WIP: TVD objective in GST #506

rileyjmurray · 2024-11-20T15:52:09Z

I noticed there's a total variance distance objective function class which is almost finished. The only missing functions for the code to run without formally raising an error were RawTVDFunction.dterms and RawTVDFunction.chi2k_distributed_qty. The former function has a very simple "canonical" implementation, but it's numerically problematic since the 1-norm isn't differentiable at any vector with a zero component. I don't know what a mathematically correct version of the second function looks like (or if there is such an implementation) so I just have it return -1 to indicate "this doesn't mean what you think it means."

I'm mostly opening this PR to facilitate QPL-internal discussion on the topic.

You can run GST where the final iteration is TVD-based instead of log-likelihood using something like the following.

models, opt_results, cache = pygsti.algorithms.run_iterative_gst(
    dataset, starting_model, lsgst_circuit_list_of_lists,
    optimizer={'tol': 1e-5}, resource_alloc=None, verbosity=4,
    iteration_objfn_builders=['chi2'], final_objfn_builders=['tvd']
)

…me a file in test/performance so that it doesnt get discovered by testing frameworks like pytest

…t commit will simpify further, which may have consequences from the perspective of floating point arithmetic. Checking in NOW in case we want to revert to this version for numerical reasons.

…eparately, modify docstring for run_long_sequence_gst to be consistent with its implementation (including leaving notes for later).

…uge_group. Having the casting occur here makes sense, since a model has access to the associated StateSpace, Basis, and Evotype.

…tchboard.objfn_builder was used when switchboard.objfn_builder_modvi should have been used

rileyjmurray · 2025-02-17T20:29:33Z

@pcwysoc I merged in changes from develop and now I'm getting errors in my example script. I'm working on getting this ready for you to experiment with.

…g in develop)

rileyjmurray · 2025-02-18T14:17:41Z

@pcwysoc my latest push includes an example notebook you can use to get started. LMK if you have any questions!

rileyjmurray · 2025-04-17T00:52:16Z

pygsti/drivers/longsequence.py

+
+        HISTORICAL NOTE: "XX" indicates that we've at least _intended_ for the
+        keyword argument to be removed. I've removed documentation for options
+        that we never reference in the code (truncScheme, estimate_label, and
+        missingDataAction).


Could move this change (and related changes below) to another PR.

rileyjmurray · 2025-04-17T00:52:42Z

pygsti/layouts/copalayout.py

Need to document why changes to this file were needed.

rileyjmurray · 2025-04-17T00:52:57Z

pygsti/models/explicitmodel.py

Can move this change to another PR.

rileyjmurray · 2025-04-17T00:53:50Z

pygsti/objectivefns/objectivefns.py

+    # QUESTION: could I just call the base class's implementation of terms and then scale the 
+    # result? If so then I can avoid a ton of code duplication.


Need to resolve this question before merge.

rileyjmurray · 2025-04-17T00:54:09Z

pygsti/objectivefns/objectivefns.py

+    # QUESTION: could I just call the base class's implementation of dterms and then scale the 
+    # leading rows of the result? If so then I can avoid a ton of code duplication.


Need to resolve this before merge.

rileyjmurray · 2025-04-17T00:55:51Z

pygsti/report/factory.py

+            switchBd.objfn_builder[d, i] =  _objfns.ObjectiveFunctionBuilder.create_from('logl') # est.parameters.get('final_objfn_builder', _objfns.ObjectiveFunctionBuilder.create_from('logl'))
+            switchBd.objfn_builder_modvi[d, i] = _objfns.ObjectiveFunctionBuilder.create_from('logl')
+                # est_modvi.parameters.get('final_objfn_builder', _objfns.ObjectiveFunctionBuilder.create_from('logl'))


Need to explain why these changes were made.

rileyjmurray · 2025-04-17T00:56:16Z

pygsti/report/factory.py

+        - skip_sections : tuple[str], optional
+             Contains names of standard report sections that should be skipped
+             in this particular report. Strings will be cast to lowercase, 
+             stripped of white space, and then mapped to omitted Section classes
+             as follows
+
+                {
+                    'summary'         : SummarySection,
+                    'goodness'        : GoodnessSection,
+                    'colorbox'        : GoodnessColorBoxPlotSection,
+                    'invariantgates'  : GaugeInvariantsGatesSection,
+                    'invariantgerms'  : GaugeInvariantsGermsSection,
+                    'variant'         : GaugeVariantSection,
+                    'variantraw'      : GaugeVariantsRawSection,
+                    'variantdecomp'   : GaugeVariantsDecompSection,
+                    'varianterrorgen' : GaugeVariantsErrorGenSection,
+                    'input'           : InputSection,
+                    'meta'            : MetaSection,
+                    'help'            : HelpSection
+                }
+
+            A KeyError will be raised if skip_sections contains a string
+            that is not in the keys of the above dict (after casting to
+            lower case and stripping white space).
+
+


