release / genetic-gpu

Commit 34f743bedd107688353aca4e9797416a5acaa4ec

Authored by David Mayerich
1 parent a801c828

simplified CMakeLists.txt and changed the main.cpp to main.cu to facilitate comp… 

…iling in Visual Studio
Showing 2 changed files with 8 additions and 588 deletions   Show diff stats
Inline Side-by-side
@@ -2,40 +2,23 @@ @@ -2,40 +2,23 @@
2 cmake_minimum_required(VERSION 3.16) 2 cmake_minimum_required(VERSION 3.16)
3 3
4 #Name your project here 4 #Name your project here
5 -project(ga-gpu) 5 +project(genetic-gpu)
6 6
7 #set the module directory 7 #set the module directory
8 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}") 8 set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}")
9 9
10 -#default to release mode  
11 -if(NOT CMAKE_BUILD_TYPE)  
12 - set(CMAKE_BUILD_TYPE Release)  
13 -endif(NOT CMAKE_BUILD_TYPE) 10 +
14 11
15 #build the executable in the binary directory on MS Visual Studio 12 #build the executable in the binary directory on MS Visual Studio
16 if ( MSVC ) 13 if ( MSVC )
17 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_DEBUG "${OUTPUT_DIRECTORY}") 14 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_DEBUG "${OUTPUT_DIRECTORY}")
18 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_RELEASE "${OUTPUT_DIRECTORY}") 15 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_RELEASE "${OUTPUT_DIRECTORY}")
19 endif ( MSVC ) 16 endif ( MSVC )
20 -#MAYBE REMOVE-----------------  
21 -#set C++11 flags if using GCC  
22 -if( CMAKE_COMPILER_IS_GNUCC )  
23 -# SET( CMAKE_CXX_FLAGS "-std=c++11")  
24 - set(CMAKE_CXX_FLAGS "-std=c++11 -D_FORCE_INLINES")  
25 -# SET( CUDA_NVCC_FLAGS "-std=c++11")  
26 -endif( CMAKE_COMPILER_IS_GNUCC )  
27 17
28 -SET( CUDA_NVCC_FLAGS "--gpu-architecture=compute_50 --gpu-code=sm_50,compute_50")  
29 -#-----------------------------  
30 18
31 19
32 20
33 #find packages----------------------------------- 21 #find packages-----------------------------------
34 -#find OpenCV  
35 -#find_package(OpenCV REQUIRED)  
36 -#add_definitions(-DUSING_OPENCV)  
37 -  
38 -#find the pthreads package  
39 find_package(Threads) 22 find_package(Threads)
40 23
41 #find the X11 package 24 #find the X11 package
@@ -57,89 +40,39 @@ if( CMAKE_COMPILER_IS_GNUCC ) @@ -57,89 +40,39 @@ if( CMAKE_COMPILER_IS_GNUCC )
57 endif() 40 endif()
58 endif() 41 endif()
59 42
60 -#find FANN  
61 -#find_package(FANN REQUIRED)  
62 -  
63 -#find the GLUT library for visualization  
64 -#find_package(OpenGL REQUIRED)  
65 -#find_package(GLUT REQUIRED)  
66 -#if(WIN32)  
67 -# find_package(GLEW REQUIRED)  
68 -# include_directories(${GLEW_INCLUDE_DIR})  
69 -#endif(WIN32)  
70 -  
71 #find LAPACK and supporting link_libraries 43 #find LAPACK and supporting link_libraries
72 find_package(LAPACKE REQUIRED) 44 find_package(LAPACKE REQUIRED)
73 45
74 #include include directories 46 #include include directories
75 include_directories(${CUDA_INCLUDE_DIRS} 47 include_directories(${CUDA_INCLUDE_DIRS}
76 - ${OpenCV_INCLUDE_DIRS}  
77 ${LAPACKE_INCLUDE_DIR} 48 ${LAPACKE_INCLUDE_DIR}
78 ${STIM_INCLUDE_DIRS} 49 ${STIM_INCLUDE_DIRS}
79 - ${OpenGL_INCLUDE_DIRS}  
80 -# ${GLUT_INCLUDE_DIR}  
81 - ${FANN_INCLUDE_DIRS}  
82 "${CMAKE_SOURCE_DIR}/src" 50 "${CMAKE_SOURCE_DIR}/src"
83 ) 51 )
84 52
85 -#Assign a variable for all of the header files in this project 53 +#collect all source files
86 include_directories("${CMAKE_SOURCE_DIR}/src") 54 include_directories("${CMAKE_SOURCE_DIR}/src")
87 -#file(GLOB GACPU_H "${CMAKE_SOURCE_DIR}/src/gacpu/*.h")  
88 -file(GLOB GAGPU_H "${CMAKE_SOURCE_DIR}/src/*.h")  
89 -#file(GLOB GA_H "${CMAKE_SOURCE_DIR}/src/*.h")  
90 -  
91 -#Assign source files to the appropriate variables to easily associate them with executables  
92 -#file(GLOB GA_CPU_SRC "${CMAKE_SOURCE_DIR}/src/gacpu/*.cpp")  
93 -file(GLOB GA_GPU_SRC "${CMAKE_SOURCE_DIR}/src/*.c*") 55 +file(GLOB GA_GPU_SRC "${CMAKE_SOURCE_DIR}/src/*")
94 56
95 57
96 #create an executable file 58 #create an executable file
97 -cuda_add_executable(ga-gpu  
98 - ${GAGPU_H}  
99 -# ${GA_H} 59 +cuda_add_executable(genetic-gpu
100 ${GA_GPU_SRC} 60 ${GA_GPU_SRC}
