| 1 | """ |
|---|
| 2 | Functions used to calculate the root mean square deviation. |
|---|
| 3 | |
|---|
| 4 | Duncan Gray, GA - 2007 |
|---|
| 5 | |
|---|
| 6 | """ |
|---|
| 7 | |
|---|
| 8 | |
|---|
| 9 | #---------------------------------------------------------------------------- |
|---|
| 10 | # Import necessary modules |
|---|
| 11 | #---------------------------------------------------------------------------- |
|---|
| 12 | |
|---|
| 13 | # Standard modules |
|---|
| 14 | import os |
|---|
| 15 | from csv import writer |
|---|
| 16 | from time import localtime, strftime |
|---|
| 17 | from os.path import join |
|---|
| 18 | |
|---|
| 19 | # Related major packages |
|---|
| 20 | from Numeric import zeros, Float, where, greater, less, compress, sqrt, sum |
|---|
| 21 | from anuga.shallow_water.data_manager import csv2dict |
|---|
| 22 | from anuga.utilities.numerical_tools import ensure_numeric, err, norm |
|---|
| 23 | from anuga.utilities.interp import interp |
|---|
| 24 | |
|---|
| 25 | # Scenario specific imports |
|---|
| 26 | import project |
|---|
| 27 | |
|---|
| 28 | def get_max_min_condition_array(min, max, vector): |
|---|
| 29 | """ |
|---|
| 30 | Given a vector of values, and minimum and maximum values, return a |
|---|
| 31 | vector of 0/1's that can be used to cut arrays so only the times |
|---|
| 32 | in the min max range are used. |
|---|
| 33 | |
|---|
| 34 | precondition: The vector values are ascending. |
|---|
| 35 | |
|---|
| 36 | """ |
|---|
| 37 | |
|---|
| 38 | SMALL_MIN = -1e10 # Not that small, but small enough |
|---|
| 39 | vector = ensure_numeric(vector) |
|---|
| 40 | assert min > SMALL_MIN |
|---|
| 41 | no_maxs = where(less(vector,max), vector, SMALL_MIN) |
|---|
| 42 | band_condition = greater(no_maxs, min) |
|---|
| 43 | return band_condition |
|---|
| 44 | |
|---|
| 45 | |
|---|
| 46 | def auto_rrms(outputdir_tag, scenarios, quantity='stage', |
|---|
| 47 | y_location_tag=':0.0'): |
|---|
| 48 | """ |
|---|
| 49 | Given a list of scenarios that have CSV guage files, calc the |
|---|
| 50 | err, Number_of_samples and rmsd for all gauges in each scenario. |
|---|
| 51 | Write this info to a file for each scenario. |
|---|
| 52 | """ |
|---|
| 53 | for run_data in scenarios: |
|---|
| 54 | location_sims = [] |
|---|
| 55 | location_exps = [] |
|---|
| 56 | for gauge_x in run_data['gauge_x']: |
|---|
| 57 | gauge_x = str(gauge_x) |
|---|
| 58 | location_sims.append(gauge_x + y_location_tag) |
|---|
| 59 | location_exps.append(gauge_x) |
|---|
| 60 | |
|---|
| 61 | id = run_data['scenario_id'] |
|---|
| 62 | outputdir_name = id + outputdir_tag |
|---|
| 63 | file_sim = join(project.output_dir,outputdir_name + '_' + \ |
|---|
| 64 | quantity + ".csv") |
|---|
| 65 | file_exp = id + '_exp_' + quantity + '.csv' |
|---|
| 66 | file_err = join(project.output_dir,outputdir_name + "_" + \ |
|---|
| 67 | quantity + "_err.csv") |
|---|
| 68 | |
|---|
| 69 | |
|---|
| 70 | simulation, _ = csv2dict(file_sim) |
|---|
| 71 | experiment, _ = csv2dict(file_exp) |
|---|
| 72 | |
|---|
| 73 | time_sim = [float(x) for x in simulation['time']] |
