Analysis is easy to use reduction to absurdity.
Let's set the midpoint C(xo, 0) of AB,
Then xo=(x 1+x2)/2, from P+Q = 1, Q >;; =p, and both p and q are positive real numbers, so it is easy to get 0.
Then (px1+qx2)-XO = px1+(1-p) x2-(x1+x2)/2 = (x1-x2) (2p-/kloc. =0,(0 & ltx 1 & lt; x2)
Get (px 1+qx2) >: =xo .
The first derivative g'(x)=2/x-2x-a is obtained by g (x) = 2lnx-x 2-ax.
Then take the derivative of g'(x) and get its second derivative g "(x) =-2/x2-2.
Knowing that g'(x) is in x>0, and g' (px 1+qx2) < =g'(xo),
So it is necessary to prove G' (px 1+qx2).
Let's prove it by reducing to absurdity
Suppose g' (XO) >; =0 holds.
Combine what is known to be available
2lnx 1-x 1^2-ax 1=0.....( 1),
2lnx2-x2^2-ax2=0......(2),
2/XO-2xo-a & gt; =0......(3),
xo=(x 1+x2)/2......(4),
There are 0
ln(x2/x 1)-2(x2-x 1)/(x2+x 1)& lt; =0,
That is ln (x2/x1)-2 [(x2/x1)-1]/[(x2/x1)+1]
Let x2/x 1 = t, (t >;; 1) and h (t) = LNT-2 (t-1)/(t+1), (t >;; 1)
Derivation is easy to get H' (t) = (t-1) 2/[t (t+1) 2] > 0, (t> 1).
Then h (t) at t > 1 increases monotonically, and h(t) can be continuous when t= 1.
So h (t) gt; H( 1)=0, (t> 1) is LNT-2 (t-1)/(t+1) > 0.
That is ln (x2/x1)-2 [(x2/x1)-1]/[(x2/x1)+1] > 0.
But it contradicts the formula (5), so g' (XO) >: Hypothesis =0 is not valid.
Then there is g' (XO) < 0, so g' (px 1+qx2).
Therefore g' (px 1+qx2)
comment