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
CMakeLists.txt
  Show/Hide comments   View file @34f743b
... ... @@ -2,40 +2,23 @@
2 2 cmake_minimum_required(VERSION 3.16)
3 3  
4 4 #Name your project here
5   -project(ga-gpu)
  5 +project(genetic-gpu)
6 6  
7 7 #set the module directory
8 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 12 #build the executable in the binary directory on MS Visual Studio
16 13 if ( MSVC )
17 14 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_DEBUG "${OUTPUT_DIRECTORY}")
18 15 SET( CMAKE_RUNTIME_OUTPUT_DIRECTORY_RELEASE "${OUTPUT_DIRECTORY}")
19 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 21 #find packages-----------------------------------
34   -#find OpenCV
35   -#find_package(OpenCV REQUIRED)
36   -#add_definitions(-DUSING_OPENCV)
37   -
38   -#find the pthreads package
39 22 find_package(Threads)
40 23  
41 24 #find the X11 package
... ... @@ -57,89 +40,39 @@ if( CMAKE_COMPILER_IS_GNUCC )
57 40 endif()
58 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(${GLEW_INCLUDE_DIR})
69   -#endif(WIN32)
70   -
71 43 #find LAPACK and supporting link_libraries
72 44 find_package(LAPACKE REQUIRED)
73 45  
74 46 #include include directories
75 47 include_directories(${CUDA_INCLUDE_DIRS}
76   - ${OpenCV_INCLUDE_DIRS}
77 48 ${LAPACKE_INCLUDE_DIR}
78 49 ${STIM_INCLUDE_DIRS}
79   - ${OpenGL_INCLUDE_DIRS}
80   -# ${GLUT_INCLUDE_DIR}
81   - ${FANN_INCLUDE_DIRS}
82 50 "${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 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 58 #create an executable file
97   -cuda_add_executable(ga-gpu
98   - ${GAGPU_H}
99   -# ${GA_H}
  59 +cuda_add_executable(genetic-gpu
100 60 ${GA_GPU_SRC}
101 61 )
102   -target_link_libraries(ga-gpu ${CUDA_LIBRARIES}
  62 +
  63 +target_link_libraries(genetic-gpu ${CUDA_LIBRARIES}
103 64 ${CUDA_CUBLAS_LIBRARIES}
104 65 ${CUDA_CUFFT_LIBRARIES}
105 66 ${LAPACKE_LIBRARIES}
106 67 ${LAPACK_LIBRARIES}
107 68 ${BLAS_LIBRARIES}
108   - ${CMAKE_THREAD_LIBS_INIT}
109 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 73 #if Boost is found, set an environment variable to use with preprocessor directives
133 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 76 ${Boost_SYSTEM_LIBRARY}
143 77 )
144   -# endif(BUILD_GAGPU)
145 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   -