strip whitespaces

27 files changed, 259 insertions, 259 deletions

diff --git a/include/gensvm.h b/include/gensvm.h
index 5101b41..ef85157 100644
--- a/include/gensvm.h
+++ b/include/gensvm.h
@@ -31,7 +31,7 @@ struct GenModel {
 	double epsilon;
 	///< stopping criterion for the IM algorithm.
 	double p;
-	///< parameter for the L-p norm in the loss function	
+	///< parameter for the L-p norm in the loss function
 	double kappa;
 	///< parameter for the Huber hinge function
 	double lambda;
@@ -51,16 +51,16 @@ struct GenModel {
 	///< 3D simplex difference matrix
 	double *Q;
 	///< error matrix
-	double *H; 
+	double *H;
 	///< Huber weighted error matrix
 	double *R;
-	///< 0-1 auixiliary matrix, this matrix is n x K, with for row i a 0 on 
+	///< 0-1 auixiliary matrix, this matrix is n x K, with for row i a 0 on
 	///< column y[i]-1, and 1 everywhere else.
-	double *rho; 	
+	double *rho;
 	///< vector of instance weights
 	double training_error;
 	///< loss function value after training has finished
-	char *data_file; 
+	char *data_file;
 	///< filename of the data
 	KernelType kerneltype;
 	///< type of kernel used in the model
@@ -94,13 +94,13 @@ struct GenData {
  	long *y;
 	///< array of class labels, 1..K
 	double *Z;
-	///< augmented data matrix (either equal to RAW or to the eigenvectors 
+	///< augmented data matrix (either equal to RAW or to the eigenvectors
 	///< of the kernel matrix)
 	double *RAW;
 	///< augmented raw data matrix
 	double *Sigma;
-	KernelType kerneltype; 	
- 	double *kernelparam; 
+	KernelType kerneltype;
+ 	double *kernelparam;
 };
 
 #endif
diff --git a/include/gensvm_io.h b/include/gensvm_io.h
index 35b6a5a..73c24bd 100644
--- a/include/gensvm_io.h
+++ b/include/gensvm_io.h
@@ -24,7 +24,7 @@ void gensvm_read_data(struct GenData *dataset, char *data_file);
 void gensvm_read_model(struct GenModel *model, char *model_filename);
 void gensvm_write_model(struct GenModel *model, char *output_filename);
 
-void gensvm_write_predictions(struct GenData *data, long *predy, 
+void gensvm_write_predictions(struct GenData *data, long *predy,
 		char *output_filename);
 
 #endif
diff --git a/include/gensvm_kernel.h b/include/gensvm_kernel.h
index d5c5e8d..d01eb88 100644
--- a/include/gensvm_kernel.h
+++ b/include/gensvm_kernel.h
@@ -6,7 +6,7 @@
  *
  * @details
  * Contains function declarations for computing the kernel matrix
- * in nonlinear MSVMGen. Additional kernel functions should be 
+ * in nonlinear MSVMGen. Additional kernel functions should be
  * included here and in gensvm_kernel.c
  *
  */
diff --git a/include/gensvm_lapack.h b/include/gensvm_lapack.h
index 7ac4fc9..a664cba 100644
--- a/include/gensvm_lapack.h
+++ b/include/gensvm_lapack.h
@@ -3,9 +3,9 @@
  * @author Gertjan van den Burg
  * @date August, 2013
  * @brief Header file for gensvm_lapack.c
- * 
+ *
  * @details
- * Function declarations for external LAPACK functions 
+ * Function declarations for external LAPACK functions
  *
  */
 
diff --git a/include/gensvm_train_dataset.h b/include/gensvm_train_dataset.h
index 16f8e07..c743bd7 100644
--- a/include/gensvm_train_dataset.h
+++ b/include/gensvm_train_dataset.h
@@ -74,8 +74,8 @@ struct Queue {
  *
  * @param traintype 		type of training to use
  * @param kerneltype 		type of kernel to use throughout training
- * @param repeats 		number of repeats to be done after the grid 
- * 				search to find the parameter set with the 
+ * @param repeats 		number of repeats to be done after the grid
+ * 				search to find the parameter set with the
  * 				most consistent high performance
  * @param folds 		number of folds in cross validation
  * @param Np 			size of the array of p values
@@ -87,7 +87,7 @@ struct Queue {
  * @param Nc 			size of the array of coef values
  * @param Nd 			size of the array of degree values
  * @param *weight_idxs 		array of weight_idxs
- * @param *ps 			array of p values 
+ * @param *ps 			array of p values
  * @param *lambdas 		array of lambda values
  * @param *kappas 		array of kappa values
  * @param *epsilons 		array of epsilon values
diff --git a/include/gensvm_util.h b/include/gensvm_util.h
index fe8d2a3..18aa553 100644
--- a/include/gensvm_util.h
+++ b/include/gensvm_util.h
@@ -3,7 +3,7 @@
  * @author Gertjan van den Burg
  * @date August, 2013
  * @brief Header file for util.c
- * 
+ *
  * @details
  * Function declarations for utility functions of the program.
  *
diff --git a/include/globals.h b/include/globals.h
index 46cc3d2..dfd65ad 100644
--- a/include/globals.h
+++ b/include/globals.h
@@ -8,10 +8,10 @@
  * This header file contains defines and includes which are used in many
  * parts of the program. Most notable are the Calloc, Malloc and Memset
  * defines, which are commonly used to allocate memory. These functions
- * are shorthands for their lowercase counterparts. 
+ * are shorthands for their lowercase counterparts.
  *
- * Furthermore, a maximum and minimum function are defined here. These 
- * functions have their own include guards, to ensure potential linked 
+ * Furthermore, a maximum and minimum function are defined here. These
+ * functions have their own include guards, to ensure potential linked
  * libraries don't conflict with these definitions.
  *
  */
diff --git a/include/libGenSVM.h b/include/libGenSVM.h
index cfa2815..dbf0903 100644
--- a/include/libGenSVM.h
+++ b/include/libGenSVM.h
@@ -11,16 +11,16 @@
  */
 
 /**
- * @todo 
+ * @todo
  * rename this file and libGenSVM.c to correspond with the lowercase convention.
  * Also change the name of the include guard.
- */ 
+ */
 #ifndef LIBGENSVM_H
 #define LIBGENSVM_H
 
 #include "globals.h"
 
-// forward declarations 
+// forward declarations
 struct GenData;
 struct GenModel;
 
diff --git a/include/types.h b/include/types.h
index 1cbcba0..329401a 100644
--- a/include/types.h
+++ b/include/types.h
@@ -4,7 +4,7 @@
  * @date August, 2013
  * @brief Definitions of common types
  *
- * @details 
+ * @details
  * Here common types used throughout the program are defined.
  *
  */
diff --git a/src/GenSVMgrid.c b/src/GenSVMgrid.c
index 0b6d33e..3cf1346 100644
--- a/src/GenSVMgrid.c
+++ b/src/GenSVMgrid.c
@@ -8,7 +8,7 @@
  * This is a command line interface to the parameter grid search functionality
  * of the algorithm. The grid search is specified in a separate file, thereby
  * reducing the number of command line arguments. See
- * read_training_from_file() for documentation on the training file. 
+ * read_training_from_file() for documentation on the training file.
  *
  * The program runs a grid search as specified in the training file. If
  * desired the grid search can incorporate consistency checks to find the
@@ -82,7 +82,7 @@ int main(int argc, char **argv)
 	struct GenData *test_data = gensvm_init_data();
 	struct Queue *q = gensvm_init_queue();
 
-	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help") 
+	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help")
 			|| gensvm_check_argv_eq(argc, argv, "-h") )
 		exit_with_help();
 	parse_command_line(argc, argv, input_filename);
@@ -150,7 +150,7 @@ void parse_command_line(int argc, char **argv, char *input_filename)
 				i--;
 				break;
 			default:
-				fprintf(stderr, "Unknown option: -%c\n", 
+				fprintf(stderr, "Unknown option: -%c\n",
 						argv[i-1][1]);
 				exit_with_help();
 		}
@@ -184,10 +184,10 @@ KernelType parse_kernel_str(char *kernel_line)
  *
  * @details
  * Read the Training struct from a file. The training file follows a specific
- * format specified in @ref spec_training_file. 
+ * format specified in @ref spec_training_file.
  *
  * Commonly used string functions in this function are all_doubles_str() and
- * all_longs_str(). 
+ * all_longs_str().
  *
  * @param[in] 	input_filename 	filename of the training file
  * @param[in] 	training 	Training structure to place the parsed
@@ -206,7 +206,7 @@ void read_training_from_file(char *input_filename, struct Training *training)
 
 	fid = fopen(input_filename, "r");
 	if (fid == NULL) {
-		fprintf(stderr, "Error opening training file %s\n", 
+		fprintf(stderr, "Error opening training file %s\n",
 				input_filename);
 		exit(1);
 	}
@@ -216,7 +216,7 @@ void read_training_from_file(char *input_filename, struct Training *training)
 		Memset(lparams, long, MAX_LINE_LENGTH);
 		if (str_startswith(buffer, "train:")) {
 			sscanf(buffer, "train: %s\n", train_filename);
-			training->train_data_file = Calloc(char, 
+			training->train_data_file = Calloc(char,
 					MAX_LINE_LENGTH);
 			strcpy(training->train_data_file, train_filename);
 		} else if (str_startswith(buffer, "test:")) {
diff --git a/src/GenSVMpred.c b/src/GenSVMpred.c
index ef2c97c..57680b1 100644
--- a/src/GenSVMpred.c
+++ b/src/GenSVMpred.c
@@ -36,8 +36,8 @@ extern FILE *GENSVM_OUTPUT_FILE;
 
