"plus" and "cross" shear correlation functions? about treecorr

[rmjarvis]

I'm not sure exactly what you want, but the xi+ and xi- are defined as xi+ = <g_+ g_+ + g_x g_x> and xi- = <g_+ g_+ - g_x g_x>. So you can recover the separate components as <g_+ g_+> = (xi+ + xi-)/2 and <g_x g_x> = (xi+ - xi-)/2.

However, these g_+ and g_x values are defined relative to the line connecting the pairs of galaxies. So they are not what you would get by setting either g1 or g2 to zero. So is this what you had in mind? Or did you want something like <g1 g1> directly without any rotations applied?

Also, I'm curious what use case you have for these + and x correlations separately. I don't know of any case where those statistics are more informative than xi+ and xi-. The latter really are the natural formulations of the shear correlation function.

[chazshapiro]

I need and (and perhaps ). If I understand the code usage correctly, when I process a catalog of (X, Y, g1, g2), g1 would be the plus shear component defined in the coordinate system for (X,Y) so that g1>0 is an elongation along the X direction.

The use case is to study detector-induced shape measurement systematics. When X and Y are aligned with the pixel grid, we naturally find much larger biases in g1.

[rmjarvis]

I suspect one point statistics would be more diagnostic in your case. (i.e. just <g1>, not <g1 g1>.) But if you really wan the two point correlation of each of the shear components separately, treating them as scalars rather than a spinor, you can get that with TreeCorr using KKCorrelation. It computes the two-point correlation function of a scalar field. In you case, you can separately run with k=g1 and k=g2.

[chazshapiro]

that works, thanks!

---

Sent from Gmail