dncnwtts / astro-stomp Goto Github PK

unit tests for most of the python port of the library have yet to be written

On two linux systems using g++ (Ubuntu & CentOS), the following compilation
error occured.  Previous versions of astro-stomp (r50) compiled fine on
both systems.

% make 
.. SUCCESS UNTIL ...
/bin/bash ../libtool --tag=CXX   --mode=link g++  -g -O2 -Wall  -lstomp  -o
stomp_adapt_map stomp_adapt_map.o  -L../stomp/ 
mkdir .libs
g++ -g -O2 -Wall -o .libs/stomp_adapt_map stomp_adapt_map.o 
/home/cameron/code/astro-stomp/stomp/.libs/libstomp.so
-L/home/cameron/code/astro-stomp/stomp 
/home/cameron/code/astro-stomp/stomp/.libs/libstomp.so: undefined reference
to
`Stomp::IndexedTreePixel::FindKNearestNeighbors(Stomp::AngularCoordinate&,
unsigned char, std::vector<Stomp::IndexedAngularCoordinate,
std::allocator<Stomp::IndexedAngularCoordinate> >&)'
collect2: ld returned 1 exit status
make[2]: *** [stomp_adapt_map] Error 1
make[2]: Leaving directory `/raid/data/cameron/code/astro-stomp/examples'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/raid/data/cameron/code/astro-stomp'
make: *** [all] Error 2

Calculate an auto-correlation with errors for a given data set and Map.  
Re-calculate the auto-
correlation and errors using the cross-correlation code.

This should produce identical values, but the errors for the cross-correlation 
are considerably 
larger.

Calculating pair-based correlation functions on small data sets (~1000s of 
objects) results in a 
correlation function that flattens out at medium angular scales (~0.1 degrees). 
 It's unclear exactly 
why this is happening, if it's due to a bug in the code or insufficient random 
points or what.

What steps will reproduce the problem?
1. Run FindPairs on an IndexedTreeMap() with a large separation angle

Expected output:
A list of indices

Error given by stomp:
" Illegal theta value on x index: Lambda,Eta = -22.4984,-32.7357,theta = 117.229
  x = 164/143"

The python interface generated via SWIG works well for passing objects by 
references, but fails to 
properly pass data primitives (doubles, ints, etc) properly.  For example, in 
python:

>>> import stomp
>>> ang = stomp.AngularCoordinate(60.0, 20.0)
>>> pix = stomp.Pixel(ang, 4)
>>> near_distance = 0.0
>>> far_distance = 0.0
>>> pix.EdgeDistances(ang, near_distance, far_distance)
True
>>> print near_distance, far_distance
0.0 0.0

This should store the distance from the input AngularCoordinate to the near and 
far Pixel edge 
distances into the near_distance and far_distance variables.  However, the 
values of these 
variables are unchanged.