101 ) 61 )
102 -target_link_libraries(ga-gpu ${CUDA_LIBRARIES} 62 +
  63 +target_link_libraries(genetic-gpu ${CUDA_LIBRARIES}
103 ${CUDA_CUBLAS_LIBRARIES} 64 ${CUDA_CUBLAS_LIBRARIES}
104 ${CUDA_CUFFT_LIBRARIES} 65 ${CUDA_CUFFT_LIBRARIES}
105 ${LAPACKE_LIBRARIES} 66 ${LAPACKE_LIBRARIES}
106 ${LAPACK_LIBRARIES} 67 ${LAPACK_LIBRARIES}
107 ${BLAS_LIBRARIES} 68 ${BLAS_LIBRARIES}
108 - ${CMAKE_THREAD_LIBS_INIT}  
109 ${X11_LIBRARIES} 69 ${X11_LIBRARIES}
110 - ${OpenCV_LIBS}  
111 ) 70 )
112 71
113 72
114 -#create the PROC executable----------------------------------------------  
115 -  
116 -#create an executable file  
117 -#add_executable(hsiga  
118 -# ${GACPU_H}  
119 -# ${GA_H}  
120 -# ${GA_CPU_SRC}  
121 -#)  
122 -#target_link_libraries(hsiga ${LAPACKE_LIBRARIES}  
123 -# ${LAPACK_LIBRARIES}  
124 -# ${BLAS_LIBRARIES}  
125 -# ${CMAKE_THREAD_LIBS_INIT}  
126 -# ${X11_LIBRARIES}  
127 -# ${OpenCV_LIBS}  
128 -#)  
129 -  
130 -  
131 -  
132 #if Boost is found, set an environment variable to use with preprocessor directives 73 #if Boost is found, set an environment variable to use with preprocessor directives
133 if(Boost_FILESYSTEM_FOUND) 74 if(Boost_FILESYSTEM_FOUND)
134 -# if(BUILD_GACPU)  
135 -# target_link_libraries(hsiga ${Boost_FILESYSTEM_LIBRARIES}  
136 -# ${Boost_SYSTEM_LIBRARY}  
137 -# )  
138 - #message(${Boost_FILESYSTEM_LIBRARIES})  
139 -# endif(BUILD_GACPU)  
140 -# if(BUILD_GAGPU)  
141 - target_link_libraries(ga-gpu ${Boost_FILESYSTEM_LIBRARIES} 75 + target_link_libraries(genetic-gpu ${Boost_FILESYSTEM_LIBRARIES}
142 ${Boost_SYSTEM_LIBRARY} 76 ${Boost_SYSTEM_LIBRARY}
143 ) 77 )
144 -# endif(BUILD_GAGPU)  
145 endif(Boost_FILESYSTEM_FOUND) 78 endif(Boost_FILESYSTEM_FOUND)
src/main.cpp deleted
View file @a801c82
1 -#include <iostream>  
2 -  
3 -//stim libraries  
4 -#include <stim/envi/envi.h>  
5 -#include <stim/image/image.h>  
6 -#include <stim/ui/progressbar.h>  
7 -#include <stim/parser/filename.h>  
8 -#include <stim/parser/table.h>  
9 -#include <stim/parser/arguments.h>  
10 -//input arguments  
11 -stim::arglist args;  
12 -#include <fstream>  
13 -#include <thread>  
14 -#include <random>  
15 -#include <vector>  
16 -#include <math.h>  
17 -#include <limits>  
18 -  
19 -#define NOMINMAX  
20 -  
21 -  
22 -  
23 -//GA  
24 -#include "ga_gpu.h"  
25 -#include "enviload.h"  
26 -  
27 -  
28 -//envi input file and associated parameters  
29 -stim::envi E; //ENVI binary file object  
30 -unsigned int B; //shortcuts storing the spatial and spectral size of the ENVI image  
31 -//mask and class information used for training  
32 -//std::vector< stim::image<unsigned char> > C; //2D array used to access each mask C[m][p], where m = mask# and p = pixel#  
33 -std::vector<unsigned int> nP; //array holds the number of pixels in each mask: nP[m] is the number of pixels in mask m  
34 -size_t nC = 0; //number of classes  
35 -size_t tP = 0; //total number of pixels in all masks: tP = nP[0] + nP[1] + ... + nP[nC]  
36 -float* fea;  
37 -  
38 -//ga_gpu class object  
39 -ga_gpu ga;  
40 -bool debug;  
41 -bool binaryClass;  
42 -int binClassOne;  
43 -  
44 -//creating struct to pass to thread functions as it limits number of arguments to 3  
45 -typedef struct {  
46 - float* S;  
47 - float* Sb;  
48 - float* Sw;  
49 - float* lda;  
50 -}gnome;  
51 -gnome gnom;  
52 -  
53 -  
54 -void gpuComputeEignS( size_t g, size_t fea){  
55 - //eigen value computation will return r = (nC-1) eigen vectors so new projected data will have dimension of r rather than f  
56 - // std::thread::id this_id = std::this_thread::get_id();  
57 - // std::cout<<"thread id is "<< this_id<<std::endl;  
58 - size_t f = fea;  
59 - //std::thread::id g = std::this_thread::get_id();  
60 - float* LeftEigVectors_a = (float*) malloc(f * f * sizeof(float));  
61 - float* gSw_a = (float*) malloc(f * f * sizeof(float)); //copy of between class scatter  
62 - std::memcpy(gSw_a, &gnom.Sw[g * f * f], f * f *sizeof(float));  
63 - if(debug){  
64 - std::cout<<"From Eigen function: Sb and Sw "<<std::endl;  
65 - displayS(gSw_a, f); //display Sb  
