import json import os import time import sys import getopt import re import pprint from multiprocessing import Pool, Manager, cpu_count from kneed import KneeLocator from sklearn.cluster import AgglomerativeClustering import matplotlib.pyplot as plt import ruptures as rpt import numpy as np from dfatool.functions import analytic from dfatool.loader import RawData from dfatool import parameters from dfatool.model import ParallelParamFit, PTAModel from dfatool.utils import by_name_to_by_param # from scipy.cluster.hierarchy import dendrogram, linkage # for graphical display # py bin\Proof_Of_Concept_PELT.py --filename="..\data\TX.json" --jump=1 --pen_override=28 --refinement_thresh=100 # py bin\Proof_Of_Concept_PELT.py --filename="..\data\TX.json" --jump=1 --pen_override=28 --refinement_thresh=100 --cache_dicts --cache_loc="..\data\TX2_cache" from dfatool.validation import CrossValidator def plot_data_from_json(filename, trace_num, x_axis, y_axis): with open(filename, 'r') as file: tx_data = json.load(file) print(tx_data[trace_num]['parameter']) plt.plot(tx_data[trace_num]['offline'][0]['uW']) plt.xlabel(x_axis) plt.ylabel(y_axis) plt.show() def plot_data_vs_mean(signal, x_axis, y_axis): plt.plot(signal) average = np.mean(signal) plt.hlines(average, 0, len(signal)) plt.xlabel(x_axis) plt.ylabel(y_axis) plt.show() def plot_data_vs_data_vs_means(signal1, signal2, x_axis, y_axis): plt.plot(signal1) lens = max(len(signal1), len(signal2)) average = np.mean(signal1) plt.hlines(average, 0, lens, color='red') plt.vlines(len(signal1), 0, 100000, color='red', linestyles='dashed') plt.plot(signal2) average = np.mean(signal2) plt.hlines(average, 0, lens, color='green') plt.vlines(len(signal2), 0, 100000, color='green', linestyles='dashed') plt.xlabel(x_axis) plt.ylabel(y_axis) plt.show() def get_bkps(algo, pen, q): res = pen, len(algo.predict(pen=pen)) q.put(pen) return res def find_knee_point(data_x, data_y, S=1.0, curve='convex', direction='decreasing'): kneedle = KneeLocator(data_x, data_y, S=S, curve=curve, direction=direction) kneepoint = (kneedle.knee, kneedle.knee_y) return kneepoint def calc_pelt(signal, penalty, model="l1", jump=5, min_dist=2, plotting=False): # default params in Function if model is None: model = "l1" if jump is None: jump = 5 if min_dist is None: min_dist = 2 if plotting is None: plotting = False # change point detection. best fit seemingly with l1. rbf prods. RuntimeErr for pen > 30 # https://ctruong.perso.math.cnrs.fr/ruptures-docs/build/html/costs/index.html # model = "l1" #"l1" # "l2", "rbf" algo = rpt.Pelt(model=model, jump=jump, min_size=min_dist).fit(signal) if penalty is not None: bkps = algo.predict(pen=penalty) if plotting: fig, ax = rpt.display(signal, bkps) plt.show() return bkps print_error("No Penalty specified.") sys.exit(-1) def calculate_penalty_value(signal, model="l1", jump=5, min_dist=2, range_min=0, range_max=50, num_processes=8, refresh_delay=1, refresh_thresh=5, S=1.0, pen_modifier=None, show_plots=False): # default params in Function if model is None: model = "l1" if jump is None: jump = 5 if min_dist is None: min_dist = 2 if range_min is None: range_min = 0 if range_max is None: range_max = 50 if num_processes is None: num_processes = 8 if refresh_delay is None: refresh_delay = 1 if refresh_thresh is None: refresh_thresh = 5 if S is None: S = 1.0 if pen_modifier is None: pen_modifier = 1 # change point detection. best fit seemingly with l1. rbf prods. RuntimeErr for pen > 30 # https://ctruong.perso.math.cnrs.fr/ruptures-docs/build/html/costs/index.html # model = "l1" #"l1" # "l2", "rbf" algo = rpt.Pelt(model=model, jump=jump, min_size=min_dist).fit(signal) ### CALC BKPS WITH DIFF PENALTYS if range_max != range_min: # building args array for parallelizing args = [] # for displaying progression m = Manager() q = m.Queue() for i in range(range_min, range_max + 1): args.append((algo, i, q)) print_info("starting kneepoint calculation.") # init Pool with num_proesses with Pool(min(num_processes, len(args))) as p: # collect results from pool result = p.starmap_async(get_bkps, args) # monitor loop percentage = -100 # Force display of 0% i = 0 while True: if result.ready(): break size = q.qsize() last_percentage = percentage percentage = round(size / (range_max - range_min) * 100, 2) if percentage >= last_percentage + 2 or i >= refresh_thresh: print_info("Current progress: " + str(percentage) + "%") i = 0 else: i += 1 time.sleep(refresh_delay) res = result.get() print_info("Finished kneepoint calculation.") # DECIDE WHICH PENALTY VALUE TO CHOOSE ACCORDING TO ELBOW/KNEE APPROACH # split x and y coords to pass to kneedle pen_val = [x[0] for x in res] fitted_bkps_val = [x[1] for x in res] # # plot to look at res knee = find_knee_point(pen_val, fitted_bkps_val, S=S) # TODO: Find plateau on pen_val vs fitted_bkps_val # scipy.find_peaks() does not find plateaus if they extend through the end of the data. # to counter that, add one extremely large value to the right side of the data # after negating it is extremely small -> Almost certainly smaller than the # found plateau therefore the plateau does not extend through the border # -> scipy.find_peaks finds it. Choose value from within that plateau. # fitted_bkps_val.append(100000000) # TODO: Approaching over find_peaks might not work if the initial decrease step to the # "correct" number of changepoints and additional decrease steps e.g. underfitting # take place within the given penalty interval. find_peak only finds plateaus # of peaks. If the number of chpts decreases after the wanted plateau the condition # for local peaks is not satisfied anymore. Therefore this approach will only work # if the plateau