Answers to frequently asked questions

1.  Java errors

1.1  Exception in thread "main" java.lang.OutOfMemoryError: Java heap space

This is caused by insufficient RAM. The default maximum heap size is 1000 Mb, but this can be increased as long as there is available RAM on your computer. Set the environment variable CAMINO_HEAP_SIZE to allow Java to use more memory, for example export CAMINO_HEAP_SIZE=2000.

Camino uses double precision floats internally, so it can require quite a bit of memory.

1.2  Camino won't compile or nothing will run

Unless something has gone really wrong, this should not happen. The nightly download is compiled and tested using Java 7 on Linux. If nothing works, then it's likely that you have an incompatible version of Java. Both javac and java need to be Sun / Oracle Java version 1.6 or later. It's not sufficient to have the correct Java somewhere on your computer, you must ensure that a call to javac or java on the command line runs the correct thing. A good way to test this is with java -version.

2.  Common Camino-specific exceptions and warnings

2.1  SEVERE: Exception in thread "main" class data.DataSourceException: No more voxels in data source.

Camino ran out of data prematurely. This usually happens when the data type is incorrectly specified, causing Camino to run out of input data because it's reading more bytes per data point than it should. This exception can also occur when the image dimensions are incorrectly specified.

2.2  SEVERE: Exception in thread "main" class data.DataSourceException: End of file reached without completing voxel.

This means the data source ran out of data in the middle of a voxel. This only happens for multi-component input, such as diffusion tensors or DWI data. It usually means the number of measurements in the scheme file doesn't match the number of measurements in the data file, or the wrong input model has been specified.

2.3  optimizers.MarquardtMinimiserException

An exception of this type is thrown if there is some problem with nonlinear optimization. The program will continue to run. A small percentage or optimizations may fail even on very high quality data, but you should not see thousands of these messages unless something is wrong.

Programs that fit nonlinear models usually output an exit code (as explained in the man pages) that will be non-zero in voxels where the exceptions occurred. Visualizing this image (for example, using voxel2image) will show the extent and distribution of optimization errors.

2.4  WARNING: WARNING: couldn't parse arg

This means Camino didn't understand one of the command line arguments. Any subsequent errors are likely a result of this problem.

Camino parses a bunch of common arguments (eg, -schemefile) automatically, so you won't see this error even if you add -schemefile a.scheme where it's not needed. You will see this error if you mistype a flag or use one of the more specialized ones with the wrong program.

3.  Other

3.1  The squid

The squid dates back to a CMIC retreat many Moons ago. We were...inspired. It represents the multi-tool nature of Camino and the contribution of cephalopods to our study of the human brain.