66 - displayS(&gnom.Sb[g * f * f], f); //display Sw  
67 - std::cout<<std::endl;  
68 - }  
69 -  
70 - std::vector<unsigned int> features = ga.getGnome(g);  
71 - std::vector<unsigned int> featuresunique;  
72 - int flag = 0;  
73 - std::sort(features.begin(), features.end()); // 1 1 2 2 3 3 3 4 4 5 5 6 7  
74 - std::unique_copy(features.begin(), features.end(), std::back_inserter(featuresunique));  
75 - if(featuresunique.size()< features.size()){  
76 - f = featuresunique.size();  
77 - }  
78 -  
79 - size_t r = nC-1; //LDA projected dimension (limited to number of classes - 1 by rank)  
80 - if(r > f){  
81 - r = f;  
82 - }  
83 -  
84 - int info;  
85 - float* EigenvaluesI_a = (float*)malloc(f * sizeof(float));  
86 - float* Eigenvalues_a = (float*)malloc(f * sizeof(float));  
87 - int *IPIV = (int*) malloc(sizeof(int) * f);  
88 - //computing inverse of matrix Sw  
89 - memset(IPIV, 0, f * sizeof(int));  
90 - LAPACKE_sgetrf(LAPACK_COL_MAJOR, (int)f, (int)f, gSw_a, (int)f, IPIV);  
91 - // DGETRI computes the inverse of a matrix using the LU factorization computed by DGETRF.  
92 - LAPACKE_sgetri(LAPACK_COL_MAJOR, (int)f, gSw_a, (int)f, IPIV);  
93 -  
94 - float* gSbSw_a = (float*)calloc(f * f, sizeof(float));  
95 - //mtxMul(gSbSw_a, gSw_a, &gnom.Sb[g * f * f * sizeof(float)], f, f, f,f);  
96 - mtxMul(gSbSw_a, gSw_a, &gnom.Sb[g * f * f], f, f, f,f);  
97 - if(debug){  
98 - std::cout<<"From Eigen function: inverse of sw and ratio of sb and sw (Sb/Sw)";  
99 - displayS(gSw_a, f); //display inverse of Sw (1/Sw)  
100 - displayS(gSbSw_a, f); //display ratio of Sb and Sw (Sb/Sw)  
101 - }  
102 -  
103 - //compute left eigenvectors for current gnome from ratio of between class scatter and within class scatter: Sb/Sw  
104 - info = LAPACKE_sgeev(LAPACK_COL_MAJOR, 'V', 'N', (int)f, gSbSw_a, (int)f, Eigenvalues_a, EigenvaluesI_a, LeftEigVectors_a, (int)f, 0, (int)f);  
105 - //sort eignevalue indices in descending order  
106 - size_t* sortedindx = sortIndx(Eigenvalues_a, f);  
107 - //displayS(LeftEigVectors_a, f); //display Eignevectors (Note these are -1 * matlab eigenvectors does not change fitness score results but keep in mind while projecting data on it)  
108 - //sorting left eigenvectors (building forward transformation matrix As)  
109 - for (size_t rowE = 0; rowE < r; rowE++){  
110 - for (size_t colE = 0; colE < f; colE++){  
111 - size_t ind1 = g * r * f + rowE * f + colE;  
112 - //size_t ind1 = rowE * f + colE;  
113 - size_t ind2 = sortedindx[rowE] * f + colE; //eigenvector as row vector  
114 - gnom.lda[ind1] = LeftEigVectors_a[ind2];  
115 - }  
116 - }  
117 -  
118 - if(debug){  
119 - std::cout<<"Eigenvalues are"<<std::endl;  
120 - for(size_t n = 0 ; n < f; n ++){  
121 - std::cout << Eigenvalues_a[n] << ", " ;  
122 - }  
123 - std::cout<< std::endl;  
124 - std::cout<<"From Eigen function: Eignevector"<<std::endl;  
125 -  
126 - std::cout<<"LDA basis is "<<std::endl;  
127 - std::cout << "r is " << r << std::endl;  
128 - for(size_t l = 0 ; l < r; l++){  
129 - for(size_t n = 0 ; n < f; n ++){  
130 - std::cout << gnom.lda[g * l * f + l * f + n] << ", " ;  
131 - }  
132 - std::cout<<std::endl;  
133 - }  
134 -  
135 - }  
136 - //Extract only r eigne vectors as a LDA projection basis  
137 - float* tempgSb = (float*)calloc(r * f, sizeof(float));  
138 - //mtxMul(tempgSb, &gnom.lda[g * r * f * sizeof(float)], &gnom.Sb[g * f * f * sizeof(float)], r, f, f,f);  
139 - //mtxMul(tempgSb, &lda[g * r * f ], gSb, r, f, f,f);  
140 - mtxMul(tempgSb, &gnom.lda[g * r * f], &gnom.Sb[g * f * f], r, f, f,f);  
141 - float* nSb = (float*)calloc(r * r, sizeof(float));  
142 - mtxMultranspose(nSb, tempgSb, &gnom.lda[g * r * f], r, f, r, f);  
143 -  
144 - float* tempgSw = (float*)calloc(r * f, sizeof(float));  
145 - //mtxMul(tempgSw, &gnom.lda[g * r * f * sizeof(float)], &gnom.Sw[g * f * f * sizeof(float)], r, f, f,f);  
146 - mtxMul(tempgSw, &gnom.lda[g * r * f], &gnom.Sw[g * f * f], r, f, f,f);  
147 - float* nSw = (float*)calloc(r * r, sizeof(float));  
148 - mtxMultranspose(nSw, tempgSw, &gnom.lda[g * r * f], r, f, r, f);  