|---|
| 74 | time_exp = [float(x) for x in experiment['Time']] |
|---|
| 75 | time_sim = ensure_numeric(time_sim) |
|---|
| 76 | time_exp = ensure_numeric(time_exp) |
|---|
| 77 | condition = get_max_min_condition_array(run_data['wave_times'][0], |
|---|
| 78 | run_data['wave_times'][1], |
|---|
| 79 | time_exp) |
|---|
| 80 | time_exp_cut = compress(condition, time_exp) |
|---|
| 81 | |
|---|
| 82 | print "Writing to ", file_err |
|---|
| 83 | |
|---|
| 84 | err_list = [] |
|---|
| 85 | points = [] |
|---|
| 86 | rmsd_list = [] |
|---|
| 87 | for location_sim, location_exp in map(None, location_sims, |
|---|
| 88 | location_exps): |
|---|
| 89 | quantity_sim = [float(x) for x in simulation[location_sim]] |
|---|
| 90 | quantity_exp = [float(x) for x in experiment[location_exp]] |
|---|
| 91 | |
|---|
| 92 | quantity_exp_cut = compress(condition, quantity_exp) |
|---|
| 93 | |
|---|
| 94 | # Now let's do interpolation |
|---|
| 95 | quantity_sim_interp = interp(quantity_sim, time_sim, time_exp_cut) |
|---|
| 96 | |
|---|
| 97 | assert len(quantity_sim_interp) == len(quantity_exp_cut) |
|---|
| 98 | norm = err(quantity_sim_interp, |
|---|
| 99 | quantity_exp_cut, |
|---|
| 100 | 2, relative = False) # 2nd norm (rel. RMS) |
|---|
| 101 | err_list.append(norm) |
|---|
| 102 | points.append(len(quantity_sim_interp)) |
|---|
| 103 | rmsd_list.append(norm/sqrt(len(quantity_sim_interp))) |
|---|
| 104 | assert len(location_exps) == len(err_list) |
|---|
| 105 | |
|---|
| 106 | # Writing the file out for one scenario |
|---|
| 107 | a_writer = writer(file(file_err, "wb")) |
|---|
| 108 | a_writer.writerow(["x location", "err", "Number_of_samples", "rmsd"]) |
|---|
| 109 | a_writer.writerows(map(None, |
|---|
| 110 | location_exps, |
|---|
| 111 | err_list, |
|---|
| 112 | points, |
|---|
| 113 | rmsd_list)) |
|---|
| 114 | |
|---|
| 115 | |
|---|
| 116 | |
|---|
| 117 | def load_sensors(quantity_file): |
|---|
| 118 | """ |
|---|
| 119 | Load a csv file, where the first row is the column header and |
|---|
| 120 | the first colum explains the rows. |
|---|
| 121 | |
|---|
| 122 | returns the data as two vectors and an array. |
|---|
| 123 | |
|---|
| 124 | """ |
|---|
| 125 | |
|---|
| 126 | # Read the depth file |
|---|
| 127 | dfid = open(quantity_file) |
|---|
| 128 | lines = dfid.readlines() |
|---|
| 129 | dfid.close() |
|---|
| 130 | |
|---|
| 131 | title = lines.pop(0) |
|---|
| 132 | n_time = len(lines) |
|---|
| 133 | n_sensors = len(lines[0].split(','))-1 # -1 to remove time |
|---|
| 134 | times = zeros(n_time, Float) #Time |
|---|
| 135 | depths = zeros(n_time, Float) # |
|---|
| 136 | sensors = zeros((n_time,n_sensors), Float) |
|---|
| 137 | quantity_locations = title.split(',') |
|---|
| 138 | quantity_locations.pop(0) # remove 'time' |
|---|
| 139 | |
|---|
| 140 | # Doing j.split(':')[0] drops the y location |
|---|
| 141 | locations = [float(j.split(':')[0]) for j in quantity_locations] |
|---|
| 142 | |
|---|
| 143 | for i, line in enumerate(lines): |
|---|
| 144 | fields = line.split(',') |
|---|
| 145 | fields = [float(j) for j in fields] |