extends over the right border of the penalty interval. # peaks, peak_plateaus = find_peaks(- np.array(fitted_bkps_val), plateau_size=1) # Since the data is monotonously decreasing only one plateau can be found. # assuming the plateau is constant start_index = -1 end_index = -1 longest_start = -1 longest_end = -1 prev_val = -1 for i, num_bkpts in enumerate(fitted_bkps_val[knee[0]:]): if num_bkpts != prev_val: end_index = i - 1 if end_index - start_index > longest_end - longest_start: # currently found sequence is the longest found yet longest_start = start_index longest_end = end_index start_index = i if i == len(fitted_bkps_val[knee[0]:]) - 1: # end sequence with last value end_index = i # # since it is not guaranteed that this is the end of the plateau, assume the mid # # of the plateau was hit. # size = end_index - start_index # end_index = end_index + size # However this is not the clean solution. Better if search interval is widened if end_index - start_index > longest_end - longest_start: # last found sequence is the longest found yet longest_start = start_index longest_end = end_index start_index = i prev_val = num_bkpts if show_plots: plt.xlabel('Penalty') plt.ylabel('Number of Changepoints') plt.plot(pen_val, fitted_bkps_val) plt.vlines(longest_start + knee[0], 0, max(fitted_bkps_val), linestyles='dashed') plt.vlines(longest_end + knee[0], 0, max(fitted_bkps_val), linestyles='dashed') plt.show() # choosing pen from plateau mid_of_plat = longest_start + (longest_end - longest_start) // 2 knee = (mid_of_plat + knee[0], fitted_bkps_val[mid_of_plat + knee[0]]) # modify knee according to options. Defaults to 1 * knee knee = (knee[0] * pen_modifier, knee[1]) else: # range_min == range_max. has the same effect as pen_override knee = (range_min, None) print_info(str(knee[0]) + " has been selected as penalty.") if knee[0] is not None: return knee print_error("With the current thresh-hold S=" + str(S) + " it is not possible to select a penalty value.") sys.exit(-1) # very short benchmark yielded approx. 1/3 of speed compared to solution with sorting # def needs_refinement_no_sort(signal, mean, thresh): # # linear search for the top 10%/ bottom 10% # # should be sufficient # length_of_signal = len(signal) # percentile_size = int() # percentile_size = length_of_signal // 100 # upper_percentile = [None] * percentile_size # lower_percentile = [None] * percentile_size # fill_index_upper = percentile_size - 1 # fill_index_lower = percentile_size - 1 # index_smallest_val = fill_index_upper # index_largest_val = fill_index_lower # # for x in signal: # if x > mean: # # will be in upper percentile # if fill_index_upper >= 0: # upper_percentile[fill_index_upper] = x # if x < upper_percentile[index_smallest_val]: # index_smallest_val = fill_index_upper # fill_index_upper = fill_index_upper - 1 # continue # # if x > upper_percentile[index_smallest_val]: # # replace smallest val. Find next smallest val # upper_percentile[index_smallest_val] = x # index_smallest_val = 0 # i = 0 # for y in upper_percentile: # if upper_percentile[i] < upper_percentile[index_smallest_val]: # index_smallest_val = i # i = i + 1 # # else: # if fill_index_lower >= 0: # lower_percentile[fill_index_lower] = x # if x > lower_percentile[index_largest_val]: # index_largest_val = fill_index_upper # fill_index_lower = fill_index_lower - 1 # continue # if x < lower_percentile[index_largest_val]: # # replace smallest val. Find next smallest val # lower_percentile[index_largest_val] = x # index_largest_val = 0 # i = 0 # for y in lower_percentile: # if lower_percentile[i] > lower_percentile[index_largest_val]: # index_largest_val = i # i = i + 1 # # # should have the percentiles # lower_percentile_mean = np.mean(lower_percentile) # upper_percentile_mean = np.mean(upper_percentile) # dist = mean - lower_percentile_mean # if dist > thresh: # return True # dist = upper_percentile_mean - mean # if dist > thresh: # return True # return False # raw_states_calc_args.append((num_measurement, measurement, penalty, opt_model # , opt_jump)) def calc_raw_states_func(num_measurement, measurement, penalty, model, jump): signal = np.array(measurement['uW']) normed_signal = norm_signal(signal) bkpts = calc_pelt(normed_signal, penalty, model=model, jump=jump) calced_states = list() start_time = 0 end_time = 0 for bkpt in bkpts: # start_time of state is end_time of previous one # (Transitions are instantaneous) start_time = end_time end_time = bkpt power_vals = signal[start_time: end_time] mean_power = np.mean(power_vals) std_dev = np.std(power_vals) calced_state = (start_time, end_time, mean_power, std_dev) calced_states.append(calced_state) num = 0 new_avg_std = 0 for s in calced_states: # print_info("State " + str(num) + " starts at t=" + str(s[0]) # + " and ends at t=" + str(s[1]) # + " while using " + str(s[2]) # + "uW with sigma=" + str(s[3])) num = num + 1 new_avg_std = new_avg_std + s[3] new_avg_std = new_avg_std / len(calced_states) change_avg_std = measurement['uW_std'] - new_avg_std # print_info("The average standard deviation for the newly found states is " # + str(new_avg_std)) # print_info("That is a reduction of " + str(change_avg_std)) return num_measurement, calced_states, new_avg_std, change_avg_std def calc_raw_states(arg_list, num_processes=8): m = Manager() with Pool(processes=min(num_processes, len(arg_list))) as p: # collect results from pool result = p.starmap(calc_raw_states_func, arg_list) return result # Very short benchmark yielded approx. 