149 - if(debug){  
150 - std::cout<<"From Eigen function: projected Sb sn Sw"<<std::endl;  
151 - displayS(nSb, r); //display Sb  
152 - displayS(nSw, r); //display Sw  
153 - std::cout<<std::endl;  
154 - }  
155 -  
156 - ///DETERMINANT CURRENTLY REQUIRES OPENCV  
157 - std::cout<<"ERROR: This code requires a fix to remove an OpenCV dependence."<<std::endl;  
158 - mtxOutputFile("newSw.csv", nSw, r, r);  
159 - exit(1);  
160 - //FIX BY REPLACING THE FOLLOWING THREE LINES OF CODE USING LAPACK  
161 - //cv::Mat newSw = cv::Mat((int)r, (int)r, CV_32FC1, nSw); //within scatter of gnome g in the population  
162 - //cv::Mat newSb = cv::Mat((int)r, (int)r, CV_32FC1, nSb); //within scatter of gnome g in the population  
163 -  
164 - //fisher's ratio from ratio of projected sb and sw  
165 - float fisherRatio = 0;// = cv::determinant(newSb) /cv::determinant(newSw);  
166 - gnom.S[g] = 1/fisherRatio;  
167 - if (debug) {  
168 - std::cout<<"Score["<<g<<"]: "<< gnom.S[g]<<std::endl;  
169 -  
170 - std::cout << "best gnoem is " << std::endl;  
171 - for (size_t i = 0; i < f; i++)  
172 - std::cout << ga.P[ga.f * g + i] << ", ";  
173 - std::cout << std::endl;  
174 - }  
175 -  
176 -  
177 - if(IPIV!= NULL) std::free(IPIV);  
178 - if(gSw_a!= NULL) std::free(gSw_a);  
179 - if(gSbSw_a!= NULL) std::free(gSbSw_a);  
180 - if(Eigenvalues_a!= NULL) std::free(Eigenvalues_a);  
181 - if(EigenvaluesI_a!= NULL) std::free(EigenvaluesI_a);  
182 - if(tempgSb!= NULL) std::free(tempgSb);  
183 - if(tempgSw!= NULL) std::free(tempgSw);  
184 -  
185 -}  
186 -  
187 -  
188 -void fitnessFunction( float* sb, float* sw, float* lda, float* M, float* cM, size_t f, cudaDeviceProp props, size_t gen, std::ofstream& profilefile){  
189 -  
190 - size_t tP = 0; //total number of pixels  
191 - std::for_each(nP.begin(), nP.end(), [&] (size_t n) {  
192 - tP += n;  
193 - });  
194 - size_t nC = nP.size(); //total number of classes  
195 -  
196 - //--------------Compute between class scatter  
197 -  
198 - ga.gpu_computeSbSw(sb, sw, nC, tP, props, gen, debug, profilefile);  
199 -  
200 - if(debug){  
201 - std::cout<<"From fitness function: gpu results of Sb sn Sw"<<std::endl;  
202 - displayS(sb, ga.f); //display Sb  
203 - displayS(sw, ga.f); //display Sw  
204 - std::cout<<std::endl;  
205 - }  
206 -  
207 - // ----------------------- Linear discriminant Analysis --------------------------------------  
208 - gnom.S = ga.S;  
209 - gnom.Sw = sw;  
210 - gnom.Sb = sb;  
211 - gnom.lda = lda;  
212 -  
213 - //calling function without using threads  
214 - for (size_t i = 0; i<ga.p; i++){  
215 - //calling function for eigencomputation  
216 - gpuComputeEignS(i, f);  
217 - }  
218 -  
219 - const auto elapsed1 = timer.time_elapsed();  
220 - if(gen > ga.gnrtn - 2){  
221 - std::cout << "gpu_eigen time "<<std::chrono::duration_cast<std::chrono::microseconds>(elapsed1).count() << "us" << std::endl;  
222 - profilefile<< "gpu_eigen time "<<std::chrono::duration_cast<std::chrono::microseconds>(elapsed1).count() << "us" << std::endl;  
223 - }  
224 -  
225 -}//end of fitness function  
226 -  
227 -void binaryclassifier(int classnum){  
228 - unsigned int* target = (unsigned int*) calloc(tP, sizeof(unsigned int));  
229 - memcpy(target, ga.T, tP * sizeof(unsigned int));  
230 - for(int i = 0 ; i < tP; i++){  
231 - if(target[i]==classnum){  
232 - ga.T[i] = 1;  
233 -  
234 - }else  
235 - ga.T[i] = 0;  
236 - }  
237 -}  
238 -  
239 -  
240 -  
241 -void advertisement() {  
242 - std::cout << std::endl;  
243 - std::cout << "=========================================================================" << std::endl;  
244 - std::cout << "Thank you for using the GA-GPU features selection for spectroscopic image!" << std::endl;  
245 - std::cout << "=========================================================================" << std::endl << std::endl;  
246 -}  
247 -  
248 -int main(int argc, char* argv[]){  
249 -  
250 -//Add the argument options and set some of the default parameters  
251 - args.add("help", "print this help");  
252 - args.section("Genetic Algorithm");  
