# Horizontal Morphing Demo

Sven Kreiss, Kyle Cranmer

June 2013

Drag the sliders.

## Small KD-trees

When walking the tree, it is easy to keep track of the volume of the rectangles. The opacity of the rectangles is proportional to the inverse of the volume. As it turns out, this might be a poor approximation as a probability density, so something else like KEYS needs to be layered on top.

\(m_H\):

## Two Gaussians

These are 2x 200 points drawn from two Gaussians. For small mH, the Gaussian is on the left and narrow along m34 / MEKD, and for larger mH, the Gaussian is on the right and is wider in m34 / MEKD.

\(m_H\):

## Four Gaussians, Two Parameters

These are 4x 200 points drawn from four Gaussians, two narrow and two broad. This is supposed to mimic the situation of shifting shapes and event-by-event errors.

\(m_H\):

\(m_{4l,err}\):