 // function declarations
 void exit_with_help();
-void parse_command_line(int argc, char **argv, 
-		char *input_filename, char *output_filename, 
+void parse_command_line(int argc, char **argv,
+		char *input_filename, char *output_filename,
 		char *model_filename);
 
 /**
@@ -80,7 +80,7 @@ int main(int argc, char **argv)
 	char model_filename[MAX_LINE_LENGTH];
 	char output_filename[MAX_LINE_LENGTH];;
 
-	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help") 
+	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help")
 			|| gensvm_check_argv_eq(argc, argv, "-h") )
 		exit_with_help();
 	parse_command_line(argc, argv, input_filename, output_filename,
@@ -137,7 +137,7 @@ int main(int argc, char **argv)
  *
  * @param[in] 	argc 			number of command line arguments
  * @param[in] 	argv 			array of command line arguments
- * @param[in] 	input_filename 		pre-allocated array for the input 
+ * @param[in] 	input_filename 		pre-allocated array for the input
  * 					filename
  * @param[in] 	output_filename 	pre-allocated array for the output
  * 					filename
@@ -165,7 +165,7 @@ void parse_command_line(int argc, char **argv, char *input_filename,
 				i--;
 				break;
 			default:
-				fprintf(stderr, "Unknown option: -%c\n", 
+				fprintf(stderr, "Unknown option: -%c\n",
 						argv[i-1][1]);
 				exit_with_help();
 		}
diff --git a/src/GenSVMtrain.c b/src/GenSVMtrain.c
index 63e2512..b9dc250 100644
--- a/src/GenSVMtrain.c
+++ b/src/GenSVMtrain.c
@@ -6,8 +6,8 @@
  *
  * @details
  * This is a command line program for training a single model on a given
- * dataset. To run a grid search over a number of parameter configurations, 
- * see trainGenSVMdataset.c. 
+ * dataset. To run a grid search over a number of parameter configurations,
+ * see trainGenSVMdataset.c.
  *
  */
 
@@ -28,7 +28,7 @@ extern FILE *GENSVM_OUTPUT_FILE;
 
 // function declarations
 void exit_with_help();
-void parse_command_line(int argc, char **argv, struct GenModel *model, 
+void parse_command_line(int argc, char **argv, struct GenModel *model,
 		char *input_filename, char *output_filename, char *model_filename);
 
 /**
@@ -81,24 +81,24 @@ int main(int argc, char **argv)
 	struct GenModel *model = gensvm_init_model();
 	struct GenData *data = gensvm_init_data();
 
-	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help") 
-			|| gensvm_check_argv_eq(argc, argv, "-h") ) 
+	if (argc < MINARGS || gensvm_check_argv(argc, argv, "-help")
+			|| gensvm_check_argv_eq(argc, argv, "-h") )
 		exit_with_help();
-	parse_command_line(argc, argv, model, input_filename, 
+	parse_command_line(argc, argv, model, input_filename,
 			output_filename, model_filename);
 
 	// read data file
 	gensvm_read_data(data, input_filename);
 
-	// copy dataset parameters to model	
+	// copy dataset parameters to model
 	model->n = data->n;
 	model->m = data->m;
 	model->K = data->K;
 	model->data_file = input_filename;
-	
+
 	// allocate model
 	gensvm_allocate_model(model);
-	
+
 	// initialize kernel (if necessary)
 	//gensvm_make_kernel(model, data);
 
@@ -131,7 +131,7 @@ int main(int argc, char **argv)
 	// free model and data
 	gensvm_free_model(model);
 	gensvm_free_data(data);
-	
+
 	return 0;
 }
 
@@ -155,12 +155,12 @@ int main(int argc, char **argv)
  * 					filename
  *
  */
-void parse_command_line(int argc, char **argv, struct GenModel *model, 
+void parse_command_line(int argc, char **argv, struct GenModel *model,
 		char *input_filename, char *output_filename, char *model_filename)
 {
 	int i;
-	double gamma = 1.0, 
-	       degree = 2.0, 
+	double gamma = 1.0,
+	       degree = 2.0,
 	       coef = 0.0;
 
 	GENSVM_OUTPUT_FILE = stdout;
@@ -210,12 +210,12 @@ void parse_command_line(int argc, char **argv, struct GenModel *model,
 				i--;
 				break;
 			default:
-				fprintf(stderr, "Unknown option: -%c\n", 
+				fprintf(stderr, "Unknown option: -%c\n",
 						argv[i-1][1]);
 				exit_with_help();
 		}
 	}
-	
+
 	// read input filename
 	if (i >= argc)
 		exit_with_help();
diff --git a/src/GenSVMtraintest.c b/src/GenSVMtraintest.c
index c5f09f8..955f4fa 100644
--- a/src/GenSVMtraintest.c
+++ b/src/GenSVMtraintest.c
@@ -5,7 +5,7 @@
  * @brief Command line interface for training and testing with a GenSVM model
  *
  * @details
- * This is a command line program for training and testing on a single model 
+ * This is a command line program for training and testing on a single model
  * with specified model parameters.
  *
  */
@@ -126,13 +126,13 @@ int main(int argc, char **argv)
 	// start training
 	gensvm_optimize(model, traindata);
 
-	// if we also have a test set, predict labels and write to predictions 
+	// if we also have a test set, predict labels and write to predictions
 	// to an output file if specified
 	if (with_test) {
 		// predict labels
 		predy = Calloc(long, testdata->n);
 		gensvm_predict_labels(testdata, model, predy);
-	
+
 		if (testdata->y != NULL) {
 			performance = gensvm_prediction_perf(testdata, predy);
 			note("Predictive performance: %3.2f%%\n", performance);
@@ -181,7 +181,7 @@ void parse_command_line(int argc, char **argv, struct GenModel *model,
 	double gamma = 1.0,
 	       degree = 2.0,
 	       coef = 0.0;
-	
+
 	GENSVM_OUTPUT_FILE = stdout;
 
 	// parse options
diff --git a/src/gensvm_crossval.c b/src/gensvm_crossval.c
index 864e692..6930166 100644
--- a/src/gensvm_crossval.c
+++ b/src/gensvm_crossval.c
@@ -1,13 +1,13 @@
 /**
  * @file crossval.c
- * @author Gertjan van den Burg 
+ * @author Gertjan van den Burg
  * @date January 7, 2014
  * @brief Functions for cross validation
  *
  * @details
  * This file contains functions for performing cross validation. The funtion
  * gensvm_make_cv_split() creates a cross validation vector for non-stratified
- * cross validation. The function gensvm_get_tt_split() creates a train and 
+ * cross validation. The function gensvm_get_tt_split() creates a train and
  * test dataset from a given dataset and a pre-determined CV partition vector.
  * See individual function documentation for details.
  *
@@ -22,17 +22,17 @@
  *
  * @details
  * A pre-allocated vector of length N is created which can be used to define
- * cross validation splits. The folds are contain between 
- * @f$ \lfloor N / folds \rfloor @f$ and @f$ \lceil N / folds \rceil @f$ 
- * instances. An instance is mapped to a partition randomly until all folds 
- * contain @f$ N \% folds @f$ instances. The zero fold then contains 
+ * cross validation splits. The folds are contain between
+ * @f$ \lfloor N / folds \rfloor @f$ and @f$ \lceil N / folds \rceil @f$
+ * instances. An instance is mapped to a partition randomly until all folds
+ * contain @f$ N \% folds @f$ instances. The zero fold then contains
  * @f$ N / folds + N \% folds @f$ instances. These remaining @f$ N \% folds @f$
- * instances are then distributed over the first @f$ N \% folds @f$ folds. 
+ * instances are then distributed over the first @f$ N \% folds @f$ folds.
  *
  * @param[in] 		N 	number of instances
  * @param[in] 		folds 	number of folds
  * @param[in,out] 	cv_idx 	array of size N which contains the fold index
- * 				for each observation on exit	
+ * 				for each observation on exit
  *
  */
 void gensvm_make_cv_split(long N, long folds, long *cv_idx)
@@ -44,7 +44,7 @@ void gensvm_make_cv_split(long N, long folds, long *cv_idx)
 