253 - args.add("features", "select features selection algorithm parameters","10", "number of features to be selected");  
254 - args.add("classes", "image masks used to specify classes", "", "class1.bmp class2.bmp class3.bmp");  
255 - args.add("population", "total number of feature subsets in puplation matrix", "1000");  
256 - args.add("generations", "number of generationsr", "50");  
257 - args.add("initial_guess", "initial guess of featues", "");  
258 - args.add("debug", "display intermediate data for debugging");  
259 - args.add("binary", "Calculate features based on class1 vs. all other classes", "");  
260 - args.add("trim", "this gives wavenumber to use in trim option of siproc which trims all bands from envi file except gagpu selected bands");  
261 -  
262 - args.parse(argc,argv); //parse the command line arguments  
263 -  
264 -//Print the help text if set  
265 - if(args["help"].is_set()){ //display the help text if requested  
266 - advertisement();  
267 - std::cout<<std::endl<<"usage: ga-gpu input_ENVI output.txt --classes class1.bmp class2.bmp ... --option [A B C ...]"<<std::endl;  
268 - std::cout<<std::endl<<std::endl;  
269 - std::cout<<args.str()<<std::endl;  
270 - exit(1);  
271 - }  
272 - if (args.nargs() < 2) { //if the user doesn't provide input and output files  
273 - std::cout << "ERROR: GA-GPU requires an input (ENVI) file and an output (features, text) file." << std::endl;  
274 - return 1;  
275 - }  
276 - if (args["classes"].nargs() < 2) { //if the user doesn't specify at least two class images  
277 - std::cout << "ERROR: GA-GPU requires at least two class images to be specified using the --classes option" << std::endl;  
278 - return 1;  
279 - }  
280 -  
281 - std::string outfile = args.arg(1); //outfile is text file where bnad index, LDA-basis, wavelength and if --trim option is set then trim wavelengths are set respectively  
282 - std::string profile_file = "profile_" + outfile ;  
283 - std::ofstream profilefile(profile_file.c_str(), std::ios::out); //open outfstream for outfile  
284 -  
285 - time_t t_start = time(NULL); //start a timer for file reading  
286 - E.open(args.arg(0), std::string(args.arg(0)) + ".hdr"); //open header file  
287 - size_t X = E.header.samples; //total number of pixels in X dimension  
288 - size_t Y = E.header.lines; //total number of pixels in Y dimension  
289 - B = (unsigned int)E.header.bands; //total number of bands (features)  
290 - std::vector<double> wavelengths = E.header.wavelength; //wavelengths of each band  
291 -  
292 - if(E.header.interleave != stim::envi_header::BIP){ //this code can only load bip files and hence check that in header file  
293 - std::cout<<"this code works for only bip files. please convert file to bip file"<<std::endl;  
294 - exit(1); //if file is not bip file exit code execution  
295 - }  
296 -  
297 -///--------------------------Load features---------------------------------------------  
298 - nP = ga_load_class_images(argc, args, &nC, &tP); //load supervised class images  
299 - ga.F = load_features( nC, tP, B, E, nP); //generate the feature matrix  
300 - ga.T = ga_load_responses(tP, nC, nP); //load the responses for RF training  
301 - E.close(); //close the hyperspectral file  
302 - time_t t_end = time(NULL);  
303 - std::cout<<"Total time: "<<t_end - t_start<<" s"<<std::endl;  
304 -  
305 -///--------------------------Genetic algorith configurations with defult paramets and from argument values---------------------  
306 - ga.f = args["features"].as_int(0); //number of features to be selected by user default value is 10  
307 - ga.p = args["population"].as_int(0); //population size to be selected by user default value is 1000  
308 - ga.gnrtn = args["generations"].as_int(0); //number of generations to be selected by user default value is 50  
309 - if(args["binary"]) { //set this option when features are to be selected as binary clas features (class vs stroma)  
310 - binClassOne = args["binary"].as_int(0); //sel class number here, if 2 then features are selected for (class-2 vs stroma)  