3 times the speed of solution not using sort def needs_refinement(signal, thresh): sorted_signal = sorted(signal) length_of_signal = len(signal) percentile_size = int() percentile_size = length_of_signal // 100 lower_percentile = sorted_signal[0:percentile_size] upper_percentile = sorted_signal[length_of_signal - percentile_size: length_of_signal] lower_percentile_mean = np.mean(lower_percentile) upper_percentile_mean = np.mean(upper_percentile) median = np.median(sorted_signal) dist = median - lower_percentile_mean if dist > thresh: return True dist = upper_percentile_mean - median if dist > thresh: return True return False def print_info(str_to_prt): str_lst = str_to_prt.split(sep='\n') for str_prt in str_lst: print("[INFO]" + str_prt) def print_warning(str_to_prt): str_lst = str_to_prt.split(sep='\n') for str_prt in str_lst: print("[WARNING]" + str_prt) def print_error(str_to_prt): str_lst = str_to_prt.split(sep='\n') for str_prt in str_lst: print("[ERROR]" + str_prt, file=sys.stderr) def norm_signal(signal, scaler=25): # TODO: maybe refine normalisation of signal max_val = max(signal) normed_signal = np.zeros(shape=len(signal)) for i, signal_i in enumerate(signal): normed_signal[i] = signal_i / max_val normed_signal[i] = normed_signal[i] * scaler # plt.plot(normed_signal) # plt.show() return normed_signal def norm_values_to_cluster(values_to_cluster): new_vals = np.array(values_to_cluster) num_samples = len(values_to_cluster) num_params = len(values_to_cluster[0]) for i in range(num_params): param_vals = [] for sample in new_vals: param_vals.append(sample[i]) max_val = np.max(np.abs(param_vals)) for num_sample, sample in enumerate(new_vals): values_to_cluster[num_sample][i] = sample[i] / max_val return new_vals def get_state_num(state_name, distinct_states): for state_num, states in enumerate(distinct_states): if state_name in states: return state_num return -1 if __name__ == '__main__': # OPTION RECOGNITION opt = dict() optspec = ( "filename= " "v " "model= " "jump= " "min_dist= " "range_min= " "range_max= " "num_processes= " "refresh_delay= " "refresh_thresh= " "S= " "pen_override= " "pen_modifier= " "plotting= " "refinement_thresh= " "cache_dicts " "cache_loc= " ) opt_filename = None opt_verbose = False opt_model = None opt_jump = None opt_min_dist = None opt_range_min = None opt_range_max = None opt_num_processes = cpu_count() opt_refresh_delay = None opt_refresh_thresh = None opt_S = None opt_pen_override = None opt_pen_modifier = None opt_plotting = False opt_refinement_thresh = None opt_cache_loc = None try: raw_opts, args = getopt.getopt(sys.argv[1:], "", optspec.split(" ")) for option, parameter in raw_opts: optname = re.sub(r"^--", "", option) opt[optname] = parameter if 'filename' not in opt: print_error("No file specified!") sys.exit(-1) else: opt_filename = opt['filename'] if 'v' in opt: opt_verbose = True opt_plotting = True if 'model' in opt: opt_model = opt['model'] if 'jump' in opt: try: opt_jump = int(opt['jump']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'min_dist' in opt: try: opt_min_dist = int(opt['min_dist']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'range_min' in opt: try: opt_range_min = int(opt['range_min']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'range_max' in opt: try: opt_range_max = int(opt['range_max']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'num_processes' in opt: try: opt_num_processes = int(opt['num_processes']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'refresh_delay' in opt: try: opt_refresh_delay = int(opt['refresh_delay']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'refresh_thresh' in opt: try: opt_refresh_thresh = int(opt['refresh_thresh']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'S' in opt: try: opt_S = float(opt['S']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'pen_override' in opt: try: opt_pen_override = int(opt['pen_override']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'pen_modifier' in opt: try: opt_pen_modifier = float(opt['pen_modifier']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'refinement_thresh' in opt: try: opt_refinement_thresh = int(opt['refinement_thresh']) except ValueError as verr: print(verr, file=sys.stderr) sys.exit(-1) if 'cache_dicts' in opt: if 'cache_loc' in opt: opt_cache_loc = opt['cache_loc'] else: print_error("If \"cache_dicts\" is set, \"cache_loc\" must be provided.") sys.exit(-1) except getopt.GetoptError as err: print(err, file=sys.stderr) sys.exit(-1) # OPENING DATA if ".json" in opt_filename: # open file with trace data from json print_info( "Will only refine the state which is present in " + opt_filename + " if necessary.") with open(opt_filename, 'r') as f: configurations = json.load(f) # for i in range(0, 7): # signal = np.array(configurations[i]['offline'][0]['uW']) # plt.plot(signal) # plt.xlabel('Time [us]') # plt.ylabel('Power [mW]') # plt.show() # sys.exit() # loop through all traces check if refinement is necessary # resulting_sequence_list = [] # search for param_names, by_param and by_name files by_param_file = None by_name_file = None param_names_file = None from_cache = False not_accurate = False if opt_cache_loc is not None: flag = False by_name_loc = os.path.join(opt_cache_loc, "by_name.txt") by_param_loc = os.path.join(opt_cache_loc, "by_param.txt") param_names_loc = os.path.join(opt_cache_loc, "param_names.txt") if os.path.isfile(by_name_loc) and os.path.getsize(by_name_loc) > 0: by_name_file = open(by_name_loc, "r") else: print_error("In " + opt_cache_loc + " is no by_name.txt.") flag = True if os.path.isfile(by_param_loc) and os.path.getsize(by_param_loc) > 0: by_param_file = open(by_param_loc, "r") else: print_error("In " + opt_cache_loc + " is no by_param.txt.") flag = True if os.path.isfile(param_names_loc) and os.path.getsize(param_names_loc) > 0: param_names_file = open(param_names_loc, "r") else: print_error("In " + opt_cache_loc + " is no param_names.txt.") flag = True if flag: print_info("The cache will be build.") else: print_warning("THE OPTION \"cache_dicts\" IS FOR DEBUGGING PURPOSES ONLY! " "\nDO NOT USE FOR REGULAR APPLICATIONS!" "\nThe script will not run to the end properly." "\nNo final parametrization will be done.") from_cache = True if None in (by_param_file, by_name_file, param_names_file): state_durations_by_config = [] state_consumptions_by_config = [] for num_config, measurements_by_config in enumerate(configurations): # loop through all occurrences of the looked at state print_info("Looking at state '" + measurements_by_config['name'] + "' with params: " + str(measurements_by_config['parameter']) + "(" + str( num_config + 1) + "/" + str(len(configurations)) + ")") refine = False print_info("Checking if refinement is necessary...") for measurement in measurements_by_config['offline']: # loop through measurements of particular state # an check if state needs refinement signal = measurement['uW'] # mean = measurement['uW_mean'] if needs_refinement(signal, opt_refinement_thresh) and not refine: print_info("Refinement is necessary!") refine = True if not refine: print_info( "No refinement necessary for state '" + measurements_by_config['name'] + "' with params: " + str(measurements_by_config['parameter'])) else: # assume that all measurements of the same param configuration are fundamentally # similar -> calculate penalty for first measurement, use it for all if opt_pen_override is None: signal = np.array(measurements_by_config['offline'][0]['uW']) normed_signal = norm_signal(signal) penalty = calculate_penalty_value(normed_signal, model=opt_model, range_min=opt_range_min, range_max=opt_range_max, num_processes=opt_num_processes, jump=opt_jump, S=opt_S, pen_modifier=opt_pen_modifier) penalty = penalty[0] else: penalty = opt_pen_override # build arguments for parallel excecution print_info("Starting raw_states calculation.") raw_states_calc_args = [] for num_measurement, measurement in enumerate( measurements_by_config['offline']): raw_states_calc_args.append((num_measurement, measurement, penalty, opt_model, opt_jump)) raw_states_list = [None] * len(measurements_by_config['offline']) raw_states_res = calc_raw_states(raw_states_calc_args, opt_num_processes) # extracting result and putting it in correct order -> index of raw_states_list # entry still corresponds with index of measurement in measurements_by_states # -> If measurements are discarded the correct ones are easily recognized for ret_val in raw_states_res: num_measurement = ret_val[0] raw_states = ret_val[1] avg_std = ret_val[2] change_avg_std = ret_val[3] # TODO: Wieso gibt mir meine IDE hier eine Warning aus? Der Index müsste doch # int sein oder nicht? Es scheint auch vernünftig zu klappen... raw_states_list[num_measurement] = raw_states print_info("The average standard deviation for the newly found states in " + "measurement No. " + str(num_measurement) + " is " + str( avg_std)) print_info("That is a reduction of " + str(change_avg_std)) # l_signal = measurements_by_config['offline'][num_measurement]['uW'] # l_bkpts = [s[1] for s in raw_states] # fig, ax = rpt.display(np.array(l_signal), l_bkpts) # plt.show() print_info("Finished raw_states calculation.") num_states_array = [int()] * len(raw_states_list) i = 0 for i, x in enumerate(raw_states_list): num_states_array[i] = len(x) avg_num_states = np.mean(num_states_array) num_states_dev = np.std(num_states_array) print_info("On average " + str(avg_num_states) + " States have been found. The standard deviation" + " is " + str(num_states_dev)) # TODO: MAGIC NUMBER if num_states_dev > 1: print_warning("The number of states varies strongly across measurements." " Consider choosing a larger value for S or using the " "pen_modifier option.") time.sleep(5) # TODO: Wie bekomme ich da jetzt raus, was die Wahrheit ist? # Einfach Durchschnitt nehmen? # Preliminary decision: Further on only use the traces, which have the most # frequent state count counts = np.bincount(num_states_array) num_raw_states = np.argmax(counts) print_info("Choose " + str(num_raw_states) + " as number of raw_states.") # iterate through all found breakpoints and determine start and end points as well # as power consumption num_measurements = len(raw_states_list) states_duration_list = [list()] * num_raw_states states_consumption_list = [list()] * num_raw_states for num_elem, _ in enumerate(states_duration_list): states_duration_list[num_elem] = [0] * num_measurements states_consumption_list[num_elem] = [0] * num_measurements num_used_measurements = 0 for num_measurement, raw_states in enumerate(raw_states_list): if len(raw_states) == num_raw_states: num_used_measurements = num_used_measurements + 1 for num_state, s in enumerate(raw_states): states_duration_list[num_state][num_measurement] = s[1] - s[0] states_consumption_list[num_state][num_measurement] = s[2] # calced_state = (start_time, end_time, mean_power, std_dev) # for num_state, s in enumerate(raw_states): # state_duration = s[1] - s[0] # state_consumption = s[2] # states_duration_list[num_state] = \ # states_duration_list[num_state] + state_duration # states_consumption_list[num_state] = \ # states_consumption_list[num_state] + state_consumption else: print_info("Discarding measurement No. " + str(num_measurement) + " because it did not recognize the number of " "raw_states correctly.") # l_signal = measurements_by_config['offline'][num_measurement]['uW'] # l_bkpts = [s[1] for s in raw_states] # fig, ax = rpt.display(np.array(l_signal), l_bkpts) # plt.show() # for i, x in