Conversation
There was a problem hiding this comment.
Pull request overview
Refactors Arkane’s project_rotors orthonormalization/projection workflow to improve numerical robustness by replacing manual Gram–Schmidt steps with QR decomposition, and adds diagnostics to detect rank/orthogonality issues during rotor projection.
Changes:
- Replaced Gram–Schmidt orthonormalization with NumPy QR decomposition for projection and internal rotor subspaces.
- Updated transformation matrix construction (
T) to use QR-derived orthonormal bases. - Added debug/warning logging for orthogonality, reconstruction checks, and near-zero
Rdiagonals.
Regression Testing Results
Detailed regression test results.Regression test aromatics:Reference: Execution time (DD:HH:MM:SS): 00:00:00:55 aromatics Passed Core Comparison ✅Original model has 15 species. aromatics Failed Edge Comparison ❌Original model has 106 species. Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-(Cds-Cds)CsHH) + group(Cds-CdsCsCs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + polycyclic(s3_4_5_ene_3) + polycyclic(s2_4_5_diene_1_5) + polycyclic(s2_5_5_diene_1_5) - ring(Cyclobutene) - ring(Cyclopentene) - ring(Cyclopentene) + radical(cyclopentene-allyl) Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-CsCsHH) + group(Cds-CdsCsCs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + Estimated bicyclic component: polycyclic(s2_3_5_ane) - ring(Cyclopropane) - ring(Cyclopentane) + ring(Cyclopentene) + ring(Cyclopropane) + polycyclic(s2_3_6_ene_1) + polycyclic(s3_5_6_diene_1_5) - ring(Cyclopropane) - ring(Cyclopentene) - ring(Cyclohexene) + radical(cyclopentene-4) Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-CsCsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsHH) + group(Cds- Cds(Cds-Cds)Cs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-Cds(Cds-Cds)H) + polycyclic(s2_3_5_ene_1) + polycyclic(s2_3_6_ene_1) + Estimated bicyclic component: polycyclic(s3_5_6_ane) - ring(Cyclopentane) - ring(Cyclohexane) + ring(Cyclopentene) + ring(Cyclohexene) - ring(Cyclopropane) - ring(Cyclopentene) - ring(Cyclohexene) + radical(cyclopentene-allyl) Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: DetailsObservables Test Case: Aromatics Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! aromatics Passed Observable Testing ✅Regression test liquid_oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:01:56 liquid_oxidation Passed Core Comparison ✅Original model has 37 species. liquid_oxidation Failed Edge Comparison ❌Original model has 214 species. Non-identical kinetics! ❌
kinetics: DetailsObservables Test Case: liquid_oxidation Comparison✅ All Observables varied by less than 0.100 on average between old model and new model in all conditions! liquid_oxidation Passed Observable Testing ✅Regression test nitrogen:Reference: Execution time (DD:HH:MM:SS): 00:00:01:03 nitrogen Passed Core Comparison ✅Original model has 41 species. nitrogen Passed Edge Comparison ✅Original model has 133 species. DetailsObservables Test Case: NC Comparison✅ All Observables varied by less than 0.200 on average between old model and new model in all conditions! nitrogen Passed Observable Testing ✅Regression test oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:01:46 oxidation Passed Core Comparison ✅Original model has 59 species. oxidation Passed Edge Comparison ✅Original model has 230 species. DetailsObservables Test Case: Oxidation Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! oxidation Passed Observable Testing ✅Errors occurred during observable testing
WARNING:root:Initial mole fractions do not sum to one; normalizing.
|
Regression Testing Results
Detailed regression test results.Regression test aromatics:Reference: Execution time (DD:HH:MM:SS): 00:00:00:57 aromatics Passed Core Comparison ✅Original model has 15 species. aromatics Failed Edge Comparison ❌Original model has 106 species. Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-(Cds-Cds)CsHH) + group(Cds-CdsCsCs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + polycyclic(s3_4_5_ene_3) + polycyclic(s2_4_5_diene_1_5) + polycyclic(s2_5_5_diene_1_5) - ring(Cyclobutene) - ring(Cyclopentene) - ring(Cyclopentene) + radical(cyclopentene-allyl) Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-(Cds-Cds)(Cds-Cds)CsH) + group(Cs-CsCsHH) + group(Cds-CdsCsCs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + Estimated bicyclic component: polycyclic(s2_3_5_ane) - ring(Cyclopropane) - ring(Cyclopentane) + ring(Cyclopentene) + ring(Cyclopropane) + polycyclic(s2_3_6_diene_0_3) + Estimated bicyclic component: polycyclic(s3_5_6_ane) - ring(Cyclopentane) - ring(Cyclohexane) + ring(Cyclopentene) + ring(1,4-Cyclohexadiene) - ring(Cyclopropane) - ring(Cyclopentene) - ring(1,4-Cyclohexadiene) + radical(cyclopentene-4) Non-identical thermo! ❌
thermo: Thermo group additivity estimation: group(Cs-CsCsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsCsH) + group(Cs-(Cds-Cds)CsHH) + group(Cds- Cds(Cds-Cds)Cs) + group(Cds-CdsCsH) + group(Cds-CdsCsH) + group(Cds-Cds(Cds-Cds)H) + polycyclic(s2_3_5_ene_1) + polycyclic(s2_3_6_diene_1_3) + Estimated bicyclic component: polycyclic(s3_5_6_ane) - ring(Cyclopentane) - ring(Cyclohexane) + ring(Cyclopentene) + ring(1,3-Cyclohexadiene) - ring(Cyclopropane) - ring(Cyclopentene) - ring(1,3-Cyclohexadiene) + radical(cyclopentene-allyl) Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: Non-identical kinetics! ❌
kinetics: DetailsObservables Test Case: Aromatics Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! aromatics Passed Observable Testing ✅Regression test liquid_oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:02:02 liquid_oxidation Passed Core Comparison ✅Original model has 37 species. liquid_oxidation Failed Edge Comparison ❌Original model has 214 species. DetailsObservables Test Case: liquid_oxidation Comparison✅ All Observables varied by less than 0.100 on average between old model and new model in all conditions! liquid_oxidation Passed Observable Testing ✅Regression test nitrogen:Reference: Execution time (DD:HH:MM:SS): 00:00:01:05 nitrogen Failed Core Comparison ❌Original model has 41 species. nitrogen Failed Edge Comparison ❌Original model has 133 species. DetailsObservables Test Case: NC Comparison✅ All Observables varied by less than 0.200 on average between old model and new model in all conditions! nitrogen Passed Observable Testing ✅Regression test oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:01:48 oxidation Passed Core Comparison ✅Original model has 59 species. oxidation Passed Edge Comparison ✅Original model has 230 species. DetailsObservables Test Case: Oxidation Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! oxidation Passed Observable Testing ✅Errors occurred during observable testing
WARNING:root:Initial mole fractions do not sum to one; normalizing.
