library(OSLdecomposition)
source("../R/decompose_OSLcurve.R")
# For measuring execution time
library(tictoc)
# For printing tables
library(knitr)Module tests for decompose_OSLcurve()
This document collects scripts for developing, bug-fixing and pre-release validating a specific function.
Sourcing
Is the function successfully sourced?
Basic algorithm test
Create test curve
# Some common quartz CW-OSL measurement with a PMT
sim_components <- data.frame(name = c("Fast", "Medium", "Slow"),
lambda = c(1.5, 0.2, 0.01),
n = c(1000, 1000, 20000),
n.error = NA)
# Simulate the curve, add shot noise, detector noise and detector background
curve <- simulate_OSLcomponents(sim_components,
simulate.curve = TRUE,
add.poisson.noise = TRUE,
add.gaussian.noise = 2,
add.background = 10)
# How does the curve look like?
plot(x = curve$time, y = curve$signal, type = "l")
Cycle algorithms, without background correction
results <- data.frame(NULL)
for (arg_algorithm in c("det",
"nls",
"det+nls")) {
for (arg_error_est in c("empiric",
"poisson",
"only.bin.RSS",
"none",
2)) {
case_title <- paste("Algorithm =", arg_algorithm,
"| Error est. =", arg_error_est)
cat("\n", case_title, "\n")
# Function call with return
components <- decompose_OSLcurve(curve,
sim_components,
background.fitting = FALSE,
algorithm = arg_algorithm,
error.estimation = arg_error_est,
verbose = TRUE)
# Call the function now 1000 times to test performance and see if errors occur
tic("Execution time")
for (i in 1:1000) {
suppressWarnings({
decompose_OSLcurve(curve,
sim_components,
background.fitting = FALSE,
algorithm = arg_algorithm,
error.estimation = arg_error_est,
verbose = FALSE)
})
}
tictoc <- toc()
elapsed <- as.numeric((tictoc$toc - tictoc$tic)/1000)
result <- data.frame(test = case_title,
t(components$n),
t(components$n.error),
time = elapsed)
results <- rbind(results, result)
}
}
Algorithm = det | Error est. = empiric
name n n.error bin.RSS bin.error initial.signal n.residual
Fast 954.4813 50.82649 1143.165 34.91956 0.7559 0
Medium 1127.6177 114.88515 3481.258 59.21104 0.1269 0
Slow 20620.8451 333.07933 3379.889 58.25730 0.1172 13823
Execution time: 0.33 sec elapsed
Algorithm = det | Error est. = poisson
name n n.error bin.error initial.signal n.residual
Fast 954.4813 54.69045 38.78144 0.7559 0
Medium 1127.6177 117.68928 59.53150 0.1269 0
Slow 20620.8451 353.84590 61.90315 0.1172 13823
Execution time: 0.29 sec elapsed
Algorithm = det | Error est. = only.bin.RSS
name n n.error bin.RSS
Fast 954.4813 NA 1143.165
Medium 1127.6177 NA 3481.258
Slow 20620.8451 NA 3379.889
Execution time: 0.22 sec elapsed
Algorithm = det | Error est. = none
name n n.error
Fast 954.4813 NA
Medium 1127.6177 NA
Slow 20620.8451 NA
Execution time: 0.16 sec elapsed
Algorithm = det | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting =
FALSE, : Invalid error.estimation argument. name n n.error initial.signal n.residual
Fast 954.4813 NA 0.7559 0
Medium 1127.6177 NA 0.1269 0
Slow 20620.8451 NA 0.1172 13823
Execution time: 0.2 sec elapsed
Algorithm = nls | Error est. = empiric
name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.55 sec elapsed
Algorithm = nls | Error est. = poisson Warning in decompose_OSLcurve(curve, sim_components, background.fitting =
FALSE, : Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.57 sec elapsed
Algorithm = nls | Error est. = only.bin.RSS Warning in decompose_OSLcurve(curve, sim_components, background.fitting =
