Likelihood contour scans in 1D/2D #14
Description
Contour scans using constrained minimizers to constrain the -2 delta log(L) = n^2*sigma and scan the contour. This reduces the dimensionality by 1.
- In a 1D likelihood contour scan it is just a fit to 2 points to minimize/maximize the parameter to be scanned. In 2D it would be a 1D scan.
- In 2D the scan would be done by fixing the angle and maximizing the radius and then iterating the angle through 360 degree.
- Ideally we make use of the hessian to define initial value.
- In 2D we can use the correlation and scan on ellipse.