|---|
| 146 | times[i] = fields[0] |
|---|
| 147 | sensors[i] = fields[1:] # 1: to remove time |
|---|
| 148 | |
|---|
| 149 | return times, locations, sensors |
|---|
| 150 | |
|---|
| 151 | |
|---|
| 152 | def err_files(scenarios, outputdir_tag, quantity='stage'): |
|---|
| 153 | """ |
|---|
| 154 | Create a list of err files, for a list of scenarios. |
|---|
| 155 | """ |
|---|
| 156 | file_errs = [] |
|---|
| 157 | for scenario in scenarios: |
|---|
| 158 | id = scenario['scenario_id'] |
|---|
| 159 | outputdir_name = id + outputdir_tag |
|---|
| 160 | file_err = join(project.output_dir,outputdir_name + "_" + \ |
|---|
| 161 | quantity + "_err.csv") |
|---|
| 162 | file_errs.append(file_err) |
|---|
| 163 | return file_errs |
|---|
| 164 | |
|---|
| 165 | |
|---|
| 166 | def compare_different_settings(outputdir_tag, scenarios, quantity='stage'): |
|---|
| 167 | """ |
|---|
| 168 | Calculate the RMSD for all the tests in a scenario |
|---|
| 169 | """ |
|---|
| 170 | files = err_files(scenarios, outputdir_tag, quantity=quantity) |
|---|
| 171 | err = 0.0 |
|---|
| 172 | number_of_samples = 0 |
|---|
| 173 | for run_data, file in map(None, scenarios, files): |
|---|
| 174 | |
|---|
| 175 | simulation, _ = csv2dict(file) |
|---|
| 176 | err_list = [float(x) for x in simulation['err']] |
|---|
| 177 | number_of_samples_list = [float(x) for x in \ |
|---|
| 178 | simulation['Number_of_samples']] |
|---|
| 179 | |
|---|
| 180 | if number_of_samples is not 0: |
|---|
| 181 | err_list.append(err) |
|---|
| 182 | number_of_samples_list.append(number_of_samples) |
|---|
| 183 | err, number_of_samples = err_addition(err_list, number_of_samples_list) |
|---|
| 184 | rmsd = err/sqrt(number_of_samples) |
|---|
| 185 | print outputdir_tag + " " + str(rmsd) |
|---|
| 186 | |
|---|
| 187 | |
|---|
| 188 | |
|---|
| 189 | def err_addition(err_list, number_of_samples_list): |
|---|
| 190 | """ |
|---|
| 191 | This function 'sums' a list of errs and sums a list of samples |
|---|
| 192 | |
|---|
| 193 | err is the err value (sqrt(sum_over_x&y((xi - yi)^2))) for a set of values. |
|---|
| 194 | number_of_samples is the number of values associated with the err. |
|---|
| 195 | |
|---|
| 196 | If this function gets used alot, maybe pull this out and make it an object |
|---|
| 197 | """ |
|---|
| 198 | err = norm(ensure_numeric(err_list)) |
|---|
| 199 | number_of_samples = sum(ensure_numeric(number_of_samples_list)) |
|---|
| 200 | |
|---|
| 201 | return err, number_of_samples |
|---|
| 202 | |
|---|
| 203 | |
|---|
| 204 | #------------------------------------------------------------- |
|---|
| 205 | if __name__ == "__main__": |
|---|
| 206 | |
|---|
| 207 | from scenarios import scenarios |
|---|
| 208 | |
|---|
| 209 | # Change the outputdir_tag to change which set of data |
|---|
| 210 | # the RMSD is calculated for. |
|---|
| 211 | outputdir_tag = "_nolmts_wdth_0.1_z_0.0_ys_0.01_mta_0.01_A" |
|---|
| 212 | auto_rrms(outputdir_tag, scenarios, "stage", y_location_tag=':0.0') |
|---|
| 213 | compare_different_settings(outputdir_tag, scenarios, "stage") |
|---|
| 214 | |
|---|