 	long big_folds = N%folds;
 	long small_fold_size = N/folds;
-	
+
 	j = 0;
 	for (i=0; i<small_fold_size*folds; i++)
 		while (1) {
@@ -73,9 +73,9 @@ void gensvm_make_cv_split(long N, long folds, long *cv_idx)
  * @details
  * Given a GenData structure for the full dataset, a previously created
  * cross validation split vector and a fold index, a training and test dataset
- * are created. 
+ * are created.
  *
- * @param[in] 		full_data 	a GenData structure for the entire 
+ * @param[in] 		full_data 	a GenData structure for the entire
  * 					dataset
  * @param[in,out] 	train_data 	an initialized GenData structure which
  * 					on exit contains the training dataset
@@ -83,7 +83,7 @@ void gensvm_make_cv_split(long N, long folds, long *cv_idx)
  * 					on exit contains the test dataset
  * @param[in] 		cv_idx 		a vector of cv partitions created by
  * 					gensvm_make_cv_split()
- * @param[in] 		fold_idx 	index of the fold which becomes the 
+ * @param[in] 		fold_idx 	index of the fold which becomes the
  * 					test dataset
  */
 void gensvm_get_tt_split(struct GenData *full_data, struct GenData *train_data,
diff --git a/src/gensvm_init.c b/src/gensvm_init.c
index 6be2706..228e9fe 100644
--- a/src/gensvm_init.c
+++ b/src/gensvm_init.c
@@ -4,9 +4,9 @@
  * @date January 7, 2014
  * @brief Functions for initializing model and data structures
  *
- * @details 
+ * @details
  * This file contains functions for initializing a GenModel instance
- * and a GenData instance. In addition, default values for these 
+ * and a GenData instance. In addition, default values for these
  * structures are defined here (and only here). Functions for allocating
  * memory for the model structure and freeing of the model and data structures
  * are also included.
@@ -59,7 +59,7 @@ struct GenModel *gensvm_init_model()
  * @brief Initialize a GenData structure
  *
  * @details
- * A GenData structure is initialized and default values are set. 
+ * A GenData structure is initialized and default values are set.
  * A pointer to the initialized data is returned.
  *
  * @returns 	initialized GenData
@@ -161,8 +161,8 @@ void gensvm_allocate_model(struct GenModel *model)
  * @brief Reallocate memory for GenModel
  *
  * @details
- * This function can be used to reallocate existing memory for a GenModel, 
- * upon a change in the model dimensions. This is used in combination with 
+ * This function can be used to reallocate existing memory for a GenModel,
+ * upon a change in the model dimensions. This is used in combination with
  * kernels.
  *
  * @param[in] 	model 	GenModel to reallocate
diff --git a/src/gensvm_io.c b/src/gensvm_io.c
index 01f1db5..d81f19b 100644
--- a/src/gensvm_io.c
+++ b/src/gensvm_io.c
@@ -6,7 +6,7 @@
  *
  * @details
  * This file contains functions for reading and writing model files, and data
- * files. 
+ * files.
  *
  */
 
@@ -18,12 +18,12 @@
 
 /**
  * @brief Read data from file
- * 
+ *
  * @details
  * Read the data from the data_file. The data matrix X is augmented
  * with a column of ones, to get the matrix Z. The data is expected
  * to follow a specific format, which is specified in the @ref spec_data_file.
- * The class labels are corrected internally to correspond to the interval 
+ * The class labels are corrected internally to correspond to the interval
  * [1 .. K], where K is the total number of classes.
  *
  * @todo
@@ -105,11 +105,11 @@ void gensvm_read_data(struct GenData *dataset, char *data_file)
 	}
 
 	if (nr < n * m) {
-		fprintf(stderr, "ERROR: not enough data found in %s\n", 
+		fprintf(stderr, "ERROR: not enough data found in %s\n",
 				data_file);
 		exit(0);
 	}
-	
+
 	// Set the column of ones
 	for (i=0; i<n; i++)
 		matrix_set(dataset->RAW, m+1, i, 0, 1.0);
@@ -145,7 +145,7 @@ void gensvm_read_model(struct GenModel *model, char *model_filename)
 
 	fid = fopen(model_filename, "r");
 	if (fid == NULL) {
-		fprintf(stderr, "Error opening model file %s\n", 
+		fprintf(stderr, "Error opening model file %s\n",
 				model_filename);
 		exit(1);
 	}
@@ -158,7 +158,7 @@ void gensvm_read_model(struct GenModel *model, char *model_filename)
 	model->lambda = get_fmt_double(fid, model_filename, "lambda = %lf");
 	model->kappa = get_fmt_double(fid, model_filename, "kappa = %lf");
 	model->epsilon = get_fmt_double(fid, model_filename, "epsilon = %lf");
-	model->weight_idx = (int) get_fmt_long(fid, model_filename, 
+	model->weight_idx = (int) get_fmt_long(fid, model_filename,
 			"weight_idx = %li");
 
 	// skip to data section
@@ -167,7 +167,7 @@ void gensvm_read_model(struct GenModel *model, char *model_filename)
 
 	// read filename of data file
 	if (fgets(buffer, MAX_LINE_LENGTH, fid) == NULL) {
-		fprintf(stderr, "Error reading model file %s\n", 
+		fprintf(stderr, "Error reading model file %s\n",
 				model_filename);
 		exit(1);
 	}
@@ -193,7 +193,7 @@ void gensvm_read_model(struct GenModel *model, char *model_filename)
 	}
 	if (nr != (model->m+1)*(model->K-1)) {
 		fprintf(stderr, "Error reading model file %s. "
-				"Not enough elements of V found.\n", 
+				"Not enough elements of V found.\n",
 				model_filename);
 		exit(1);
 	}
@@ -207,7 +207,7 @@ void gensvm_read_model(struct GenModel *model, char *model_filename)
  * UTC + offset. The model file further corresponds to the @ref
  * spec_model_file.
  *
- * @param[in] 	model 		GenModel which contains an estimate for 
+ * @param[in] 	model 		GenModel which contains an estimate for
  * 				GenModel::V
  * @param[in] 	output_filename the output file to write the model to
  *
@@ -221,7 +221,7 @@ void gensvm_write_model(struct GenModel *model, char *output_filename)
 	// open output file
 	fid = fopen(output_filename, "w");
 	if (fid == NULL) {
-		fprintf(stderr, "Error opening output file %s", 
+		fprintf(stderr, "Error opening output file %s",
 				output_filename);
 		exit(1);
 	}
@@ -246,8 +246,8 @@ void gensvm_write_model(struct GenModel *model, char *output_filename)
 	fprintf(fid, "Output:\n");
 	for (i=0; i<model->m+1; i++) {
 		for (j=0; j<model->K-1; j++) {
-			fprintf(fid, "%+15.16f ", 
-					matrix_get(model->V, 
+			fprintf(fid, "%+15.16f ",
+					matrix_get(model->V,
 						model->K-1, i, j));
 		}
 		fprintf(fid, "\n");
@@ -261,17 +261,17 @@ void gensvm_write_model(struct GenModel *model, char *output_filename)
  *
  * @details
  * Write the given predictions to an output file, such that the resulting file
- * corresponds to the @ref spec_data_file. 
+ * corresponds to the @ref spec_data_file.
  *
  * @param[in] 	data 		GenData with the original instances
- * @param[in] 	predy 		predictions of the class labels of the 
+ * @param[in] 	predy 		predictions of the class labels of the
  * 				instances in the given GenData. Note that the
  * 				order of the instances is assumed to be the
  * 				same.
  * @param[in] 	output_filename the file to which the predictions are written
  *
  */
-void gensvm_write_predictions(struct GenData *data, long *predy, 
+void gensvm_write_predictions(struct GenData *data, long *predy,
 		char *output_filename)
 {
 	long i, j;
@@ -279,18 +279,18 @@ void gensvm_write_predictions(struct GenData *data, long *predy,
 
 	fid = fopen(output_filename, "w");
 	if (fid == NULL) {
-		fprintf(stderr, "Error opening output file %s", 
+		fprintf(stderr, "Error opening output file %s",
 				output_filename);
 		exit(1);
 	}
-	
+
 	fprintf(fid, "%li\n", data->n);
 	fprintf(fid, "%li\n", data->m);
 
 	for (i=0; i<data->n; i++) {
-		for (j=0; j<data->m; j++) 
-			fprintf(fid, "%f ", 
-					matrix_get(data->Z, 
+		for (j=0; j<data->m; j++)
+			fprintf(fid, "%f ",
+					matrix_get(data->Z,
 						data->m+1, i, j+1));
 		fprintf(fid, "%li\n", predy[i]);
 	}
diff --git a/src/gensvm_kernel.c b/src/gensvm_kernel.c
index a6bc9fc..f53bcce 100644
--- a/src/gensvm_kernel.c
+++ b/src/gensvm_kernel.c
@@ -5,7 +5,7 @@
  * @brief Defines main functions for use of kernels in GenSVM.
  *
  * @details
- * Functions for constructing different kernels using user-supplied 
+ * Functions for constructing different kernels using user-supplied
  * parameters. Also contains the functions for decomposing the
  * kernel matrix using several decomposition methods.
  *
@@ -34,7 +34,7 @@ void gensvm_kernel_preprocess(struct GenModel *model, struct GenData *data)
 		return;
 	}
 
