is defined below,<br />

if 2^ 1 and S, ^ 0, i= l,...,n.<br />

Proof Let Kfi = {& \ e'x = 1, x ^ 0, /(x) ^ /?}. Let pL> 0 be the smallest<br />

0 for which A^ ^ 0. Clearly J?L ^

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

Saved successfully!

Ooh no, something went wrong!