DiagnosticsToolbox

[andrewlee94]

Fixes None

Summary/Motivation:

This PR adds the first iteration of the new DiagnosticsToolbox class. Documentation and tests are coming, but we need input and feedback on the API before it gets set in stone.

A draft tutorial can be found here: https://app.box.com/folder/218647852677 (for those with access to the IDAES Box folder).

Changes proposed in this PR:

• New DiagnosticsToolbox class

Legal Acknowledgement

By contributing to this software project, I agree to the following terms and conditions for my contribution:

1. I agree my contributions are submitted under the license terms described in the LICENSE.txt file at the top level of this directory.
2. I represent I am authorized to make the contributions and grant the license. If my employer has rights to intellectual property that includes these contributions, I represent that I have received permission to make contributions and grant the required license on behalf of that employer.

[codecov]

Codecov Report

Patch coverage: 86.26% and project coverage change: +0.11% 🎉

Comparison is base (2e41a88) 76.59% compared to head (deeb154) 76.70%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1228      +/-   ##
==========================================
+ Coverage   76.59%   76.70%   +0.11%     
==========================================
  Files         382      382              
  Lines       60822    61182     +360     
  Branches    11200    11289      +89     
==========================================
+ Hits        46587    46932     +345     
  Misses      11818    11818              
- Partials     2417     2432      +15     

Files Changed	Coverage Δ
idaes/core/util/model_diagnostics.py	71.85% <85.38%> (+22.32%)	⬆️
idaes/core/util/__init__.py	100.00% <100.00%> (ø)
idaes/core/util/model_statistics.py	97.38% <100.00%> (+0.67%)	⬆️
idaes/core/util/scaling.py	90.47% <100.00%> (+1.99%)	⬆️

... and 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Robbybp]

Overall, this looks really good. Several minor comments, then I'd recommend printing out the independent diagonal blocks of the overconstrained/underconstrained subsystems individually, as this is often much easier to read if these sets decompose.

[dallan-keylogic]

I want to register a complaint that list_badly_scaled_variables is pretty feeble. This is what I got with it:

====================================================================================
The following variable(s) are poorly scaled:

    fs.condenser_flash.split.Liq_state[0.0].log_mole_frac_phase_comp[Liq,H2O]: 0.0009862360871915779

while this is what I got with extreme_jacobian_columns:

    2.00e-03, [fs.condenser_flash.control_volume.properties_in[0.0].log_mole_frac_comp[N2]]
    2.00e-03, [fs.condenser_flash.control_volume.properties_in[0.0].log_mole_frac_phase_comp[Vap,N2]]
    2.00e-03, [fs.condenser_flash.control_volume.properties_out[0.0].log_mole_frac_comp[N2]]
    2.00e-03, [fs.condenser_flash.control_volume.properties_out[0.0].log_mole_frac_phase_comp[Vap,N2]]
    5.29e-05, [fs.condenser_flash.split.Vap_state[0.0].temperature_dew[Liq,Vap]]
    2.00e-03, [fs.condenser_flash.split.Vap_state[0.0].log_mole_frac_comp[N2]]
    2.08e-03, [fs.condenser_flash.split.Vap_state[0.0].log_mole_frac_phase_comp[Vap,N2]]
    1.33e-13, [fs.condenser_flash.split.Liq_state[0.0].phase_frac[Liq]]
    1.33e-13, [fs.condenser_flash.split.Liq_state[0.0].phase_frac[Vap]]

I suppose 2e-3 is above the cutoff for the default setting, so that isn't an issue. It did miss the issue with the dew point and the enormous, matrix-is-singular issue with the phase fractions.

Also you need to make clear to the user that, whatever these methods return, they aren't going to be inerrant signs a variable or constraint is badly scaled, but just a sign you need to look at that variable or constraint to ensure it is well-scaled.

[andrewlee94]

@dallan-keylogic The poorly scaled variables check currently uses the badly_scaled_variables method out of the scaling toolbox, so it is a rather simple check. We used this for the prototype in order to have a basic demonstration using existing tools that were robust - I hope to either supplement or replace this with an svd tool once we have time to address the current issues and limitation with that tool.

[dallan-keylogic]

@andrewlee94 , You don't need to do a full SVD in order to use the extreme columns/rows tools, you just need a representation of the Jacobian. Creating a Jacobian from the block takes a bit of time, but not nearly as much as doing a full SVD.

[Robbybp]

Other than the docstring formatting, I think this is ready.