rkwitt

1) Compile your code (using mex) with -g flag
2) Call gdb via MATLAB's 'matlab -Dgdb' functionality
3) Set a breakpoint for instance by 'break testfile.cpp:9'
3) Run MATLAB without the JVM by 'run -nojvm'
4) Now, run the segfaulting mex function 
5) Get a backtrace via 'bt' and use your debugging magic

Example:

$ cd '/mycode/

rkwitt

"""kw-techtip-04-03-2013-p3.py

Compute BoW representations and train / test a SVM classifier in a N-fold 
cross-validation setup. 

Usage
-----

$ python kw-techtip-04-03-2013-p3.py <DataFile> <IndexFile> <ClassInfoFile>
"""

rkwitt

"""kw-techtip-04-03-2013-p2.py

Fine-structure analysis of a graph.

Usage
-----
    
Assuming that the file "graph.txt" generated by kw-techtip-04-03-2013-p1.py
resides in "/tmp", run
    

rkwitt

"""kw-techtip-04-03-2013-p1.py
 
Simple example of how to construct a graph from a {0,1} adjacency matrix and a
file with vertex labels.
 
Usage
-----
 
$ python kw-techtip-04-03-2013-p1.py <AdjacencyFile> <LabelFile>
 

rkwitt

2014-09-29 09:29:04 +0200

make
all

clang -O   -c -o vecLibFort.o vecLibFort.c
clang -shared -O -DVECLIBFORT_INTERPOSE -o libvecLibFortI.dylib -O vecLibFort.c \
		-Wl,-reexport_framework -Wl,Accelerate \
		-install_name /usr/local/lib/libvecLibFortI.dylib
ar -cru libvecLibFort.a vecLibFort.o

rkwitt

2014-09-29 09:29:04 +0200

make
all

clang -O   -c -o vecLibFort.o vecLibFort.c
clang -shared -O -DVECLIBFORT_INTERPOSE -o libvecLibFortI.dylib -O vecLibFort.c \
		-Wl,-reexport_framework -Wl,Accelerate \
		-install_name /usr/local/lib/libvecLibFortI.dylib
ar -cru libvecLibFort.a vecLibFort.o

rkwitt

function [pos_ev_idx,neg_ev_idx,seed,data] = test_negative_type_simple(options)
% TEST_NEGATIVE_TYPE_SIMPLE evaluate the eigenvalues of Gram matrices 
%   
%   This function evaluates the number of positive/negative eigenvalues 
%   of the Gram matrix M, either constructed from the (1) q-Wasserstein 
%   distance or (2) the bottleneck distance.
%
%   Both distances lead to nonnegative (symmetric) Gram matrices and
%   the question is if the Gram matrices are (conditionally) negative
%   definite (see suppl. material). 

rkwitt

int f(int a, int b, int c)
{
  int c;
  c = a + b;
  return c;
}

int main()
{
  int a;

rkwitt

int a = 1111;

int main() {

  a = 3333;

}

rkwitt

int g(int x, int y) {
  if (y > 0)
    return g(x, y - 1) + 1  ;
  else
    return x;
}

int main() {
  int a;