311 - //feture selection for class selected by user with user arguments (make it binary class data by making chosen class label as 1 and al other class labels 0 from multiclass data )  
312 - //to select feature for all classes in joint class data using binary class system need to write a script with loop covering all classes  
313 - binaryclassifier(binClassOne);  
314 - } ///not fully implemented yet  
315 -  
316 - ga.ub = B; //upper bound is number of bands (i.e. size of z dimension) Note: for this particular application and way code is written lower bound is 0 and upper bound is size of z dimension  
317 - ga.uniformRate = 0.5; //uniform rate is used in crossover  
318 - ga.mutationRate = 0.5f; //in percentage for mutation operation on gnome  
319 - ga.tournamentSize = 5; //for crossover best parents are selected from tournament of gnomes  
320 - ga.elitism = true; // if it is true then best gnome of current generation is passed to next generation  
321 - //initial guess of population  
322 - ga.i_guess = (unsigned int*) calloc(ga.f, sizeof(unsigned int));  
323 - debug = args["debug"];  
324 -  
325 -//==================Generate intial population =================================================  
326 - std::vector<unsigned int> i_guess(ga.f);  
327 - for (size_t b = 0; b < ga.f; b++) //generate default initial guess  
328 - i_guess[b] = rand() % B + 0;  
329 -  
330 - if (args["initial_guess"].is_set()) {//if the user specifies the --initialguess option & provides feature indices as initial guess  
331 - size_t nf = args["initial_guess"].nargs(); //get the number of arguments after initial_guess  
332 - if (nf == 1 || nf == ga.f) { //check if file with initial guessed indices is given or direct indices are given as argument  
333 - if (nf == 1) { //if initial guessed feature indices are given in file  
334 - std::ifstream in; //open the file containing the baseline points  
335 - in.open(args["initial_guess"].as_string().c_str());  
336 - if (in.is_open()){ //if file is present and can be opened then read it  
337 - unsigned int b_ind;  
338 - while (in >> b_ind) //get each band index and push it into the vector  
339 - i_guess.push_back(b_ind);  
340 - }  
341 - else  
342 - std::cout << "cannot open file of initial_guess indices" << std::endl;  
343 - }  
344 - else if (nf == ga.f) { //if direct indices are given as argument  
345 - for (size_t b = 0; b < nf; b++) //for each band given by the user  
346 - i_guess[b] = args["initial_guess"].as_int(b); //store that band in the i_guess array  
347 - }  
348 - }  
349 - }  
350 -  
351 - ga.initialize_population(i_guess, debug); //initialize first population set for first generation, user can pass manually preferred features from command line  
352 - //display_gnome(0);  
353 -  
354 -//------------------Calculate class means and total mean of features----------------------------  
355 - float* M = (float*)calloc( B , sizeof(float)); //total mean of entire feature martix for all features (bands B)  
356 - ga.ttlMean(M, tP, B); //calculate total mean, ga.F is entire feature matrix, M is mean for all bands B(features)  
357 - if(debug) ga.dispalymean(M); //if option --debug is used display all bands mean  
358 -  
359 - //display band index of bands with mean zero, this indicates that band has all zero values  
360 - std::cout<<"Display features indices with zero mean "<<std::endl;  
361 - for(unsigned int i = 0; i < B; i++){  
362 - if(M[i]== 0)  
363 - std::cout<<"\t"<<i;  
364 - }  
365 - std::cout<<std::endl;  
366 -// std::cout << "pixel target is " << ga.T[0] << " " << ga.T[1] << " " << ga.T[tP - 2] << " " << ga.T[tP - 1]<<std::endl;  
367 - float* cM = (float*)calloc(nC * B , sizeof(float)); //cM is nC X B matrix with each row as mean of all samples in one class for all features (bands B)  
368 - ga.classMean(cM, tP, nC, B, nP); //calculate class mean, ga.F is entire feature matrix, M is mean for all bands B(features)  
369 - if(debug) ga.dispalyClassmean(cM, nC);  
370 -  
371 -//------------------------------------GPU init----------------------------------------------------  
372 - //checking for cuda device  
