Generals 2000 II 2
From PlasmaWiki

Writing this as:

With:
![\phi\left(w,t\right)=i\lambda\left[\ln\left(w\right)-\ln\left(1-w\right)-4wt^{2}\right]](http://localhost/mediawiki/images/math/4/b/1/4b159a7da27cb30dfe3fc1605f30b446.png)
Setting the derivative to zero to find saddles:

Rearranging:


Solving for ws:

Which we will now call
.
[edit] t<<1 limit
The saddle points are off the axis. Looking at the second derivative:
![\phi^{\prime\prime}\left(w\right)=i\lambda\left[-\frac{1}{w^{2}}+\frac{1}{\left(1-w\right)^{2}}\right]](http://localhost/mediawiki/images/math/8/a/0/8a0c27540a0888c8b857543fe6b54a40.png)
So that:

Using
we taylor expand:
![\phi^{\prime\prime}\left(w_{+}\right)=i\lambda\left[2t^{2}\left(1+2it-2t^{2}\right)-2t^{2}\left(1-2it-2t^{2}\right)\right]\approx-8\lambda t^{3}](http://localhost/mediawiki/images/math/1/d/9/1d9d46e53075d61a91d3727a4eac0f26.png)
So that the direction
is along the real axis. For the other direction:

Using
we taylor expand:
![\phi^{\prime\prime}\left(w_{-}\right)=i\lambda\left[2t^{2}\left(1-2it-2t^{2}\right)-2t^{2}\left(1+2it-2t^{2}\right)\right]\approx8\lambda t^{3}](http://localhost/mediawiki/images/math/f/e/3/fe3e349ea2993a387d63942ab59138f3.png)
So that the direction
is purely imaginary and cannot be used. We then get from the saddle
at w + :

Expressing
approximately by taking
:
![\phi\left(w_{+},t\right)=i\lambda\left[\ln\left(\frac{1+\sqrt{1-t^{-2}}}{1-\sqrt{1-t^{-2}}}\right)-4\left(1+\sqrt{1-t^{-2}}\right)t^{2}\right]\approx i\lambda\left[\ln\left(-1\right)-4it\right]=4\lambda t-\lambda\pi](http://localhost/mediawiki/images/math/5/b/3/5b3ac64a836b3b853f61d34188530ac7.png)
Putting this in:

[edit] t>>1 limit
For this limit, the saddle points are on the real axis. Finding the
second derivative and using
:

So the direction of integration is
.
For the other saddle point:

So the direction of integration is
.
We integrate through both saddles. Evaluating φ at the first
saddle, with
:
![\phi\left(w_{+},t\right)=i\lambda\left[\ln\left(\frac{1+\sqrt{1-t^{-2}}}{1-\sqrt{1-t^{-2}}}\right)-4\left(1+\sqrt{1-t^{-2}}\right)t^{2}\right]\approx i\lambda\left[\ln\left(\frac{2}{t^{-2}/2}\right)-8t^{2}\right]=2i\lambda\ln\left(2t\right)+8i\lambda t^{2}](http://localhost/mediawiki/images/math/1/c/a/1ca37b681b98b959be68d14e8987b958.png)
So this is just a phase factor. At the other saddle:
![\phi\left(w_{-},t\right)=i\lambda\left[\ln\left(\frac{1-\sqrt{1-t^{-2}}}{1+\sqrt{1-t^{-2}}}\right)-4\left(1+\sqrt{1-t^{-2}}\right)t^{2}\right]\approx i\lambda\left[\ln\left(\frac{t^{-2}/2}{2}\right)-8t^{2}\right]=-2i\lambda\ln\left(t/2\right)+8i\lambda t^{2}](http://localhost/mediawiki/images/math/b/2/9/b29b3a37800f7849e48aeca3f74f85b4.png)
So this is another phase factor. Then the saddles contributions are:

Plugging in:

Which simplifies to:


