Fifth week - Major updates on literally everything, and non-LTE benchmarking
1. Fitting method benchmarking
The ideal of benchmarking result is to test and assess under what conditions, such as fitting method, pipeline, refinement, etc., the fitting process can achieve a stable and robust result. Firstly, I want to test the fitting method and see which ones are the best to put into the module as the default method. As we use LMFIT.Minimizer, we have 23 fitting methods in total:
-
leastsq
: Levenberg-Marquardt (default). -
least_squares
: Least-Squares minimization, using Trust Region Reflective method. -
differential_evolution
: differential evolution. -
brute
: brute force method. -
basinhopping
: Basin-hopping method. -
ampgo
: Adaptive Memory Programming for Global Optimization. -
nelder
: Nelder-Mead. -
lbfgsb
: Limited-memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS-B). -
powell
: Powell’s method. -
cg
: Conjugate-Gradient. -
newton
: Newton-Conjugate-Gradient. -
cobyla
: Cobyla. -
bfgs
: Broyden–Fletcher–Goldfarb–Shanno (BFGS). -
tnc
: Truncated Newton. -
trust-ncg
: Newton-Conjugate-Gradient trust-region. -
trust-exact
: nearly exact trust-region. -
trust-krylov
: Newton’s Generalized Lanczos Trust-Region (GLTR). -
trust-constr
: trust-region for constrained optimization. -
dogleg
: Dog-leg trust-region. -
slsqp
: Sequential Linear Squares Programming. -
emcee
: Maximum likelihood via Monte-Carlo Markov Chain. -
shgo
: Simplicial Homology Global Optimization. -
dual_annealing
: Dual Annealing optimization.
In this list, there are 5 methods - newton
, trust-ncg
, trust_exact
, trust-krylov
and dogleg
- that require Jacobian function to work, which adds more complexity into our fitting process and codebase, hence I remove them from the benchmark and never use them again. There are also emcee
method that, for some unknown reasons, the fitting procedure never stops even after passing the loop limit, thus I have to remove it. Now we have 17 methods left that are stable enough to compare. Additionally, I set the max number of fitting loops as 200, so this means that any method that have equal of higher than 200 loops means that they are most likely unable to stop. I will conduct benchmarking process on the CO2_measured_spectrum_4-5um.spec
first.
The result for this method-comparing benchmark can be found in this JSON file.
Method | Last residual | Number of loops | Processing time (s) |
---|---|---|---|
leastsq | 0.0027299042272 | 17 | 6.128568887710571 |
least_squares | 0.0027299046347 | 14 | 3.8792104721069336 |
differential_evolution | 0.0027299042330 | 48 | 7.211840629577637 |
brute | 0.0027847218345 | 20 | 3.13600492477417 |
basinhopping | 0.0030471725482 | 201 | 31.650216579437256 |
ampgo | 0.0027301332094 | 201 | 36.60996413230896 |
nelder | 0.0027299042330 | 48 | 7.532714605331421 |
lbfgsb | 0.0027299043815 | 12 | 1.8955962657928467 |
powell | 0.0027299042271 | 38 | 6.310025691986084 |
cg | 0.0027299046922 | 34 | 5.223567724227905 |
cobyla | 0.0027299044752 | 22 | 3.028048515319824 |
bfgs | 0.0027299042351 | 20 | 2.9560532569885254 |
tnc | 0.0027299042284 | 36 | 5.905533313751221 |
trust-constr | 0.0027299042271 | 16 | 2.3700413703918457 |
slsqp | 0.0027299969016 | 18 | 3.160074472427368 |
shgo | 0.0027299042272 | 32 | 6.185185194015503 |
dual_annealing | 0.0221619241989 | 201 | 32.40411591529846 |
(It is important to remember that this result might differ for each run, but rest assure the common trend is unchanged)
As you can see from the data above, we have basinhopping
, ampgo
and dual_annealing
jumping out of the loop limit of 200, and it’s totally not a good thing, which I would like to exclude them out for the sake of better visualization. Then, in order to compare the rest of 14 methods, I have a scatter plot below in which I focus on analyzing the last_residual
- indicator of accuracy - on the horizontal axis, and loops
- indicator of fitting iterations needed - on the vertical axis. As time
is heavily influenced by the computational capacity of each device, I don’t prioritize it than other two criteria in the result assessment, and thus it is indicated by color code.
If we zoom in the best 8 cases marked by the red rectangle above:
In the zoomed figure, the result is quite satisfying as I expected. When we talk about the most famous curve-fitting algorithms, we can mention either leastsq
/least_squares
or bfgs
/lbfgsb
, and now we can see them taking 4 out of top 5. Now let’s focus on the two competitors: lbfgsb
and least_squares
and get some observations:
-
lbfgsb
has a little lower residual and so a little bit better in accuracy thanleast_squares
. - Although approximately same fitting loops (12 and 14), the time required for
lbfgsb
is 1.895596s, significantly lower thanleast_squares
of 3.879210s. We can also see this behavior in their neighbors:bfgs
(2.956053s) <leastsq
(6.128569s).