-	int i;	
+	int i;
 	long r,
 	     n = data->n;
 	double *P = NULL,
@@ -55,7 +55,7 @@ void gensvm_kernel_preprocess(struct GenModel *model, struct GenData *data)
 
 	// build M and set to data (leave RAW intact)
 	gensvm_make_trainfactor(data, P, Sigma, r);
-	
+
 	// Set Sigma to data->Sigma (need it again for prediction)
 	if (data->Sigma != NULL)
 		free(data->Sigma);
@@ -81,7 +81,7 @@ void gensvm_kernel_preprocess(struct GenModel *model, struct GenData *data)
 			data->kernelparam[0] = model->kernelparam[0];
 			data->kernelparam[1] = model->kernelparam[1];
 	}
-	
+
 	free(K);
 	free(P);
 }
@@ -117,13 +117,13 @@ void gensvm_make_kernel(struct GenModel *model, struct GenData *data,
 			x1 = &data->RAW[i*(data->m+1)+1];
 			x2 = &data->RAW[j*(data->m+1)+1];
 			if (model->kerneltype == K_POLY)
-				value = gensvm_dot_poly(x1, x2, 
+				value = gensvm_dot_poly(x1, x2,
 						model->kernelparam, data->m);
 			else if (model->kerneltype == K_RBF)
-				value = gensvm_dot_rbf(x1, x2, 
+				value = gensvm_dot_rbf(x1, x2,
 						model->kernelparam, data->m);
 			else if (model->kerneltype == K_SIGMOID)
-				value = gensvm_dot_sigmoid(x1, x2, 
+				value = gensvm_dot_sigmoid(x1, x2,
 						model->kernelparam, data->m);
 			else {
 				fprintf(stderr, "Unknown kernel type in "
@@ -154,11 +154,11 @@ long gensvm_make_eigen(double *K, long n, double **P, double **Sigma)
 
 	IWORK = Malloc(int, 5*n);
 	IFAIL = Malloc(int, n);
-	
-	// highest precision eigenvalues, may reduce for speed	
+
+	// highest precision eigenvalues, may reduce for speed
 	abstol = 2.0*dlamch('S');
 
-	// first perform a workspace query to determine optimal size of the 
+	// first perform a workspace query to determine optimal size of the
 	// WORK array.
 	WORK = Malloc(double, 1);
 	status = dsyevx(
@@ -183,7 +183,7 @@ long gensvm_make_eigen(double *K, long n, double **P, double **Sigma)
 			IFAIL);
 	LWORK = WORK[0];
 
-	// allocate the requested memory for the eigendecomposition 
+	// allocate the requested memory for the eigendecomposition
 	WORK = (double *)realloc(WORK, LWORK*sizeof(double));
 	status = dsyevx(
 			'V',
@@ -211,7 +211,7 @@ long gensvm_make_eigen(double *K, long n, double **P, double **Sigma)
 		exit(1);
 	}
 
-	// Select the desired number of eigenvalues, depending on their size.  
+	// Select the desired number of eigenvalues, depending on their size.
 	// dsyevx sorts eigenvalues in ascending order.
 	max_eigen = tempSigma[n-1];
 	cutoff_idx = 0;
@@ -223,23 +223,23 @@ long gensvm_make_eigen(double *K, long n, double **P, double **Sigma)
 		}
 
 	num_eigen = n - cutoff_idx;
-	
+
 	*Sigma = Calloc(double, num_eigen);
-	
+
 	for (i=0; i<num_eigen; i++) {
 		(*Sigma)[i] = tempSigma[n-1 - i];
 	}
 
-	// revert P to row-major order and copy only the the columns 
+	// revert P to row-major order and copy only the the columns
 	// corresponding to the selected eigenvalues
-	*P = Calloc(double, n*num_eigen);	
+	*P = Calloc(double, n*num_eigen);
 	for (j=n-1; j>n-1-num_eigen; j--) {
 		for (i=0; i<n; i++) {
 			(*P)[i*num_eigen + (n-1)-j] = tempP[i + j*n];
 		}
 	}
 
-	free(tempSigma);	
+	free(tempSigma);
 	free(tempP);
 	free(IWORK);
 	free(IFAIL);
@@ -322,7 +322,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
 	double value,
 	       *N = NULL,
 	       *M = NULL;
-	
+
 	n1 = traindata->n;
 	n2 = testdata->n;
 	r = traindata->r;
@@ -340,7 +340,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
 		exit(1);
 	}
 
-	// copy M from traindata->Z because we need it in dgemm without column 
+	// copy M from traindata->Z because we need it in dgemm without column
 	// of 1's.
 	for (i=0; i<n1; i++)
 		for (j=0; j<r; j++)
@@ -370,7 +370,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
 		for (i=0; i<n2; i++)
 			matrix_mul(N, r, i, j, value);
 	}
-	
+
 	// write N to Z with a column of ones
 	testdata->Z = Calloc(double, n2*(r+1));
 	if (testdata->Z == NULL) {
@@ -380,7 +380,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
 	}
 	for (i=0; i<n2; i++) {
 		for (j=0; j<r; j++) {
-			matrix_set(testdata->Z, r+1, i, j+1, 
+			matrix_set(testdata->Z, r+1, i, j+1,
 					matrix_get(N, r, i, j));
 		}
 		matrix_set(testdata->Z, r+1, i, 0, 1.0);
@@ -394,7 +394,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
 
 /**
  * @brief Compute the RBF kernel between two vectors
- * 
+ *
  * @details
  * The RBF kernel is computed between two vectors. This kernel is defined as
  * @f[
@@ -404,7 +404,7 @@ void gensvm_make_testfactor(struct GenData *testdata,
  *
  * @param[in] 	x1 		first vector
  * @param[in] 	x2 		second vector
- * @param[in] 	kernelparam 	array of kernel parameters (gamma is first 
+ * @param[in] 	kernelparam 	array of kernel parameters (gamma is first
  * 				element)
  * @param[in] 	n 		length of the vectors x1 and x2
  * @returns  			kernel evaluation
@@ -413,8 +413,8 @@ double gensvm_dot_rbf(double *x1, double *x2, double *kernelparam, long n)
 {
 	long i;
 	double value = 0.0;
-	
-	for (i=0; i<n; i++) 
+
+	for (i=0; i<n; i++)
 		value += (x1[i] - x2[i]) * (x1[i] - x2[i]);
 	value *= -kernelparam[0];
 	return exp(value);
@@ -424,7 +424,7 @@ double gensvm_dot_rbf(double *x1, double *x2, double *kernelparam, long n)
  * @brief Compute the polynomial kernel between two vectors
  *
  * @details
- * The polynomial kernel is computed between two vectors. This kernel is 
+ * The polynomial kernel is computed between two vectors. This kernel is
  * defined as
  * @f[
  * 	k(x_1, x_2) = ( \gamma \langle x_1, x_2 \rangle + c)^d
@@ -450,7 +450,7 @@ double gensvm_dot_poly(double *x1, double *x2, double *kernelparam, long n)
 
 /**
  * @brief Compute the sigmoid kernel between two vectors
- * 
+ *
  * @details
  * The sigmoid kernel is computed between two vectors. This kernel is defined
  * as
diff --git a/src/gensvm_lapack.c b/src/gensvm_lapack.c
index 6f50340..56dfc20 100644
--- a/src/gensvm_lapack.c
+++ b/src/gensvm_lapack.c
@@ -15,13 +15,13 @@
  * @brief Solve AX = B where A is symmetric positive definite.
  *
  * @details
- * Solve a linear system of equations AX = B where A is symmetric positive 
+ * Solve a linear system of equations AX = B where A is symmetric positive
  * definite. This function uses the externel LAPACK routine dposv.
  *
  * @param[in] 		UPLO 	which triangle of A is stored
  * @param[in] 		N 	order of A
  * @param[in] 		NRHS 	number of columns of B
- * @param[in,out] 	A 	double precision array of size (LDA, N). On 
+ * @param[in,out] 	A 	double precision array of size (LDA, N). On
  * 				exit contains the upper or lower factor of the
  * 				Cholesky factorization of A.
  * @param[in] 		LDA 	leading dimension of A
@@ -36,7 +36,7 @@
  * 					- >0: 	if i, the leading minor of A
  * 						was not positive definite
  *
- * See the LAPACK documentation at: 
+ * See the LAPACK documentation at:
  * http://www.netlib.org/lapack/explore-html/dc/de9/group__double_p_osolve.html
  */
 int dposv(char UPLO, int N, int NRHS, double *A, int LDA, double *B,
@@ -53,13 +53,13 @@ int dposv(char UPLO, int N, int NRHS, double *A, int LDA, double *B,
  * @brief Solve a system of equations AX = B where A is symmetric.
  *
  * @details
- * Solve a linear system of equations AX = B where A is symmetric. This 
+ * Solve a linear system of equations AX = B where A is symmetric. This
  * function uses the external LAPACK routine dsysv.
  *
  * @param[in] 		UPLO 	which triangle of A is stored
  * @param[in] 		N 	order of A
  * @param[in] 		NRHS 	number of columns of B
- * @param[in,out] 	A 	double precision array of size (LDA, N). On 
+ * @param[in,out] 	A 	double precision array of size (LDA, N). On
  * 				exit contains the block diagonal matrix D and
  * 				the multipliers used to obtain the factor U or
  * 				L from the factorization A = U*D*U**T or
@@ -96,7 +96,7 @@ int dsysv(char UPLO, int N, int NRHS, double *A, int LDA, int *IPIV,
 }
 