This can be moved to another PR, along with the changes below.

initial commit simplifying TimeIndependentMDCObjectiveFunction

bc64ecb

rileyjmurray mentioned this pull request Dec 16, 2024

Simplify the TimeIndependentMDCObjectiveFunction class #515

Merged

rileyjmurray added 16 commits December 16, 2024 16:39

bugfixes

b6e540f

improve readability

8f59891

inline comment

986e528

add pytest dispatcher for MPI test in test_packages

ccf17ef

move simple MPI test from test/test_packages into test/unit, and rena…

5959a53

…me a file in test/performance so that it doesnt get discovered by testing frameworks like pytest

implement changes discussed in 12/17 dev meeting

19316be

expand scope of exception handling

9b73727

add separate TimeIndpendentMDCObjectiveFunction.dlsvec back

fb4818c

resolve an infuriating sign error. Will probably want to clean this up.

122fd11

check in slightly simpler (but still complicated) implementation. Nex…

2ef13e2

…t commit will simpify further, which may have consequences from the perspective of floating point arithmetic. Checking in NOW in case we want to revert to this version for numerical reasons.

simplifications complete

1c23251

add comment

6e9d366

initial changes to RawTVDFunction

7d20e53

revert unncessary change

d15110e

fix a silly bug and implement per-gate weighting for TVD

b0e72b5

Merge remote-tracking branch 'origin' into TVD2

2719da0

rileyjmurray force-pushed the TVD branch from e16bee4 to 2719da0 Compare January 2, 2025 22:21

rileyjmurray and others added 6 commits February 17, 2025 13:35

Merge branch 'develop' into TVD

c33304a

add option handling for TVD objective in user-facing GST functions. S…

f8fea8e

…eparately, modify docstring for run_long_sequence_gst to be consistent with its implementation (including leaving notes for later).

add string-casting to the setter method of ExplicitOpModel.default_ga…

a67bece

…uge_group. Having the casting occur here makes sense, since a model has access to the associated StateSpace, Basis, and Evotype.

add the option of skipping report sections in construct_standard_report

2ed1c21

fix LIKELY bug in SummarySection.final_model_fit_histogram, where swi…

d028a77

…tchboard.objfn_builder was used when switchboard.objfn_builder_modvi should have been used

bugfix

48fb7ae

check in chnages for example (and getting example to run after mergin…

14c1a51

…g in develop)

rileyjmurray commented Apr 17, 2025

View reviewed changes

pygsti/layouts/copalayout.py

Copy link

Contributor Author

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Need to document why changes to this file were needed.

rileyjmurray commented Apr 17, 2025

View reviewed changes

pygsti/models/explicitmodel.py

Copy link

Contributor Author

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can move this change to another PR.

rileyjmurray commented Apr 17, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

WIP: TVD objective in GST #506

WIP: TVD objective in GST #506

Uh oh!

rileyjmurray commented Nov 20, 2024 •

edited

Loading

Uh oh!

rileyjmurray commented Feb 17, 2025

Uh oh!

rileyjmurray commented Feb 18, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Uh oh!

rileyjmurray Apr 17, 2025 •

edited

Loading

Uh oh!

Uh oh!

		# QUESTION: could I just call the base class's implementation of terms and then scale the
		# result? If so then I can avoid a ton of code duplication.

		# QUESTION: could I just call the base class's implementation of dterms and then scale the
		# leading rows of the result? If so then I can avoid a ton of code duplication.

WIP: TVD objective in GST #506

Are you sure you want to change the base?

WIP: TVD objective in GST #506

Uh oh!

Conversation

rileyjmurray commented Nov 20, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rileyjmurray commented Feb 17, 2025

Uh oh!

rileyjmurray commented Feb 18, 2025

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025

Choose a reason for hiding this comment

Uh oh!

rileyjmurray Apr 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

rileyjmurray commented Nov 20, 2024 •

edited

Loading

rileyjmurray Apr 17, 2025 •

edited

Loading