373 - int count;  
374 - HANDLE_ERROR(cudaGetDeviceCount(&count));  
375 - if(count < 1){  
376 - std::cout<<"no cuda device is available"<<std::endl;  
377 - return 1;  
378 - }  
379 - cudaDeviceProp props;  
380 - HANDLE_ERROR(cudaGetDeviceProperties(&props, 0));  
381 - ga.gpuInitializationfrommain(M, cM, nP, tP, nC);  
382 -  
383 -  
384 -//============================= GA evolution by generations ====================================================  
385 - std::vector<unsigned int> bestgnome; //holds best gnome after each generation evaluation  
386 - size_t bestG_Indx; //This gives index of best gnome in the current population to get best gnome and its fitness value  
387 - unsigned int* newPop = (unsigned int*) calloc(ga.p * ga.f, sizeof(unsigned int)); //temprory storage of new population  
388 - double* best_S = (double*) calloc (ga.gnrtn, sizeof(double)); //stores fitness value of best gnome at each iteration  
389 - float* lda = (float*) calloc (ga.p * (nC-1) * ga.f, sizeof(float)); //stores LDA basis for each gnome so that we can have best gnome's LDA basis  
390 - float* sb = (float*) calloc( ga.p * ga.f * ga.f , sizeof(float)) ; //3d matrix for between class scatter (each 2d matrix between class scatter for one gnome)  
391 - float* sw = (float*) calloc( ga.p * ga.f * ga.f , sizeof(float)) ; //3d matrix for within class scatter (each 2d matrix within class scatter for one gnome)  
392 - ga.zerobandcheck(M, true); //checking bands with all zeros and duplicated bands in a gnome replacing them with other bands avoiding duplication and zero mean  
393 - ga.zerobandcheck(M, true); //Repeating zeroband cheack as some of these bands are not replaced in previous run and gave random results  
394 - time_t gpu_timestart = time(NULL); //start a timer for total evoluation  
395 -  
396 - for (size_t gen = 0; gen < ga.gnrtn; gen++){ //for each generation find fitness value of all gnomes in population matrix and generate population for next generation  
397 - //std::cout<<"Generation: "<<gen<<std::endl;  
398 - fitnessFunction(sb, sw, lda, M , cM, ga.f, props, gen, profilefile); //Evaluate phe(feature matrix for current population) for fitness of all gnomes in current population  
399 - timer.start(); //start timer for new population generation  
400 - bestG_Indx = ga.evolvePopulation(newPop, M, debug); //evolve population to generate new generation population  
401 - const auto pop_generation = timer.time_elapsed(); // end timer for new population generation  
402 - if(gen >ga.gnrtn -2){  
403 - std::cout << "population evolution time "<<std::chrono::duration_cast<std::chrono::microseconds>(pop_generation).count() << "us" << std::endl;  
404 - profilefile<<"population evolution time "<<std::chrono::duration_cast<std::chrono::microseconds>(pop_generation).count() << "us" <<std::endl;  
405 - }  
406 -  
407 - best_S[gen] = ga.S[bestG_Indx]; //score of best gnome in current generation  
408 - bestgnome = ga.getGnome(bestG_Indx); //Best gnome of current populaation  
409 - ga.generateNewP(newPop); //replace current population with new populaiton in the ga classs object  
410 - ga.zerobandcheck(M, false); //checking bands with all zeros and duplicated bands in a gnome replacing them with other bands avoiding duplication and zero mean  
411 - ga.zerobandcheck(M, false); //Repeating zeroband cheack as some of these bands are not replaced in previous run and gave random results  
412 - }//end generation  
413 -  
414 - time_t gpu_timeend = time(NULL); //end a timer for total evoluation  
415 - std::cout<<"Total gpu time: "<<gpu_timeend - gpu_timestart<<" s"<<std::endl;  
416 - profilefile<<"Total gpu time: "<<gpu_timeend - gpu_timestart<<" s"<<std::endl;  
417 -  
418 -//================================ Results of GA ===============================================================  
419 - std::cout<<"best gnome's fitness value is "<<best_S[ga.gnrtn-1]<<std::endl;  
420 - std::cout<<"best gnome is: ";  
421 - for(size_t i = 0; i < ga.f; i++){  
422 - std::cout<<" "<<(bestgnome.at(i));  
423 - }  
424 - std::cout<<std::endl;  
425 -  