 /**
- * @brief Compute the eigenvalues and optionally the eigenvectors of a 
+ * @brief Compute the eigenvalues and optionally the eigenvectors of a
  * symmetric matrix.
  *
  * @details
diff --git a/src/gensvm_matrix.c b/src/gensvm_matrix.c
index 66e3947..c2e5986 100644
--- a/src/gensvm_matrix.c
+++ b/src/gensvm_matrix.c
@@ -17,7 +17,7 @@
 
 /**
  * @brief print a matrix
- * 
+ *
  * @details
  * Debug function to print a matrix
  *
diff --git a/src/gensvm_pred.c b/src/gensvm_pred.c
index 7baae07..7835bf6 100644
--- a/src/gensvm_pred.c
+++ b/src/gensvm_pred.c
@@ -6,7 +6,7 @@
  *
  * @details
  * This file contains functions for predicting the class labels of instances
- * and a function for calculating the predictive performance (hitrate) of 
+ * and a function for calculating the predictive performance (hitrate) of
  * a prediction given true class labels.
  *
  */
@@ -24,7 +24,7 @@
  * @brief Predict class labels of data given and output in predy
  *
  * @details
- * The labels are predicted by mapping each instance in data to the 
+ * The labels are predicted by mapping each instance in data to the
  * simplex space using the matrix V in the given model. Next, for each
  * instance the nearest simplex vertex is determined using an Euclidean
  * norm. The nearest simplex vertex determines the predicted class label,
@@ -113,7 +113,7 @@ void gensvm_predict_labels(struct GenData *testdata, struct GenModel *model,
  *
  * @details
  * The predictive performance is calculated by simply counting the number
- * of correctly classified samples and dividing by the total number of 
+ * of correctly classified samples and dividing by the total number of
  * samples, multiplying by 100.
  *
  * @param[in] 	data 	the GenData dataset with known labels
diff --git a/src/gensvm_strutil.c b/src/gensvm_strutil.c
index aab441c..57e9399 100644
--- a/src/gensvm_strutil.c
+++ b/src/gensvm_strutil.c
@@ -38,7 +38,7 @@ bool str_endswith(const char *str, const char *suf)
 {
 	size_t lensuf = strlen(suf),
 	       lenstr = strlen(str);
-	return lenstr < lensuf ? false : strncmp(str + lenstr - lensuf, suf, 
+	return lenstr < lensuf ? false : strncmp(str + lenstr - lensuf, suf,
 			lensuf) == 0;
 }
 
@@ -110,7 +110,7 @@ long get_fmt_long(FILE *fid, char *filename, const char *fmt)
  *
  * @details
  * This function is used to read a line of doubles from a buffer. All the
- * doubles found are stored in a pre-allocated array. 
+ * doubles found are stored in a pre-allocated array.
  *
  * @param[in] 	buffer 		a string buffer
  * @param[in] 	offset 		an offset of the string to start looking for
@@ -131,7 +131,7 @@ long all_doubles_str(char *buffer, long offset, double *all_doubles)
 		if (start != end) {
 			all_doubles[i] = value;
 			i++;
-		} else 
+		} else
 			break;
 		start = end;
 		end = NULL;
@@ -145,7 +145,7 @@ long all_doubles_str(char *buffer, long offset, double *all_doubles)
  *
  * @details
  * This function is used to read a line of longs from a buffer. All the
- * longs found are stored in a pre-allocated array. 
+ * longs found are stored in a pre-allocated array.
  *
  * @param[in] 	buffer 		a string buffer
  * @param[in] 	offset 		an offset of the string to start looking for
diff --git a/src/gensvm_sv.c b/src/gensvm_sv.c
index 787b869..c61347a 100644
--- a/src/gensvm_sv.c
+++ b/src/gensvm_sv.c
@@ -5,7 +5,7 @@
  * @brief Calculate the number of support vectors
  *
  * @details
- * The function in this file can be used to calculate the number of support 
+ * The function in this file can be used to calculate the number of support
  * vectors are left in a model.
  *
  */
@@ -17,9 +17,9 @@
  * @brief Calculate the number of support vectors in a model
  *
  * @details
- * If an object is correctly classified, the number of classes for which the 
- * error q is larger than 1, is K-1 (i.e., there is no error w.r.t. any of the 
- * other classes).  All objects for which this is not the case are thus support 
+ * If an object is correctly classified, the number of classes for which the
+ * error q is larger than 1, is K-1 (i.e., there is no error w.r.t. any of the
+ * other classes).  All objects for which this is not the case are thus support
  * vectors.
  *
  * @param[in] 	model 	GenModel with solution
@@ -40,6 +40,6 @@ long gensvm_num_sv(struct GenModel *model, struct GenData *data)
 		}
 		num_sv += (num_correct < data->K - 1);
 	}
-	
+
 	return num_sv;
 }
diff --git a/src/gensvm_timer.c b/src/gensvm_timer.c
index a871887..93cd77e 100644
--- a/src/gensvm_timer.c
+++ b/src/gensvm_timer.c
@@ -31,7 +31,7 @@ double elapsed_time(clock_t s_time, clock_t e_time)
  *
  * @details
  * Create a string for the current system time. Include an offset of UTC for
- * consistency. The format of the generated string is "DDD MMM D HH:MM:SS 
+ * consistency. The format of the generated string is "DDD MMM D HH:MM:SS
  * YYYY (UTC +HH:MM)", e.g. "Fri Aug 9, 12:34:56 2013 (UTC +02:00)".
  *
  * @param[in,out] 	buffer 	allocated string buffer, on exit contains
diff --git a/src/gensvm_train.c b/src/gensvm_train.c
index 680d0dd..8c32809 100644
--- a/src/gensvm_train.c
+++ b/src/gensvm_train.c
@@ -3,7 +3,7 @@
  * @author Gertjan van den Burg
  * @date August 9, 2013
  * @brief Main functions for training the GenSVM solution.
- * 
+ *
  * @details
  * Contains update and loss functions used to actually find
  * the optimal V.
@@ -32,14 +32,14 @@
  * @details
  * This function is the main training function. This function
  * handles the optimization of the model with the given model parameters, with
- * the data given. On return the matrix GenModel::V contains the optimal 
+ * the data given. On return the matrix GenModel::V contains the optimal
  * weight matrix.
  *
  * In this function, step doubling is used in the majorization algorithm after
- * a burn-in of 50 iterations. If the training is finished, GenModel::t and 
+ * a burn-in of 50 iterations. If the training is finished, GenModel::t and
  * GenModel::W are extracted from GenModel::V.
  *
- * @param[in,out] 	model 	the GenModel to be trained. Contains optimal 
+ * @param[in,out] 	model 	the GenModel to be trained. Contains optimal
  * 				V on exit.
  * @param[in] 		data 	the GenData to train the model with.
  */
@@ -54,7 +54,7 @@ void gensvm_optimize(struct GenModel *model, struct GenData *data)
 
 	double *B = Calloc(double, n*(K-1));
 	double *ZV = Calloc(double, n*(K-1));
-	double *ZAZ = Calloc(double, (m+1)*(m+1));	
+	double *ZAZ = Calloc(double, (m+1)*(m+1));
 	double *ZAZV = Calloc(double, (m+1)*(K-1));
 	double *ZAZVT = Calloc(double, (m+1)*(K-1));
 
@@ -79,16 +79,16 @@ void gensvm_optimize(struct GenModel *model, struct GenData *data)
 
 	while ((it < MAX_ITER) && (Lbar - L)/L > model->epsilon)
 	{
-		// ensure V contains newest V and Vbar contains V from 
+		// ensure V contains newest V and Vbar contains V from
 		// previous
 		gensvm_get_update(model, data, B, ZAZ, ZAZV, ZAZVT);
 		if (it > 50)
 			gensvm_step_doubling(model);
-		
+
 		Lbar = L;
 		L = gensvm_get_loss(model, data, ZV);
 
-		if (it%100 == 0) 
+		if (it%100 == 0)
 			note("iter = %li, L = %15.16f, Lbar = %15.16f, "
 			     "reldiff = %15.16f\n", it, L, Lbar, (Lbar - L)/L);
 		it++;
@@ -124,20 +124,20 @@ void gensvm_optimize(struct GenModel *model, struct GenData *data)
 