FALSE, : Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.58 sec elapsed
Algorithm = nls | Error est. = none
name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.64 sec elapsed
Algorithm = nls | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting =
FALSE, : Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.66 sec elapsed
Algorithm = det+nls | Error est. = empiric
name n n.error bin.RSS bin.error initial.signal n.residual
Fast 938.0487 50.47796 1124.097 34.62712 0.7555 0
Medium 1083.6310 114.47553 3442.138 58.87742 0.1241 0
Slow 20831.4080 333.33745 3385.314 58.30404 0.1204 13964
Execution time: 1.86 sec elapsed
Algorithm = det+nls | Error est. = poisson
name n n.error bin.error initial.signal n.residual
Fast 938.0487 54.69045 38.78144 0.7555 0
Medium 1083.6310 117.68928 59.53150 0.1241 0
Slow 20831.4080 353.84590 61.90315 0.1204 13964
Execution time: 1.83 sec elapsed
Algorithm = det+nls | Error est. = only.bin.RSS
name n n.error bin.RSS
Fast 938.0487 22.05404 1124.097
Medium 1083.6310 71.89500 3442.138
Slow 20831.4080 339.38346 3385.314
Execution time: 1.82 sec elapsed
Algorithm = det+nls | Error est. = none
name n n.error
Fast 938.0487 22.05404
Medium 1083.6310 71.89500
Slow 20831.4080 339.38346
Execution time: 1.57 sec elapsed
Algorithm = det+nls | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting =
FALSE, : Invalid error.estimation argument. name n n.error initial.signal n.residual
Fast 938.0487 22.05404 0.7555 0
Medium 1083.6310 71.89500 0.1241 0
Slow 20831.4080 339.38346 0.1204 13964
Execution time: 1.78 sec elapsedkable(results)| test | X1 | X2 | X3 | X1.1 | X2.1 | X3.1 | time |
|---|---|---|---|---|---|---|---|
| Algorithm = det | Error est. = empiric | 954.4813 | 1127.618 | 20620.85 | 50.82649 | 114.8851 | 333.0793 | 0.00033 |
| Algorithm = det | Error est. = poisson | 954.4813 | 1127.618 | 20620.85 | 54.69045 | 117.6893 | 353.8459 | 0.00029 |
| Algorithm = det | Error est. = only.bin.RSS | 954.4813 | 1127.618 | 20620.85 | NA | NA | NA | 0.00022 |
| Algorithm = det | Error est. = none | 954.4813 | 1127.618 | 20620.85 | NA | NA | NA | 0.00016 |
| Algorithm = det | Error est. = 2 | 954.4813 | 1127.618 | 20620.85 | NA | NA | NA | 0.00020 |
| Algorithm = nls | Error est. = empiric | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00155 |
| Algorithm = nls | Error est. = poisson | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00157 |
| Algorithm = nls | Error est. = only.bin.RSS | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00158 |
| Algorithm = nls | Error est. = none | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00164 |
| Algorithm = nls | Error est. = 2 | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00166 |
| Algorithm = det+nls | Error est. = empiric | 938.0487 | 1083.631 | 20831.41 | 50.47796 | 114.4755 | 333.3375 | 0.00186 |
| Algorithm = det+nls | Error est. = poisson | 938.0487 | 1083.631 | 20831.41 | 54.69045 | 117.6893 | 353.8459 | 0.00183 |
| Algorithm = det+nls | Error est. = only.bin.RSS | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00182 |
| Algorithm = det+nls | Error est. = none | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00157 |
| Algorithm = det+nls | Error est. = 2 | 938.0487 | 1083.631 | 20831.41 | 22.05404 | 71.8950 | 339.3835 | 0.00178 |
With background correction
Now do the same, with background correction.
