Shock elasticities in Dynare

As part of our project Examining Macroeconomic Models through the Lens of Asset Pricing (coauthored with Lars Peter Hansen, see my research page), we have developed tools aimed at facilitating the study of asset pricing implications of dynamic macroeconomic models solved using Dynare/Dynare++. These tools work as post-processors of Dynare solutions and exploit the exponential-quadratic framework that we are utilizing in our work.

Download and unpack the ZIP file and follow the installation instructions in the manual.

Image processing

I still get occasional emails regarding these old technical reports, so I leave them here.

• Circle detection using Hough transforms
  [4 March 2003] This homework assignment for the Image Processing and Computer Vision course at the University of Bristol has been cited a couple of times, so I provide it here as a reference.
  Report [PDF]
  The objective of this application is the recognition of circular shapes in an image. The edge-recognition procedure implements first- and second-derivative Sobel filters to determine edges and edge directions. A circular Hough transform is then applied to the thresholded edge map, followed by a convolution with a "Mexican hat" filter to enhance the "hot spots". After finding the centers of the hot spots, I accumulate in the radius space to find the most visible circles with the given center. Setting a suitable threshold allows for concentric circles. The method is evaluated using benchmark images to identify car wheels in traffic and coins strewn on a table.
  Source code [C file], header file [h file], complete archive [ZIP file]
  Images showing the detection process [ZIP file], also as TAR.GZ file
  The code uses the iplib library for reading and writing the PGM images. Other than for this purpose, the library is not necessary. A description of the iplib library together with the source code can be found at the webpage of Andrew Calway, University of Bristol. Unpack the library into the lib/ subdirectory.
• Stereo recognition: Determining depth in scene
  [5 May 2003]
  Report [PDF]
  The objective of this application is to determine the depth in a scene from two views obtained from a calibrated stereo system. This is done by reconstructing the 3-D position of detected edge points.
  Source code files in C++ [ZIP]
  Data files [ZIP]
  Corresponding edge points [tablecorr.txt] is a TXT file containing the image coordinates of matching edge points in the left and right image, format Xleft Yleft Xright Yright
  Reconstructed 3D-scene [table3d.txt] is a TXT file containing the coordinates of reconstructed edge points in the 3D scene, format X Y Z, viewable e.g. in Matlab or xgobi