 /**
  * @brief Calculate the current value of the loss function
- * 
+ *
  * @details
- * The current loss function value is calculated based on the matrix V in the 
+ * The current loss function value is calculated based on the matrix V in the
  * given model. Note that the matrix ZV is passed explicitly to avoid having
  * to reallocate memory at every step.
  *
- * @param[in] 		model 	GenModel structure which holds the current 
+ * @param[in] 		model 	GenModel structure which holds the current
  * 				estimate V
  * @param[in]  		data 	GenData structure
- * @param[in,out] 	ZV 	pre-allocated matrix ZV which is updated on 
+ * @param[in,out] 	ZV 	pre-allocated matrix ZV which is updated on
  * 				output
  * @returns 			the current value of the loss function
  */
-double gensvm_get_loss(struct GenModel *model, struct GenData *data, 
+double gensvm_get_loss(struct GenModel *model, struct GenData *data,
 		double *ZV)
 {
 	long i, j;
@@ -187,19 +187,19 @@ double gensvm_get_loss(struct GenModel *model, struct GenData *data,
  *
  * Because the function gensvm_get_update() is always called after a call to
  * gensvm_get_loss() with the same GenModel::V, it is unnecessary to calculate
- * the updated errors GenModel::Q and GenModel::H here too. This saves on 
+ * the updated errors GenModel::Q and GenModel::H here too. This saves on
  * computation time.
  *
- * In calculating the majorization coefficients we calculate the elements of a 
+ * In calculating the majorization coefficients we calculate the elements of a
  * diagonal matrix A with elements
  * @f[
- * 	A_{i, i} = \frac{1}{n} \rho_i \sum_{j \neq k} \left[ 
+ * 	A_{i, i} = \frac{1}{n} \rho_i \sum_{j \neq k} \left[
  * 		\varepsilon_i a_{ijk}^{(p)} + (1 - \varepsilon_i) \omega_i
  * 		a_{ijk}^{(p)} \right],
  * @f]
  * where @f$ k = y_i @f$.
  * Since this matrix is only used to calculate the matrix @f$ Z' A Z @f$, it is
- * efficient to update a matrix ZAZ through consecutive rank 1 updates with 
+ * efficient to update a matrix ZAZ through consecutive rank 1 updates with
  * a single element of A and the corresponding row of Z. The BLAS function
  * dsyr is used for this.
  *
@@ -225,7 +225,7 @@ double gensvm_get_loss(struct GenModel *model, struct GenData *data,
  * solving this system is done through dposv().
  *
  * @todo
- * Consider allocating IPIV and WORK at a higher level, they probably don't 
+ * Consider allocating IPIV and WORK at a higher level, they probably don't
  * change much during the iterations.
  *
  * @param [in,out] 	model 	model to be updated
@@ -301,7 +301,7 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 						b = 0.5 - kappa/2.0 - q;
 					} else if ( q <= 1.0) {
 						b = pow(1.0 - q, 3.0)/(
-							2.0*pow(kappa + 1.0, 
+							2.0*pow(kappa + 1.0,
 								2.0));
 					} else {
 						b = 0;
@@ -336,22 +336,22 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 					} else {
 						a = 0.25*pow(p, 2.0)*pow(
 							(p/(p - 2.0))*
-							(0.5 - kappa/2.0 - q), 
+							(0.5 - kappa/2.0 - q),
 							p - 2.0);
 						b = a*(2.0*q + kappa - 1.0)/
-							(p - 2.0) + 
+							(p - 2.0) +
 							0.5*p*pow(
 								p/(p - 2.0)*
 								(0.5 - kappa/
-								 2.0 - q), 
+								 2.0 - q),
 								p - 1.0);
 					}
 					if (q <= -kappa) {
 						b = 0.5*p*pow(
-							0.5 - kappa/2.0 - q, 
+							0.5 - kappa/2.0 - q,
 							p - 1.0);
 					} else if ( q <= 1.0) {
-						b = p*pow(1.0 - q, 
+						b = p*pow(1.0 - q,
 								2.0*p - 1.0)/
 							pow(2*kappa+2.0, p);
 					}
@@ -379,8 +379,8 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 		}
 		Avalue *= in * rho[i];
 
-		// Now we calculate the matrix ZAZ. Since this is 
-		// guaranteed to be symmetric, we only calculate the 
+		// Now we calculate the matrix ZAZ. Since this is
+		// guaranteed to be symmetric, we only calculate the
 		// upper part of the matrix, and then copy this over
 		// to the lower part after all calculations are done.
 		// Note that the use of dsym is faster than dspr, even
@@ -402,8 +402,8 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 		for (j=0; j<m+1; j++)
 			matrix_set(ZAZ, m+1, j, i, matrix_get(ZAZ, m+1, i, j));
 	*/
-	
-	// Calculate the right hand side of the system we 
+
+	// Calculate the right hand side of the system we
 	// want to solve.
 	cblas_dsymm(
 			CblasRowMajor,
@@ -417,7 +417,7 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 			model->V,
 			K-1,
 			0.0,
-			ZAZV, 
+			ZAZV,
 			K-1);
 
 	cblas_dgemm(
@@ -436,8 +436,8 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 			ZAZV,
 			K-1);
 
-	/* 
-	 * Add lambda to all diagonal elements except the first one. Recall 
+	/*
+	 * Add lambda to all diagonal elements except the first one. Recall
 	 * that ZAZ is of size m+1 and is symmetric.
 	 */
 	i = 0;
@@ -445,19 +445,19 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 		i += (m+1) + 1;
 		ZAZ[i] += model->lambda;
 	}
-	
+
 	// For the LAPACK call we need to switch to Column-
 	// Major order. This is unnecessary for the matrix
-	// ZAZ because it is symmetric. The matrix ZAZV 
+	// ZAZ because it is symmetric. The matrix ZAZV
 	// must be converted however.
 	for (i=0; i<m+1; i++)
 		for (j=0; j<K-1; j++)
 			ZAZVT[j*(m+1)+i] = ZAZV[i*(K-1)+j];
-	
-	// We use the lower ('L') part of the matrix ZAZ, 
+
+	// We use the lower ('L') part of the matrix ZAZ,
 	// because we have used the upper part in the BLAS
 	// calls above in Row-major order, and Lapack uses
-	// column major order. 
+	// column major order.
 
 	status = dposv(
 			'L',
@@ -470,10 +470,10 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 
 	if (status != 0) {
 		// This step should not be necessary, as the matrix
-		// ZAZ is positive semi-definite by definition. It 
+		// ZAZ is positive semi-definite by definition. It
 		// is included for safety.
 		fprintf(stderr, "GenSVM warning: Received nonzero status from "
-				"dposv: %i\n", 
+				"dposv: %i\n",
 				status);
 		int *IPIV = malloc((m+1)*sizeof(int));
 		double *WORK = malloc(1*sizeof(double));
@@ -507,7 +507,7 @@ void gensvm_get_update(struct GenModel *model, struct GenData *data, double *B,
 		free(IPIV);
 	}
 
-	// Return to Row-major order. The matrix ZAZVT contains the solution 
+	// Return to Row-major order. The matrix ZAZVT contains the solution
 	// after the dposv/dsysv call.
 	for (i=0; i<m+1; i++)
 		for (j=0; j<K-1; j++)
diff --git a/src/gensvm_train_dataset.c b/src/gensvm_train_dataset.c
index 4f42040..a70b457 100644
--- a/src/gensvm_train_dataset.c
+++ b/src/gensvm_train_dataset.c
@@ -6,7 +6,7 @@
  *
  * @details
  * The GenSVM algorithm takes a number of parameters. The functions in
- * this file are used to find the optimal parameters. 
+ * this file are used to find the optimal parameters.
  */
 