# Background fittings needs pre-defined integration intervals
sim_components <- optimise_OSLintervals(sim_components, curve,
background.component = TRUE)Maximum determinant = 0.1092 with interval dividing channels at i = 13, 81, 250 results <- data.frame(NULL)
for (arg_algorithm in c("det",
"nls",
"det+nls")) {
for (arg_error_est in c("empiric",
"poisson",
"only.bin.RSS",
"none",
2)) {
case_title <- paste("Algorithm =", arg_algorithm,
"| Error est. =", arg_error_est)
cat("\n", case_title, "\n")
# Function call with return
components <- decompose_OSLcurve(curve,
sim_components,
background.fitting = TRUE,
algorithm = arg_algorithm,
error.estimation = arg_error_est,
verbose = TRUE)
# Call the function now 1000 times to test performance and see if errors occur
tic("Execution time")
for (i in 1:1000) {
suppressWarnings({
decompose_OSLcurve(curve,
sim_components,
background.fitting = TRUE,
algorithm = arg_algorithm,
error.estimation = arg_error_est,
verbose = FALSE)
})
}
tictoc <- toc()
elapsed <- as.numeric((tictoc$toc - tictoc$tic)/1000)
result <- data.frame(test = case_title,
t(components$n),
t(components$n.error),
time = elapsed)
results <- rbind(results, result)
}
}
Algorithm = det | Error est. = empiric
name n n.error bin.RSS bin.error initial.signal n.residual
Fast 1023.58020 NA NA NA 0.7899 0
Medium 761.14579 NA NA NA 0.0835 0
Slow 29156.35126 NA NA NA 0.1615 19544
Background -62.91838 NA NA NA -0.0349 NA
Execution time: 0.45 sec elapsed
Algorithm = det | Error est. = poisson
name n n.error bin.error initial.signal n.residual
Fast 1023.58020 54.72146 36.70150 0.7899 0
Medium 761.14579 158.69096 45.23273 0.0835 0
Slow 29156.35126 4704.43965 57.12268 0.1615 19544
Background -62.91838 35.98264 47.15930 -0.0349 NA
Execution time: 0.45 sec elapsed
Algorithm = det | Error est. = only.bin.RSS
name n n.error bin.RSS
Fast 1023.58020 NA NA
Medium 761.14579 NA NA
Slow 29156.35126 NA NA
Background -62.91838 NA NA
Execution time: 0.24 sec elapsed
Algorithm = det | Error est. = none
name n n.error
Fast 1023.58020 NA
Medium 761.14579 NA
Slow 29156.35126 NA
Background -62.91838 NA
Execution time: 0.15 sec elapsed
Algorithm = det | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting = TRUE,
: Invalid error.estimation argument. name n n.error initial.signal n.residual
Fast 1023.58020 NA 0.7899 0
Medium 761.14579 NA 0.0835 0
Slow 29156.35126 NA 0.1615 19544
Background -62.91838 NA -0.0349 NA
Execution time: 0.2 sec elapsed
Algorithm = nls | Error est. = empiric
name n n.error initial.signal n.residual
Fast 949.72749 23.51746 0.7626 0
Medium 965.69785 109.91764 0.1102 0
Slow 26526.32724 4033.06206 0.1528 17781
Background -44.50749 31.40807 -0.0257 NA
Execution time: 1.98 sec elapsed
Algorithm = nls | Error est. = poisson Warning in decompose_OSLcurve(curve, sim_components, background.fitting = TRUE,
: Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 949.72749 23.51746 0.7626 0
Medium 965.69785 109.91764 0.1102 0
Slow 26526.32724 4033.06206 0.1528 17781
Background -44.50749 31.40807 -0.0257 NA
Execution time: 1.94 sec elapsed
Algorithm = nls | Error est. = only.bin.RSS Warning in decompose_OSLcurve(curve, sim_components, background.fitting = TRUE,
: Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 949.72749 23.51746 0.7626 0
Medium 965.69785 109.91764 0.1102 0
Slow 26526.32724 4033.06206 0.1528 17781
Background -44.50749 31.40807 -0.0257 NA
Execution time: 2.06 sec elapsed
Algorithm = nls | Error est. = none
name n n.error initial.signal n.residual
Fast 949.72749 23.51746 0.7626 0
Medium 965.69785 109.91764 0.1102 0
Slow 26526.32724 4033.06206 0.1528 17781
Background -44.50749 31.40807 -0.0257 NA
Execution time: 1.88 sec elapsed
Algorithm = nls | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting = TRUE,
: Specific error.estimation argument without effect when algorithm = nls name n n.error initial.signal n.residual
Fast 949.72749 23.51746 0.7626 0
Medium 965.69785 109.91764 0.1102 0
Slow 26526.32724 4033.06206 0.1528 17781
Background -44.50749 31.40807 -0.0257 NA
Execution time: 2.12 sec elapsed
Algorithm = det+nls | Error est. = empiric
name n n.error bin.RSS bin.error initial.signal n.residual
Fast 949.7275 NA NA NA 0.7626 0
Medium 965.6979 NA NA NA 0.1102 0
Slow 26526.3275 NA NA NA 0.1528 17781
Background -44.5075 NA NA NA -0.0257 NA
Execution time: 2.36 sec elapsed
Algorithm = det+nls | Error est. = poisson
name n n.error bin.error initial.signal n.residual
Fast 949.7275 54.72146 36.70150 0.7626 0
Medium 965.6979 158.69096 45.23273 0.1102 0
Slow 26526.3275 4704.43965 57.12268 0.1528 17781
Background -44.5075 35.98264 47.15930 -0.0257 NA
Execution time: 2.15 sec elapsed
Algorithm = det+nls | Error est. = only.bin.RSS
name n n.error bin.RSS
Fast 949.7275 23.51746 NA
Medium 965.6979 109.91762 NA
Slow 26526.3275 4033.06187 NA
Background -44.5075 31.40807 NA
Execution time: 1.95 sec elapsed
Algorithm = det+nls | Error est. = none
name n n.error
Fast 949.7275 23.51746
Medium 965.6979 109.91762
Slow 26526.3275 4033.06187
Background -44.5075 31.40807
Execution time: 1.72 sec elapsed
Algorithm = det+nls | Error est. = 2 Warning in decompose_OSLcurve(curve, sim_components, background.fitting = TRUE,
: Invalid error.estimation argument. name n n.error initial.signal n.residual
Fast 949.7275 23.51746 0.7626 0
Medium 965.6979 109.91762 0.1102 0
Slow 26526.3275 4033.06187 0.1528 17781
Background -44.5075 31.40807 -0.0257 NA
Execution time: 2.09 sec elapsedkable(results)| test | X1 | X2 | X3 | X4 | X1.1 | X2.1 | X3.1 | X4.1 | time |
|---|---|---|---|---|---|---|---|---|---|
| Algorithm = det | Error est. = empiric | 1023.5802 | 761.1458 | 29156.35 | -62.91838 | NA | NA | NA | NA | 0.00045 |
| Algorithm = det | Error est. = poisson | 1023.5802 | 761.1458 | 29156.35 | -62.91838 | 54.72146 | 158.6910 | 4704.440 | 35.98264 | 0.00045 |
| Algorithm = det | Error est. = only.bin.RSS | 1023.5802 | 761.1458 | 29156.35 | -62.91838 | NA | NA | NA | NA | 0.00024 |
| Algorithm = det | Error est. = none | 1023.5802 | 761.1458 | 29156.35 | -62.91838 | NA | NA | NA | NA | 0.00015 |
| Algorithm = det | Error est. = 2 | 1023.5802 | 761.1458 | 29156.35 | -62.91838 | NA | NA | NA | NA | 0.00020 |
| Algorithm = nls | Error est. = empiric | 949.7275 | 965.6979 | 26526.33 | -44.50749 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00198 |