|
rwest
force-pushed
the
arkanebug
branch
3 times, most recently
from
4b0d834 to
a1d53e9
Compare
Regression Testing Results
Detailed regression test results.Regression test aromatics:Reference: Execution time (DD:HH:MM:SS): 00:00:00:54 aromatics Passed Core Comparison ✅Original model has 15 species. aromatics Failed Edge Comparison ❌Original model has 106 species. Non-identical thermo! ❌
Identical thermo comments: Non-identical kinetics! ❌
kinetics: DetailsObservables Test Case: Aromatics Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! aromatics Passed Observable Testing ✅Regression test liquid_oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:01:59 liquid_oxidation Passed Core Comparison ✅Original model has 37 species. liquid_oxidation Failed Edge Comparison ❌Original model has 214 species. Non-identical kinetics! ❌
kinetics: DetailsObservables Test Case: liquid_oxidation Comparison✅ All Observables varied by less than 0.100 on average between old model and new model in all conditions! liquid_oxidation Passed Observable Testing ✅Regression test nitrogen:Reference: Execution time (DD:HH:MM:SS): 00:00:01:03 nitrogen Passed Core Comparison ✅Original model has 41 species. nitrogen Passed Edge Comparison ✅Original model has 133 species. DetailsObservables Test Case: NC Comparison✅ All Observables varied by less than 0.200 on average between old model and new model in all conditions! nitrogen Passed Observable Testing ✅Regression test oxidation:Reference: Execution time (DD:HH:MM:SS): 00:00:01:45 oxidation Passed Core Comparison ✅Original model has 59 species. oxidation Passed Edge Comparison ✅Original model has 230 species. DetailsObservables Test Case: Oxidation Comparison✅ All Observables varied by less than 0.500 on average between old model and new model in all conditions! oxidation Passed Observable Testing ✅Errors occurred during observable testing
WARNING:root:Initial mole fractions do not sum to one; normalizing.
|
This was originally called 'I' and during the PEP8 conversion around the py2 to py3 time, they were all renamed inertia, even though some of them represent the identity matrix not the moments of inertia
I think the previous might have left the final column in an undetermined state. And this might be quicker (fewer loops) and cleaner Python.
We were doing a manually coded Gram-Schmidt orthonormalization. This is now replaced with a QR decomposition built in to Numpy, which should be more robust (and probably faster). Added in some checks that print warnings.
Not sure it helps at all.
Motivation or Problem
When trying to project out rotors, Arkane would sometimes have major numerical issues and give very poor results.
These changes were made while trying to debug that issue - I switched to a more robust method for orthonormalization, and added some diagnostic defensive checks that would detect zero diagonal elements.
See discussion in #2757 (specifically #2757 (comment) and #2757 (comment) )
It turned out to have a different solution (ensuring Gaussian uses the same coordinate system for the Hessian matrix as the geometry, which is checked for in #2758 ) but the changes here are nevertheless probably improvements.
Description of Changes
This pull request refactors the
project_rotorsfunction inarkane/statmech.pyto modernize and simplify the orthonormalization and projection logic by replacing manual Gram-Schmidt procedures with NumPy's QR decomposition. It also adds extensive logging for debugging matrix orthogonality and projection accuracy.The most important changes are:
Refactoring and Modernization:
np.linalg.qr, streamlining the code and improving numerical stability. (arkane/statmech.py[1] [2]Tto use the orthonormalized basis from QR decomposition instead of manually normalized vectors. (arkane/statmech.pyarkane/statmech.pyR1101-R1127)Debugging and Logging Enhancements:
R(potentially indicating loss of basis vectors). (arkane/statmech.py[1] [2]arkane/statmech.pyarkane/statmech.pyL1213-R1242)Minor Corrections:
inertiatoidentityfor clarity when constructing the identity matrix for projection. (arkane/statmech.pyarkane/statmech.pyL1213-R1242)