This is explainable. While least_squares
simply calculating and minimizing the sum of the residuals of points from the comparative curves, lbfgsb
- Limited-memory BFGS uses a limited amount of computer memory to conduct Broyden-Fletcher-Goldfarb-Shanno algorithm for the minimization.
So for now, I have initial assumption that lbfgsb
performs slightly better than leastsq
. After conducting fitting process on other spectra, the lbfgsb
and leastsq
are seem to be dominant in terms of speed (based on number of loops and time elapsed) and accuracy (this depends a little bit on pipeline combination, which will be addressed in next part) compared to other methods. However, more benchmarks are needed to confirm my initla assumption.
2. Fitting pipeline comparison
Next is the benchmarking process focusing on pipeline comparison. A fitting pipeline comprises of several options, from spectrum refinement methods such as which spectral quantity to take, whether applying normalization on both spectra or not, or simply just fitting process preferences such as maxinum number of fitting loops allowed, or fitting method, or max fitting tolerance. While in the new JSON structure the users are free to adjust all of them, through this benchmarking process I would like acquire more understandings about how these pipeline might affect the quality of a fitting work.
In order to assess a fitting’s quality, I use the synthetic spectra that I generated on week 1. Although they are heavily convoluted with noises and offsets, since they are software-generated, we can know what are the experimental properties (such as path_length
, slit
, etc.). Meanwhile, regarding the experimental spectrum CO2_measured_spectrum_4-5um
, we are definitely not sure those parameters (in fact, we don’t even know whether it is LTE or non-LTE). So, I decided to test on 7 synthetic spectra:
synth-CO-1-1800-2300-cm-1-P3-t1500-v-r-mf0.1-p1-sl1nm.spec
synth-CO2-1-500-1100-cm-1-P2-t900-v-r-mf0.5-p1-sl1nm.spec
synth-CO2-1-500-3000-cm-1-P93-t740-v-r-mf0.96-p1-sl1nm.spec
synth-CO2-1-3300-3700-cm-1-P0.005-t3000-v-r-mf0.01-p1-sl1.4nm.spec
synth-H2O-1-1000-2500-cm-1-P0.5-t1500-v-r-mf0.5-p1-sl1nm.spec
synth-NH3-1-500-2000-cm-1-P10-t1000-v-r-mf0.01-p1-sl1nm.spec
synth-O2-1-7500-8000-cm-1-P1.01325-t298.15-v-r-mf0.21-p1-sl1nm.spec
For each spectra, I will modify the pipeline’s method
and normalize
properties, until the best possible fitting result is achieved - least fitting loops, and closest to the ground-truth (GT) temperature (this is why I prefer synthetic over experimental spectra, as I explained above). Here is the result:
Spec No. | GT Temp | Start Temp | Best Temp | Method | Normalize | Diff |
---|---|---|---|---|---|---|
1 | 1500 | 1300 | 1468.97 | lbfgsb |
false | 1 |
2 | 900 | 1300 | 898.84 | lbfgsb |
false | 2 |
3 | 740 | 1000 | ~740 | both | false | 3 |
4 | 3000 | 2850 | 3003.52 | leastsq |
false | 4 |
5 | 1500 | 2000 | 1507.25 | leastsq |
true | 5 |
6 | 1000 | 2250 | 994.90 | leastsq |
false | 6 |
7 | 298.15 | 660 | 297.86 | leastsq |
false | 7 |
As we can see from the table above:
- All spectra achieve near-perfect best fit results. This is because we have perfect ground-truth conditions. In real-life circumstances, such accurate ground-truth is virtually impossible to achieve, but this is the job of fitting users to measure and set them.
-
leastsq
performs quite good in most case. This is quite surprising after the result fromCO2_measured_spectrum_4-5um.spec
, but it can be explained as most likely we didn’t use the correct ground-truth conditions for it since we don’t know (seriously, who created that spectrum?). However, there is a small observation from me that, those cases wherelbfgsb
lost againsleastsq
were primarily because of number of loops. Still, it’s a win forleastsq
. - Their neighbors,
least_squares
andbfgs
, are completely underdogs. We don’t even need to mention other methods. - In nearly all the best cases,
normalize = false
is set. This is quite explainable, the more originality the better.
3. Summary
After all the benchmarking works above, I have decided to set the leastsq
as the default fitting method, in case users don’t state the method explicitly in JSON file. Later on, in the fitting tutorial, I will add some suggestions about using lbfgsb
and trying to switch the normalize
in case their fitting work is not quite good.
Nevertheless, these benchmarks helped me gain more insights about the performance of my fitting module, and most importantly, let me experience the feeling of a spectroscopist trying to fit his spectra - playing around the parameters, adjusting parameters and praying for a good result to come. Quite a physically and mentally exhausting work to be honest, since whenever the result went wrong, I didn’t know whether the error came from ground-truth conditions, or from my fitting module. There have been days and nights I sat in front of my laptop adjusting the JSON files and codebase continuously. But finally, the benchmarking process for LTE spectra is good now, and I am quite confident in my fitting module. Now let’s move on to the non-LTE spectra!