 #include <math.h>
@@ -31,7 +31,7 @@ extern FILE *GENSVM_OUTPUT_FILE;
  *
  * @details
  * A Training instance describes the grid to search over. This funtion
- * creates all tasks that need to be performed and adds these to 
+ * creates all tasks that need to be performed and adds these to
  * a Queue. Each task contains a pointer to the train and test datasets
  * which are supplied. Note that the tasks are created in a specific order of
  * the parameters, to ensure that the GenModel::V of a previous parameter
@@ -39,13 +39,13 @@ extern FILE *GENSVM_OUTPUT_FILE;
  * parameter set.
  *
  * @param[in] 	training 	Training struct describing the grid search
- * @param[in] 	queue 		pointer to a Queue that will be used to 
+ * @param[in] 	queue 		pointer to a Queue that will be used to
  * 				add the tasks to
  * @param[in] 	train_data 	GenData of the training set
  * @param[in] 	test_data 	GenData of the test set
  *
  */
-void make_queue(struct Training *training, struct Queue *queue, 
+void make_queue(struct Training *training, struct Queue *queue,
 		struct GenData *train_data, struct GenData *test_data)
 {
 	long i, j, k;
@@ -74,7 +74,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 		task->test_data = test_data;
 		task->folds = training->folds;
 		task->kerneltype = training->kerneltype;
-		task->kernelparam = Calloc(double, training->Ng + 
+		task->kernelparam = Calloc(double, training->Ng +
 				training->Nc + training->Nd);
 		queue->tasks[i] = task;
 	}
@@ -86,7 +86,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	cnt = 1;
 	i = 0;
 	while (i < N )
-		for (j=0; j<training->Np; j++) 
+		for (j=0; j<training->Np; j++)
 			for (k=0; k<cnt; k++) {
 				queue->tasks[i]->p = training->ps[j];
 				i++;
@@ -95,9 +95,9 @@ void make_queue(struct Training *training, struct Queue *queue,
 	cnt *= training->Np;
 	i = 0;
 	while (i < N )
-		for (j=0; j<training->Nl; j++) 
+		for (j=0; j<training->Nl; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->lambda = 
+				queue->tasks[i]->lambda =
 					training->lambdas[j];
 				i++;
 			}
@@ -116,7 +116,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	while (i < N )
 		for (j=0; j<training->Nw; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->weight_idx = 
+				queue->tasks[i]->weight_idx =
 					training->weight_idxs[j];
 				i++;
 			}
@@ -126,7 +126,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	while (i < N )
 		for (j=0; j<training->Ne; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->epsilon = 
+				queue->tasks[i]->epsilon =
 					training->epsilons[j];
 				i++;
 			}
@@ -136,7 +136,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	while (i < N && training->Ng > 0)
 		for (j=0; j<training->Ng; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->kernelparam[0] = 
+				queue->tasks[i]->kernelparam[0] =
 					training->gammas[j];
 				i++;
 			}
@@ -146,7 +146,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	while (i < N && training->Nc > 0)
 		for (j=0; j<training->Nc; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->kernelparam[1] = 
+				queue->tasks[i]->kernelparam[1] =
 					training->coefs[j];
 				i++;
 			}
@@ -156,7 +156,7 @@ void make_queue(struct Training *training, struct Queue *queue,
 	while (i < N && training->Nd > 0)
 		for (j=0; j<training->Nd; j++)
 			for (k=0; k<cnt; k++) {
-				queue->tasks[i]->kernelparam[2] = 
+				queue->tasks[i]->kernelparam[2] =
 					training->degrees[j];
 				i++;
 			}
@@ -285,7 +285,7 @@ struct Queue *create_top_queue(struct Queue *q)
 	}
 	nq->N = N;
 	nq->i = 0;
-	
+
 	return nq;
 }
 
@@ -298,12 +298,12 @@ struct Queue *create_top_queue(struct Queue *q)
  * The best performing tasks in the supplied Queue are found by taking those
  * Task structs that have a performance greater or equal to the 95% percentile
  * of the performance of all tasks. These tasks are then gathered in a new
- * Queue. For each of the tasks in this new Queue the cross validation run is 
- * repeated a number of times. 
+ * Queue. For each of the tasks in this new Queue the cross validation run is
+ * repeated a number of times.
  *
  * For each of the Task configurations that are repeated the mean performance,
  * standard deviation of the performance and the mean computation time are
- * reported. 
+ * reported.
  *
  * Finally, the overall best tasks are written to the specified output. These
  * tasks are selected to have both the highest mean performance, as well as the
@@ -317,7 +317,7 @@ struct Queue *create_top_queue(struct Queue *q)
  * an interval is found which contains tasks. If one or more tasks are found,
  * this loop stops.
  *
- * @param[in] 	q 		Queue of Task structs which have already been 
+ * @param[in] 	q 		Queue of Task structs which have already been
  * 				run and have a Task::performance value
  * @param[in] 	repeats 	Number of times to repeat the best
  * 				configurations for consistency
@@ -385,7 +385,7 @@ void consistency_repeats(struct Queue *q, long repeats, TrainType traintype)
 			matrix_set(perf, repeats, i, r, p);
 			mean[i] += p/((double) repeats);
 			note("%3.3f\t", p);
-			// this is done because if we reuse the V it's not a 
+			// this is done because if we reuse the V it's not a
 			// consistency check
 			gensvm_seed_model_V(NULL, model, task->train_data);
 			for (f=0; f<task->folds; f++) {
@@ -420,21 +420,21 @@ void consistency_repeats(struct Queue *q, long repeats, TrainType traintype)
 		pi = prctile(mean, N, (100.0-p));
 		pr = prctile(std, N, p);
 		pt = prctile(time, N, p);
-		for (i=0; i<N; i++) 
+		for (i=0; i<N; i++)
 			if ((pi - mean[i] < 0.0001) &&
-			    (std[i] - pr < 0.0001) && 
+			    (std[i] - pr < 0.0001) &&
 			    (time[i] - pt < 0.0001)) {
 				note("(%li)\tw = %li\te = %f\tp = %f\t"
 						"k = %f\tl = %f\t"
 						"mean: %3.3f\tstd: %3.3f\t"
 						"time: %3.3f\n",
-						nq->tasks[i]->ID, 
+						nq->tasks[i]->ID,
 						nq->tasks[i]->weight_idx,
-						nq->tasks[i]->epsilon, 
+						nq->tasks[i]->epsilon,
 						nq->tasks[i]->p,
-						nq->tasks[i]->kappa, 
+						nq->tasks[i]->kappa,
 						nq->tasks[i]->lambda,
-						mean[i], 
+						mean[i],
 						std[i],
 						time[i]);
 				breakout = true;
@@ -458,15 +458,15 @@ void consistency_repeats(struct Queue *q, long repeats, TrainType traintype)
  * @brief Check if the kernel parameters change between tasks
  *
  * @details
- * In the current strategy for training the kernel matrix is decomposed once, 
- * and tasks with the same kernel settings are performed sequentially. When a 
- * task needs to be done with different kernel parameters, the kernel matrix 
- * needs to be recalculated. This function is used to check whether this is 
+ * In the current strategy for training the kernel matrix is decomposed once,
+ * and tasks with the same kernel settings are performed sequentially. When a
+ * task needs to be done with different kernel parameters, the kernel matrix
+ * needs to be recalculated. This function is used to check whether this is
  * the case.
  *
  * @param[in] 	newtask 	the next task
  * @param[in] 	oldtask 	the old task
- * @return 			whether the kernel needs to be reevaluated 
+ * @return 			whether the kernel needs to be reevaluated
  */
 bool kernel_changed(struct Task *newtask, struct Task *oldtask)
 {
@@ -502,10 +502,10 @@ bool kernel_changed(struct Task *newtask, struct Task *oldtask)
  * cross_validation(), the optimal weights of one parameter set are used as
  * initial estimates for GenModel::V in the next parameter set. Note that to
  * optimally exploit this feature of the optimization algorithm, the order in
- * which tasks are considered is important. This is considered in 
+ * which tasks are considered is important. This is considered in
  * make_queue().
- * 
- * The performance found by cross validation is stored in the Task struct. 
+ *
+ * The performance found by cross validation is stored in the Task struct.
  *
  * @param[in,out] 	q 	Queue with Task instances to run
  */
@@ -519,7 +519,7 @@ void start_training(struct Queue *q)
 	struct GenModel *model = gensvm_init_model();
 	clock_t main_s, main_e, loop_s, loop_e;
 
-	// in principle this can change between tasks, but this shouldn't be 
+	// in principle this can change between tasks, but this shouldn't be
 	// the case TODO
 	folds = task->folds;
 
@@ -594,17 +594,17 @@ void start_training(struct Queue *q)
  * @brief Run cross validation with a given set of train/test folds
  *
  * @details
- * This cross validation function uses predefined train/test splits. Also, the  
- * the optimal parameters GenModel::V of a previous fold as initial conditions 
- * for GenModel::V of the next fold. 
+ * This cross validation function uses predefined train/test splits. Also, the
+ * the optimal parameters GenModel::V of a previous fold as initial conditions
+ * for GenModel::V of the next fold.
  *
  * @param[in] 	model 		GenModel with the configuration to train
  * @param[in] 	train_folds 	array of training datasets
  * @param[in] 	test_folds 	array of test datasets
  * @param[in] 	folds 		number of folds
- * @param[in] 	n_total 	number of objects in the union of the train 
+ * @param[in] 	n_total 	number of objects in the union of the train
  * datasets
- * @return 			performance (hitrate) of the configuration on 
+ * @return 			performance (hitrate) of the configuration on
  * cross validation
  */
 double gensvm_cross_validation(struct GenModel *model,
@@ -643,7 +643,7 @@ double gensvm_cross_validation(struct GenModel *model,
 	total_perf /= ((double) n_total);
 
 	return total_perf;
-}	
+}
 
 
 /**
@@ -735,9 +735,9 @@ void copy_model(struct GenModel *from, struct GenModel *to)
  * @brief Print the description of the current task on screen
  *
  * @details
- * To track the progress of the grid search the parameters of the current task 
- * are written to the output specified in GENSVM_OUTPUT_FILE. Since the 
- * parameters differ with the specified kernel, this function writes a 
+ * To track the progress of the grid search the parameters of the current task
+ * are written to the output specified in GENSVM_OUTPUT_FILE. Since the
+ * parameters differ with the specified kernel, this function writes a
  * parameter string depending on which kernel is used.
  *
  * @param[in] 	task 	the Task specified
diff --git a/src/gensvm_util.c b/src/gensvm_util.c
index aa4e5d9..db99e21 100644
--- a/src/gensvm_util.c
+++ b/src/gensvm_util.c
@@ -13,23 +13,23 @@
 
 #include "gensvm_util.h"
 
-FILE *GENSVM_OUTPUT_FILE; 	///< The #GENSVM_OUTPUT_FILE specifies the 
-				///< output stream to which all output is 
+FILE *GENSVM_OUTPUT_FILE; 	///< The #GENSVM_OUTPUT_FILE specifies the
+				///< output stream to which all output is
 				///< written. This is done through the
 				///< internal (!)
 				///< function gensvm_print_string(). The
-				///< advantage of using a global output 
+				///< advantage of using a global output
 				///< stream variable is that the output can
 				///< temporarily be suppressed by importing
-				///< this variable through @c extern and 
+				///< this variable through @c extern and
 				///< (temporarily) setting it to NULL.
 