| Algorithm = nls | Error est. = poisson | 949.7275 | 965.6979 | 26526.33 | -44.50749 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00194 |
| Algorithm = nls | Error est. = only.bin.RSS | 949.7275 | 965.6979 | 26526.33 | -44.50749 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00206 |
| Algorithm = nls | Error est. = none | 949.7275 | 965.6979 | 26526.33 | -44.50749 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00188 |
| Algorithm = nls | Error est. = 2 | 949.7275 | 965.6979 | 26526.33 | -44.50749 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00212 |
| Algorithm = det+nls | Error est. = empiric | 949.7275 | 965.6979 | 26526.33 | -44.50750 | NA | NA | NA | NA | 0.00236 |
| Algorithm = det+nls | Error est. = poisson | 949.7275 | 965.6979 | 26526.33 | -44.50750 | 54.72146 | 158.6910 | 4704.440 | 35.98264 | 0.00215 |
| Algorithm = det+nls | Error est. = only.bin.RSS | 949.7275 | 965.6979 | 26526.33 | -44.50750 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00195 |
| Algorithm = det+nls | Error est. = none | 949.7275 | 965.6979 | 26526.33 | -44.50750 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00172 |
| Algorithm = det+nls | Error est. = 2 | 949.7275 | 965.6979 | 26526.33 | -44.50750 | 23.51746 | 109.9176 | 4033.062 | 31.40807 | 0.00209 |
Input checks
Try out a variety of possible cases for the lambda input
component_inputs <- list(as_usual = sim_components,
lambda_only = sim_components$lambda,
named_vector = c(Schnell = 1,
Mittel = 0.1,
Langsam = 0.01),
single_lambda = 0.5,
many_lambda = c(2,1,0.5,0.2,0.1,0.05, 0.01))
for (i in 1:length(component_inputs)) {
cat("\nTest case:", names(component_inputs)[i], "\n")
decompose_OSLcurve(curve, component_inputs[[i]])
}
Test case: as_usual Warning in signal + n[k] * diff(exp(-lambda[k] * c(0, curve$time))): Länge des längeren Objektes
ist kein Vielfaches der Länge des kürzeren Objektes name n n.error bin.RSS bin.error initial.signal n.residual
Fast 1011.9862 54.11282 1353.377 38.29045 0.7838 0
Medium 874.1546 110.77269 1836.248 43.17007 0.0962 0
Slow 21590.2630 479.43393 3102.832 55.86860 0.1200 14472
Test case: lambda_only
name n n.error bin.RSS bin.error initial.signal n.residual
Component 1 954.4813 50.82649 1143.165 34.91956 0.7559 0
Component 2 1127.6177 114.88515 3481.258 59.21104 0.1269 0
Component 3 20620.8451 333.07933 3379.889 58.25730 0.1172 13823
Test case: named_vector
name n n.error bin.RSS bin.error initial.signal n.residual
Schnell 1217.363 65.84634 2394.428 49.90196 0.7863 0
Mittel 1183.630 186.50176 4568.671 67.73936 0.0799 22
Langsam 19718.174 489.82669 1748.545 41.96156 0.1338 13217
Test case: single_lambda
name n n.error bin.RSS bin.error initial.signal n.residual
Component 1 8880 1015.382 1028423 1015.382 1 0
Test case: many_lambda
name n n.error bin.RSS bin.error initial.signal
Component 1 37.34566 377.8692 560.7364 25.57719 0.0415
Component 2 1525.98218 1334.5621 396.6770 21.51255 0.8911
Component 3 -402.79441 2267.0120 870.8476 30.41834 -0.1205
Component 4 -155.15307 4563.7322 711.1425 27.01586 -0.0189
Component 5 993.77184 8230.8651 1566.6287 39.87063 0.0607
Component 6 1225.08172 8110.3724 2838.6232 53.40980 0.0375
Component 7 17713.27711 6730.8286 774.7833 28.08236 0.1086
n.residual
0
0
0
0
18
166
11874