enumerate(states_duration_list): # states_duration_list[i] = x / num_used_measurements # for i, x in enumerate(states_consumption_list): # states_consumption_list[i] = x / num_used_measurements if num_used_measurements != len(raw_states_list): if num_used_measurements / len(raw_states_list) <= 0.5: print_warning("Only used " + str(num_used_measurements) + "/" + str( len(raw_states_list)) + " Measurements for refinement. " + "Others did not recognize number of states correctly." + "\nYou should verify the integrity of the measurements.") else: print_info("Used " + str(num_used_measurements) + "/" + str(len(raw_states_list)) + " Measurements for refinement." + " Others did not recognize number of states correctly.") num_used_measurements = i # TODO: DEBUG Kram #sys.exit(0) else: print_info("Used all available measurements.") state_durations_by_config.append((num_config, states_duration_list)) state_consumptions_by_config.append((num_config, states_consumption_list)) # # TODO: # if num_config == 6: # print("BRECHE AUS") # break # combine all state durations and consumptions to parametrized model if len(state_durations_by_config) == 0: print("No refinement necessary for this state. The macromodel is usable.") sys.exit() if len(state_durations_by_config) / len(configurations) > 1 / 2 \ and len(state_durations_by_config) != len(configurations): print_warning( "Some measurements(>50%) need to be refined, however that is not true for" " all measurements. This hints a correlation between the structure of" " the underlying automaton and parameters. Only the ones which need to" " be refined will be refined. THE RESULT WILL NOT ACCURATELY DEPICT " " THE REAL WORLD.") not_accurate = True if len(state_durations_by_config) / len(configurations) < 1 / 2: print_warning( "Some measurements(<50%) need to be refined, however that is not true for" " all measurements. This hints a correlation between the structure of" " the underlying automaton and parameters. Or a poor quality of measurements." " No Refinement will be done.") sys.exit(-1) # this is only necessary because at this state only linear automatons can be modeled. num_states_array = [int()] * len(state_consumptions_by_config) for i, (_, states_consumption_list) in enumerate(state_consumptions_by_config): num_states_array[i] = len(states_consumption_list) counts = np.bincount(num_states_array) num_raw_states = np.argmax(counts) usable_configs = len(state_consumptions_by_config) # param_list identical for each raw_state param_list = [] param_names = configurations[0]['offline_aggregates']['paramkeys'][0] print_info("param_names: " + str(param_names)) for num_config, states_consumption_list in state_consumptions_by_config: if len(states_consumption_list) != num_raw_states: print_warning( "Config No." + str(num_config) + " not usable yet due to different " + "number of states. This hints a correlation between parameters and " + "the structure of the resulting automaton. This will be possibly" + " supported in a future version of this tool. HOWEVER AT THE MOMENT" " THIS WILL LEAD TO INACCURATE RESULTS!") not_accurate = True usable_configs = usable_configs - 1 else: param_list.extend(configurations[num_config]['offline_aggregates']['param']) print_info("param_list: " + str(param_list)) if usable_configs == len(state_consumptions_by_config): print_info("All configs usable.") else: print_info("Using only " + str(usable_configs) + " Configs.") if num_raw_states == 1: print_info("Upon further inspection it is clear that no refinement is necessary." " The macromodel is usable.") sys.exit(-1) by_name = {} usable_configs_2 = len(state_consumptions_by_config) for i in range(num_raw_states): consumptions_for_state = [] durations_for_state = [] for j, (_, states_consumption_list) in enumerate(state_consumptions_by_config): if len(states_consumption_list) == num_raw_states: consumptions_for_state.extend(states_consumption_list[i]) durations_for_state.extend(state_durations_by_config[j][1][i]) else: not_accurate = True usable_configs_2 = usable_configs_2 - 1 if usable_configs_2 != usable_configs: print_error("an zwei unterschiedlichen Stellen wurden unterschiedlich viele " "Messungen rausgeworfen. Bei Janis beschweren.") state_name = "state_" + str(i) state_dict = { "param": param_list, "power": consumptions_for_state, "duration": durations_for_state, "attributes": ["power", "duration"], # Da kein richtiger Automat generiert wird, gibt es auch keine Transitionen "isa": "state" } by_name[state_name] = state_dict by_param = by_name_to_by_param(by_name) if opt_cache_loc is not None: by_name_loc = os.path.join(opt_cache_loc, "by_name.txt") by_param_loc = os.path.join(opt_cache_loc, "by_param.txt") param_names_loc = os.path.join(opt_cache_loc, "param_names.txt") f = open(by_name_loc, "w") f.write(str(by_name)) f.close() f = open(by_param_loc, "w") f.write(str(by_param)) f.close() f = open(param_names_loc, "w") f.write(str(param_names)) f.close() else: by_name_text = str(by_name_file.read()) by_name = eval(by_name_text) by_param_text = str(by_param_file.read()) by_param = eval(by_param_text) param_names_text = str(param_names_file.read()) param_names = eval(param_names_text) # t = 0 # last_pow = 0 # for key in by_name.keys(): # end_t = t + np.mean(by_name[key]["duration"]) # power = np.mean(by_name[key]["power"]) # plt.vlines(t, min(last_pow, power), max(last_pow, power)) # plt.hlines(power, t, end_t) # t = end_t # last_pow = power # plt.show() stats = parameters.ParamStats(by_name, by_param, param_names, dict()) paramfit = ParallelParamFit(by_param) for state_name in by_name.keys(): for num_param, param_name in enumerate(param_names): if