426 - //create a text file to store the LDA stats (features subset and LDA-basis)  
427 - ////format of CSV file is: 1st row - band index, 2nd LDA basis depending on number of classes, 3rd - wavenumber corresponding to band index and it --trim is selected then trim wavnumbersare also given  
428 - std::ofstream csv(outfile.c_str(), std::ios::out); //open outfstream for outfile  
429 - size_t ldaindx = bestG_Indx * (nC-1) * ga.f ; //Compute LDA basis index of best gnome  
430 -  
431 -  
432 - //fitness values of best gnome is  
433 - csv<<"best gnome's fitness value is "<<best_S[ga.gnrtn-1]<<std::endl; //output fitness value of best gnome in last generation  
434 - //output gnome i.e. band index of selected featurs  
435 - csv<<(bestgnome.at(0)); //output feature subset  
436 - for(size_t i = 1; i < ga.f; i++)  
437 - csv<<","<<(bestgnome.at(i));  
438 - csv<<std::endl;  
439 -  
440 - //output LDA basis of size r X f, r is nC - 1 as LDA projection is rank limited by number of classes - 1  
441 - for (size_t i = 0; i < nC-1; i++){  
442 - csv<<lda[ldaindx + i * ga.f ];  
443 - for (size_t j = 1; j < ga.f; j++){  
444 - csv<<","<<lda[ldaindx + i * ga.f +j];  
445 - }  
446 - csv << std::endl;  
447 - }  
448 - //output actual wavelenths corresponding to those band indices  
449 - csv << (wavelengths[bestgnome.at(0)]);  
450 - for (size_t i = 1; i < ga.f; i++)  
451 - csv << "," << (wavelengths[bestgnome.at(i)]);  
452 - csv << std::endl;  
453 -  
454 -  
455 - if (args["trim"].is_set()) {  
456 - csv << "trim info" << std::endl;  
457 - std::sort(bestgnome.begin(), bestgnome.end()); //sort features index in best gnome  
458 -  
459 - std::vector<unsigned int> trimindex(ga.f); //create a vector to store temprory trim index bounds  
460 - std::vector<unsigned int> finaltrim_ind; //create a vector to store final trim index bounds  
461 - std::vector<unsigned int> trim_wv; //create a vector to store final trim wavelength bounds  
462 -  
463 - //trim index  
464 - trimindex.push_back(1); //1st trimming band index is 1  
465 - for (size_t i = 0; i < ga.f; i++) { // for each feature find its bound indexes  
466 - trimindex[i * 2] = bestgnome.at(i) - 1;  
467 - trimindex[i * 2 + 1] = bestgnome.at(i) + 1;  
468 - }  
469 - trimindex.push_back(B); //last bound index is B  
470 -  
471 - //organize trim index  
472 - int k = 0;  
473 - for (size_t i = 0; i < ga.f + 1; i++) { // find valid pair of trim indices bound excluding adjacent trim indices  
474 - if (trimindex[2 * i] < trimindex[2 * i + 1]) {  
475 - finaltrim_ind.push_back(trimindex[2 * i]); //this is left bound  
476 - finaltrim_ind.push_back(trimindex[2 * i + 1]);  
477 - k = k + 2;  
478 - }  
479 - }  
480 - //add duplicated trim indices as single index to final trim index  
481 - for (size_t i = 0; i < ga.f + 1; i++) { //check each pair of trim indices for duplications  
482 - if (trimindex[2 * i] == trimindex[2 * i + 1]) { // (duplication caused due to adjacent features)  
483 - finaltrim_ind.push_back(trimindex[2 * i]); // remove duplicated trim indices replace by one  
484 - k = k + 1;  
485 - }  
486 - }  
487 -  
488 -  
489 - ////output actual wavelenths corresponding to those trim indices  
490 - ////these wavenumber are grouped in pairs, check each pair, if duplicated numbers are there in pair delete one and keep other as band to trim, if 2nd wavnumber is smaller than 1st in a pair ignore that pair  
491 - ////e.g [1, 228, 230, 230, 232, 350,352, 351, 353, 1200] pairas [1(start)-228,230-230, 232-350, 352-351, 353-1200(end)], trimming wavenumbers are [1-228, 230, 233-350, 353-1200]  
492 - csv << (wavelengths[finaltrim_ind.at(0)]);  
493 - for (size_t i = 1; i < ga.f * + 2 ; i++)  
494 - csv << "," << (wavelengths[finaltrim_ind.at(i)]);  
495 - csv << std::endl;  
496 - } //end trim option  
497 -  
498 -  
499 - //free gpu pointers  
500 - ga.gpu_Destroy();  
501 -  
502 - //free pointers  
503 - std::free(sb);  
504 - std::free(sw);  
505 - std::free(M);  
506 - std::free(cM);  
507 - std::free(best_S);  
508 - std::free(lda);  
509 - std::free(newPop);  
510 -  
511 -}//end main  
512 -  
513 -