 /**
  * @brief Check if any command line arguments contain string
  *
  * @details
- * Check if any of a given array of command line arguments contains a given 
- * string. If the string is found, the index of the string in argv is 
+ * Check if any of a given array of command line arguments contains a given
+ * string. If the string is found, the index of the string in argv is
  * returned. If the string is not found, 0 is returned.
  *
  * This function is copied from MSVMpack/libMSVM.c.
@@ -69,7 +69,7 @@ int gensvm_check_argv(int argc, char **argv, char *str)
  * @returns 		index of the command line argument that corresponds to
  * 			the string, 0 if none matches.
  */
-int gensvm_check_argv_eq(int argc, char **argv, char *str) 
+int gensvm_check_argv_eq(int argc, char **argv, char *str)
 {
 	int i;
 	int arg_str = 0;
@@ -88,7 +88,7 @@ int gensvm_check_argv_eq(int argc, char **argv, char *str)
  *
  * @details
  * This function is used to print a given string to the output stream
- * specified by #GENSVM_OUTPUT_FILE. The stream is flushed after the string 
+ * specified by #GENSVM_OUTPUT_FILE. The stream is flushed after the string
  * is written to the stream. If #GENSVM_OUTPUT_FILE is NULL, nothing is
  * written. Note that this function is only used by note(), it should never be
  * used directly.
diff --git a/src/libGenSVM.c b/src/libGenSVM.c
index b692bdb..c9b0b3c 100644
--- a/src/libGenSVM.c
+++ b/src/libGenSVM.c
@@ -6,8 +6,8 @@
  *
  * @details
  * The functions in this file are all functions needed
- * to calculate the optimal separation boundaries for 
- * a multiclass classification problem, using the 
+ * to calculate the optimal separation boundaries for
+ * a multiclass classification problem, using the
  * GenSVM algorithm.
  *
  */
@@ -23,10 +23,10 @@ inline double rnd() { return (double) rand()/0x7FFFFFFF; }
 
 /**
  * @brief Generate matrix of simplex vertex coordinates
- * 
+ *
  * @details
- * Generate the simplex matrix. Each row of the created 
- * matrix contains the coordinate vector of a single 
+ * Generate the simplex matrix. Each row of the created
+ * matrix contains the coordinate vector of a single
  * vertex of the K-simplex in K-1 dimensions. The simplex
  * generated is a special simplex with edges of length 1.
  * The simplex matrix U must already have been allocated.
@@ -83,9 +83,9 @@ void gensvm_category_matrix(struct GenModel *model, struct GenData *dataset)
  *
  * @details
  * The simplex difference matrix is a 3D matrix which is constructed
- * as follows. For each instance i, the difference vectors between the row of 
+ * as follows. For each instance i, the difference vectors between the row of
  * the simplex matrix corresponding to the class label of instance i and the
- * other rows of the simplex matrix are calculated. These difference vectors 
+ * other rows of the simplex matrix are calculated. These difference vectors
  * are stored in a matrix, which is one horizontal slice of the 3D matrix.
  *
  * @param[in,out] 	model 	the corresponding GenModel
@@ -113,11 +113,11 @@ void gensvm_simplex_diff(struct GenModel *model, struct GenData *data)
 
 /**
  * @brief Calculate the scalar errors
- * 
+ *
  * @details
  * Calculate the scalar errors q based on the current estimate of V, and
- * store these in Q. It is assumed that the memory for Q has already been 
- * allocated. In addition, the matrix ZV is calculated here. It is assigned 
+ * store these in Q. It is assumed that the memory for Q has already been
+ * allocated. In addition, the matrix ZV is calculated here. It is assigned
  * to a pre-allocated block of memory, which is passed to this function.
  *
  * @param[in,out] 	model 	the corresponding GenModel
@@ -164,16 +164,16 @@ void gensvm_calculate_errors(struct GenModel *model, struct GenData *data,
 			}
 		}
 	}
-}	
+}
 
 /**
  * @brief Calculate the Huber hinge errors
  *
  * @details
- * For each of the scalar errors in Q the Huber hinge errors are 
- * calculated. The Huber hinge is here defined as 
+ * For each of the scalar errors in Q the Huber hinge errors are
+ * calculated. The Huber hinge is here defined as
  * @f[
- * 	h(q) = 
+ * 	h(q) =
  * 		\begin{dcases}
  * 			1 - q - \frac{\kappa + 1}{2} & \text{if } q \leq -\kappa \\
  * 			\frac{1}{2(\kappa + 1)} ( 1 - q)^2 & \text{if } q \in (-\kappa, 1] \\
@@ -183,7 +183,7 @@ void gensvm_calculate_errors(struct GenModel *model, struct GenData *data,
  *
  * @param[in,out] model 	the corresponding GenModel
  */
-void gensvm_calculate_huber(struct GenModel *model) 
+void gensvm_calculate_huber(struct GenModel *model)
 {
 	long i, j;
 	double q, value;
@@ -206,10 +206,10 @@ void gensvm_calculate_huber(struct GenModel *model)
  * @brief seed the matrix V from an existing model or using rand
  *
  * @details
- * The matrix V must be seeded before the main_loop() can start. 
- * This can be done by either seeding it with random numbers or 
+ * The matrix V must be seeded before the main_loop() can start.
+ * This can be done by either seeding it with random numbers or
  * using the solution from a previous model on the same dataset
- * as initial seed. The latter option usually allows for a 
+ * as initial seed. The latter option usually allows for a
  * significant improvement in the number of iterations necessary
  * because the seeded model V is closer to the optimal V.
  *
@@ -221,11 +221,11 @@ void gensvm_seed_model_V(struct GenModel *from_model,
 {
 	long i, j, k;
 	double cmin, cmax, value;
-	
+
 	long n = data->n;
 	long m = data->m;
 	long K = data->K;
-	
+
 	if (from_model == NULL) {
 		for (i=0; i<m+1; i++) {
 			cmin = 1e100;
@@ -243,7 +243,7 @@ void gensvm_seed_model_V(struct GenModel *from_model,
 			}
 		}
 	} else {
-		for (i=0; i<m+1; i++) 
+		for (i=0; i<m+1; i++)
 			for (j=0; j<K-1; j++) {
 				value = matrix_get(from_model->V, K-1, i, j);
 				matrix_set(to_model->V, K-1, i, j, value);
@@ -255,7 +255,7 @@ void gensvm_seed_model_V(struct GenModel *from_model,
  * @brief Use step doubling
  *
  * @details
- * Step doubling can be used to speed up the maorization algorithm. Instead of 
+ * Step doubling can be used to speed up the maorization algorithm. Instead of
  * using the value at the minimimum of the majorization function, the value
  * ``opposite'' the majorization point is used. This can essentially cut the
  * number of iterations necessary to reach the minimum in half.
@@ -283,10 +283,10 @@ void gensvm_step_doubling(struct GenModel *model)
  * @brief Initialize instance weights
  *
  * @details
- * Instance weights can for instance be used to add additional weights to 
+ * Instance weights can for instance be used to add additional weights to
  * instances of certain classes. Two default weight possibilities are
- * implemented here. The first is unit weights, where each instance gets 
- * weight 1. 
+ * implemented here. The first is unit weights, where each instance gets
+ * weight 1.
  *
  * The second are group size correction weights, which are calculated as
  * @f[
@@ -296,8 +296,8 @@ void gensvm_step_doubling(struct GenModel *model)
  * @f$ y_i = k @f$.
  *
  * @param[in] 		data 	GenData with the dataset
- * @param[in,out] 	model 	GenModel with the weight specification. On 
- * 				exit GenModel::rho contains the instance 
+ * @param[in,out] 	model 	GenModel with the weight specification. On
+ * 				exit GenModel::rho contains the instance
  * 				weights.
  */
 void gensvm_initialize_weights(struct GenData *data, struct GenModel *model)
@@ -311,12 +311,12 @@ void gensvm_initialize_weights(struct GenData *data, struct GenModel *model)
 	if (model->weight_idx == 1) {
 		for (i=0; i<n; i++)
 			model->rho[i] = 1.0;
-	} 
+	}
 	else if (model->weight_idx == 2) {
 		groups = Calloc(long, K);
-		for (i=0; i<n; i++) 
+		for (i=0; i<n; i++)
 			groups[data->y[i]-1]++;
-		for (i=0; i<n; i++) 
+		for (i=0; i<n; i++)
 			model->rho[i] = ((double) n)/((double) (groups[data->y[i]-1]*K));
 	} else {
 		fprintf(stderr, "Unknown weight specification.\n");