stats.depends_on_param(state_name, "power", param_name): paramfit.enqueue(state_name, "power", num_param, param_name) if stats.depends_on_param(state_name, "duration", param_name): paramfit.enqueue(state_name, "duration", num_param, param_name) print_info("State " + state_name + "s power depends on param " + param_name + ":" + str(stats.depends_on_param(state_name, "power", param_name)) ) print_info("State " + state_name + "s duration depends on param " + param_name + ":" + str(stats.depends_on_param(state_name, "duration", param_name)) ) paramfit.fit() fit_res_dur_dict = {} fit_res_pow_dict = {} for state_name in by_name.keys(): fit_power = paramfit.get_result(state_name, "power") fit_duration = paramfit.get_result(state_name, "duration") combined_fit_power = analytic.function_powerset(fit_power, param_names, 0) combined_fit_duration = analytic.function_powerset(fit_duration, param_names, 0) combined_fit_power.fit(by_param, state_name, "power") if not combined_fit_power.fit_success: print_warning("Fitting(power) for state " + state_name + " was not succesful!") combined_fit_duration.fit(by_param, state_name, "duration") if not combined_fit_duration.fit_success: print_warning("Fitting(duration) for state " + state_name + " was not succesful!") fit_res_pow_dict[state_name] = combined_fit_power fit_res_dur_dict[state_name] = combined_fit_duration # only raw_states with the same number of function parameters can be similar num_param_pow_dict = {} num_param_dur_dict = {} for state_name in by_name.keys(): model_function = str(fit_res_pow_dict[state_name].model_function) model_args = fit_res_pow_dict[state_name].model_args num_param_pow_dict[state_name] = len(model_args) for num_arg, arg in enumerate(model_args): replace_string = "regression_arg(" + str(num_arg) + ")" model_function = model_function.replace(replace_string, str(arg)) print_info("Power-Function for state " + state_name + ": " + model_function) for state_name in by_name.keys(): model_function = str(fit_res_dur_dict[state_name].model_function) model_args = fit_res_dur_dict[state_name].model_args num_param_dur_dict[state_name] = len(model_args) for num_arg, arg in enumerate(model_args): replace_string = "regression_arg(" + str(num_arg) + ")" model_function = model_function.replace(replace_string, str(arg)) print_info("Duration-Function for state " + state_name + ": " + model_function) similar_raw_state_buckets = {} for state_name in by_name.keys(): pow_model_function = str(fit_res_pow_dict[state_name].model_function) dur_model_function = str(fit_res_dur_dict[state_name].model_function) key_tuple = (pow_model_function, dur_model_function) if key_tuple not in similar_raw_state_buckets: similar_raw_state_buckets[key_tuple] = [] similar_raw_state_buckets[key_tuple].append(state_name) # cluster for each Key-Tuple using the function parameters distinct_states = [] for key_tuple in similar_raw_state_buckets.keys(): print_info("Key-Tuple " + str(key_tuple) + ": " + str(similar_raw_state_buckets[key_tuple])) similar_states = similar_raw_state_buckets[key_tuple] if len(similar_states) > 1: # functions are identical -> num_params is identical num_params = num_param_dur_dict[similar_states[0]] + num_param_pow_dict[ similar_states[0]] values_to_cluster = np.zeros((len(similar_states), num_params)) for num_state, state_name in enumerate(similar_states): dur_params = fit_res_dur_dict[state_name].model_args pow_params = fit_res_pow_dict[state_name].model_args j = 0 for param in pow_params: values_to_cluster[num_state][j] = param j = j + 1 for param in dur_params: values_to_cluster[num_state][j] = param j = j + 1 normed_vals_to_cluster = norm_values_to_cluster(values_to_cluster) cluster = AgglomerativeClustering(n_clusters=None, compute_full_tree=True, affinity='euclidean', linkage='ward', # TODO: Magic Number. Beim Evaluieren finetunen distance_threshold=1) cluster.fit_predict(values_to_cluster) cluster_labels = cluster.labels_ print_info("Cluster labels:\n" + str(cluster_labels)) if cluster.n_clusters_ > 1: # more than one distinct state found distinct_state_dict = {} for num_state, label in enumerate(cluster_labels): if label not in distinct_state_dict.keys(): distinct_state_dict[label] = [] distinct_state_dict[label].append(similar_states[num_state]) for distinct_state_key in distinct_state_dict.keys(): distinct_states.append(distinct_state_dict[distinct_state_key]) else: distinct_states.append(similar_states) else: distinct_states.append(similar_states) for num_state, distinct_state in enumerate(distinct_states): print("State " + str(num_state) + ": " + str(distinct_state)) num_raw_states = len(by_name.keys()) resulting_sequence = [int] * num_raw_states for i in range(num_raw_states): state_name = "state_" + str(i) state_num = get_state_num(state_name, distinct_states) if state_num == -1: print_error("Critical Error when creating the resulting sequence. raw_state state_" + str(i) + " could not be mapped to a state.") sys.exit(-1) resulting_sequence[i] = state_num print("Resulting sequence is: " + str(resulting_sequence)) # if from_cache: # print_warning( # "YOU USED THE OPTION \"cache_dicts\". THIS IS FOR DEBUGGING PURPOSES ONLY!" # "\nTHE SCRIPT WILL NOW STOP PREMATURELY," # "SINCE DATA FOR FURTHER COMPUTATION IS MISSING!") # sys.exit(0) new_by_name = {} for num_state, distinct_state in enumerate(distinct_states): state_name = "State_" + str(num_state) consumptions_for_state = [] durations_for_state = [] param_list = [] for raw_state in distinct_state: original_state_dict = by_name[raw_state] param_list.extend(original_state_dict["param"]) consumptions_for_state.extend(original_state_dict["power"]) durations_for_state.extend(original_state_dict["duration"]) new_state_dict = { "param": param_list, "power": consumptions_for_state, "duration": durations_for_state, "attributes": ["power", "duration"], # Da kein richtiger Automat generiert wird, gibt es auch keine Transitionen "isa": "state" } new_by_name[state_name] = new_state_dict new_by_param = by_name_to_by_param(new_by_name) new_stats = parameters.ParamStats(new_by_name, new_by_param, param_names, dict()) new_paramfit = ParallelParamFit(new_by_param) for state_name in new_by_name.keys(): for num_param, param_name in enumerate(param_names): if new_stats.depends_on_param(state_name, "power", param_name): new_paramfit.enqueue(state_name, "power", num_param, param_name) if new_stats.depends_on_param(state_name, "duration", param_name): new_paramfit.enqueue(state_name, "duration", num_param, param_name) print_info("State " + state_name + "s power depends on param " + param_name + ":" + str(new_stats.depends_on_param(state_name, "power", param_name)) ) print_info("State " + state_name + "s duration depends on param " + param_name + ":" + str(new_stats.depends_on_param(state_name, "duration", param_name)) ) new_paramfit.fit() new_fit_res_dur_dict = {} new_fit_res_pow_dict = {} for state_name in new_by_name.keys(): fit_power = new_paramfit.get_result(state_name, "power") fit_duration = new_paramfit.get_result(state_name, "duration") combined_fit_power = analytic.function_powerset(fit_power, param_names, 0) combined_fit_duration = analytic.function_powerset(fit_duration, param_names, 0) combined_fit_power.fit(new_by_param, state_name, "power") if not combined_fit_power.fit_success: print_warning("Fitting(power) for state " + state_name + " was not succesful!") combined_fit_duration.fit(new_by_param, state_name, "duration") if not combined_fit_duration.fit_success: print_warning("Fitting(duration) for state " + state_name + " was not succesful!") new_fit_res_pow_dict[state_name] = combined_fit_power new_fit_res_dur_dict[state_name] = combined_fit_duration for state_name in new_by_name.keys(): model_function = str(new_fit_res_pow_dict[state_name].model_function) model_args = new_fit_res_pow_dict[state_name].model_args for num_arg, arg in enumerate(model_args): replace_string = "regression_arg(" + str(num_arg) + ")" model_function = model_function.replace(replace_string, str(arg)) print("Power-Function for state " + state_name + ": " + model_function) for state_name in new_by_name.keys(): model_function = str(new_fit_res_dur_dict[state_name].model_function) model_args = new_fit_res_dur_dict[state_name].model_args for num_arg, arg in enumerate(model_args): replace_string = "regression_arg(" + str(num_arg) + ")" model_function = model_function.replace(replace_string, str(arg)) print("Duration-Function for state " + state_name + ": " + model_function) # model = PTAModel(new_by_name, param_names, dict()) # model_json = model.to_json() # param_model, _ = model.get_fitted() # param_quality = model.assess(param_model) # pprint.pprint(param_quality) # # model = PTAModel(by_name, ...) # # validator = CrossValidator(PTAModel, by_name, ...) # # param_quality = validator.kfold(lambda m: m.get_fitted()[0], 10) # validator = CrossValidator(PTAModel, new_by_name, param_names, dict()) # param_quality = validator.kfold(lambda m: m.get_fitted()[0], 10) # pprint.pprint(param_quality) if not_accurate: print_warning( "THIS RESULT IS NOT ACCURATE. SEE WARNINGLOG TO GET A BETTER UNDERSTANDING" " WHY.") # TODO: removed clustering (temporarily), since it provided too much dificultys # at the current state # i = 0 # cluster_labels_list = [] # num_cluster_list = [] # for num_trace, raw_states in enumerate(raw_states_list): # # iterate through raw states from measurements # if len(raw_states) == num_raw_states: # # build array with power values to cluster these # value_to_cluster = np.zeros((num_raw_states, 2)) # j = 0 # for s in raw_states: # value_to_cluster[j][0] = s[2] # value_to_cluster[j][1] = 0 # j = j + 1 # # linked = linkage(value_to_cluster, 'single') # # # # labelList = range(1, 11) # # # # plt.figure(figsize=(10, 7)) # # dendrogram(linked, # # orientation='top', # # distance_sort='descending', # # show_leaf_counts=True) # # plt.show() # # TODO: Automatic detection of number of clusters. Aktuell noch MAGIC NUMBER # # im distance_threshold # cluster = AgglomerativeClustering(n_clusters=None, compute_full_tree=True, # affinity='euclidean', # linkage='ward', # distance_threshold=opt_refinement_thresh * 100) # # cluster = AgglomerativeClustering(n_clusters=5, affinity='euclidean', # # linkage='ward') # cluster.fit_predict(value_to_cluster) # # print_info("Cluster labels:\n" + str(cluster.labels_)) # # plt.scatter(value_to_cluster[:, 0], value_to_cluster[:, 1], c=cluster.labels_, cmap='rainbow') # # plt.show() # cluster_labels_list.append((num_trace, cluster.labels_)) # num_cluster_list.append((num_trace, cluster.n_clusters_)) # i = i + 1 # else: # print_info("Discarding measurement No. " + str(num_trace) + " because it " # + "did not recognize the number of raw_states correctly.") # num_used_measurements = len(raw_states_list) # if i != len(raw_states_list): # if i / len(raw_states_list) <= 0.5: # print_warning("Only used " + str(i) + "/" + str(len(raw_states_list)) # + " Measurements for refinement. " # "Others did not recognize number of states correctly." # "\nYou should verify the integrity of the measurements.") # else: # print_info("Used " + str(i) + "/" + str(len(raw_states_list)) # + " Measurements for refinement. " # "Others did not recognize number of states correctly.") # num_used_measurements = i # # TODO: DEBUG Kram # sys.exit(0) # else: # print_info("Used all available measurements.") # # num_states = np.argmax(np.bincount([elem[1] for elem in num_cluster_list])) # avg_per_state_list = [None] * len(cluster_labels_list) # used_clusters = 0 # for number, (num_trace, labels) in enumerate(cluster_labels_list): # if num_cluster_list[number][1] == num_states: # avg_per_state = [0] * num_states # count_per_state = [0] * num_states # raw_states = raw_states_list[num_trace] # for num_label, label in enumerate(labels): # count_per_state[label] = count_per_state[label] + 1 # avg_per_state[label] = avg_per_state[label] + raw_states[num_label][2] # for i, _ in enumerate(avg_per_state): # avg_per_state[i] = avg_per_state[i] / count_per_state[i] # avg_per_state_list[number] = avg_per_state # used_clusters = used_clusters + 1 # else: # # hopefully this does not happen regularly # print_info("Discarding measurement " + str(number) # + " because the clustering yielded not matching results.") # num_used_measurements = num_used_measurements - 1 # if num_used_measurements == 0: # print_error("Something went terribly wrong. Discarded all measurements.") # # continue # sys.exit(-1) # # flattend version for clustering: # values_to_cluster = np.zeros((num_states * used_clusters, 2)) # index = 0 # for avg_per_state in avg_per_state_list: # if avg_per_state is not None: # for avg in avg_per_state: # values_to_cluster[index][0] = avg # values_to_cluster[index][1] = 0 # index = index + 1 # # plt.scatter(values_to_cluster[:, 0], values_to_cluster[:, 1]) # # plt.show() # cluster = AgglomerativeClustering(n_clusters=num_states) # cluster.fit_predict(values_to_cluster) # # Aktuell hast du hier ein plattes Array mit labels. Jetzt also das wieder auf die # # ursprünglichen Labels abbilden, die dann verändern mit den hier gefundenen Labels. # # Alle identischen Zustände haben identische Labels. Dann vllt bei resulting # # sequence ausgeben, wie groß die übereinstimmung bei der Stateabfolge ist. # new_labels_list = [] # new_labels = [] # i = 0 # for label in cluster.labels_: # new_labels.append(label) # i = i + 1 # if i == num_states: # new_labels_list.append(new_labels) # new_labels = [] # i = 0 # # only the selected measurements are present in new_labels. # # new_labels_index should not be incremented, if not selected_measurement is skipped # new_labels_index = 0 # # cluster_labels_list contains all measurements -> if measurement is skipped # # still increment the index # index = 0 # for elem in avg_per_state_list: # if elem is not None: # for number, label in enumerate(cluster_labels_list[index][1]): # cluster_labels_list[index][1][number] = \ # new_labels_list[new_labels_index][label] # new_labels_index = new_labels_index + 1 # else: # # override not selected measurement labels to avoid choosing the wrong ones. # for number, label in enumerate(cluster_labels_list[index][1]): # cluster_labels_list[index][1][number] = -1 # index = index + 1 # resulting_sequence = [None] * num_raw_states # i = 0 # confidence = 0 # for x in resulting_sequence: # j = 0 # test_list = [] # for arr in [elem[1] for elem in cluster_labels_list]: # if num_cluster_list[j][1] != num_states: # j = j + 1 # else: # if -1 in arr: # print_error("Bei Janis beschweren! Fehler beim Umbenennen der" # " Zustände wahrscheinlich.") # sys.exit(-1) # test_list.append(arr[i]) # j = j + 1 # bincount = np.bincount(test_list) # resulting_sequence[i] = np.argmax(bincount) # confidence = confidence + bincount[resulting_sequence[i]] / np.sum(bincount) # i = i + 1 # confidence = confidence / len(resulting_sequence) # print_info("Confidence of resulting sequence is " + str(confidence) # + " while using " + str(num_used_measurements) + "/" # + str(len(raw_states_list)) + " measurements.") # #print(resulting_sequence) # resulting_sequence_list.append((num_config, resulting_sequence)) # # TODO: Was jetzt? Hier habe ich jetzt pro Konfiguration eine Zustandsfolge. Daraus Automat # # erzeugen. Aber wie? Oder erst parametrisieren? Eigentlich brauche ich vorher die # # Loops. Wie erkenne ich die? Es können beliebig viele Loops an beliebigen Stellen # # auftreten. # # TODO: Die Zustandsfolgen werden sich nicht einfach in isomorphe(-einzelne wegfallende bzw. # # hinzukommende Zustände) Automaten übersetzten lassen. Basiert alles auf dem Problem: # # wie erkenne ich, dass zwei Zustände die selben sind und nicht nur einfach eine ähnliche # # Leistungsaufnahme haben?! Vllt Zustände 2D clustern? 1Dim = Leistungsaufnahme, # # 2Dim=Dauer? Zumindest innerhalb einer Paramkonfiguration sollte sich die Dauer eines # # Zustands ja nicht mehr ändern. Kann sicherlich immernoch Falschclustering erzeugen... # for num_config, sequence in resulting_sequence_list: # print_info("NO. config:" + str(num_config)) # print_info(sequence) # # # # elif ".tar" in opt_filename: # open with dfatool raw_data_args = list() raw_data_args.append(opt_filename) raw_data = RawData( raw_data_args, with_traces=True ) print_info("Preprocessing file. Depending on its size, this could take a while.") preprocessed_data = raw_data.get_preprocessed_data() print_info("File fully preprocessed") # TODO: Mal schauen, wie ich das mache. Erstmal nur mit json print_error("Not implemented yet. Please generate .json files first with dfatool and use" " those.") else: print_error("Unknown dataformat") sys.exit(-1) # print(tx_data[1]['parameter']) # # parse json to array for PELT # signal = np.array(tx_data[1]['offline'][0]['uW']) # # for i in range(0, len(signal)): # signal[i] = signal[i]/1000 # bkps = calc_pelt(signal, model=opt_model, range_max=opt_range_max, num_processes=opt_num_processes, jump=opt_jump, S=opt_S) # fig, ax = rpt.display(signal, bkps) # plt.xlabel('Time [us]') # plt.ylabel('Power [mW]') # plt.show()