Skip to content

Comments

Add parameter support to NLP diff engine#151

Open
Transurgeon wants to merge 6 commits intomasterfrom
nlp-param-support
Open

Add parameter support to NLP diff engine#151
Transurgeon wants to merge 6 commits intomasterfrom
nlp-param-support

Conversation

@Transurgeon
Copy link
Member

@Transurgeon Transurgeon commented Feb 17, 2026

Enable CVXPY Parameters as live nodes in the C expression tree, so parameter values can be updated via memcpy without rebuilding the tree. This is critical for parametric NLP problems (MPC, portfolio optimization).

Changes:

  • inverse_data.py: add get_param_offsets() static method
  • converters.py: add build_parameter_dict(), _normalize_shape() helper, Parameter branches in matmul/multiply converters, param_dict threading through convert_expr/convert_expressions (now returns 4-tuple)
  • c_problem.py: use convert_expressions(), register params with C problem, add update_params() method for fast parameter value refresh

Description

Please include a short summary of the change.
Issue link (if applicable):

Type of change

  • New feature (backwards compatible)
  • New feature (breaking API changes)
  • Bug fix
  • Other (Documentation, CI, ...)

Contribution checklist

  • Add our license to new files.
  • Check that your code adheres to our coding style.
  • Write unittests.
  • Run the unittests and check that they’re passing.
  • Run the benchmarks to make sure your change doesn’t introduce a regression.

Transurgeon and others added 2 commits February 17, 2026 16:26
@Transurgeon @claude
Add parameter support to NLP diff engine
d4fb774 
Enable CVXPY Parameters as live nodes in the C expression tree, so
parameter values can be updated via memcpy without rebuilding the tree.
This is critical for parametric NLP problems (MPC, portfolio optimization).

Changes:
- inverse_data.py: add get_param_offsets() static method
- converters.py: add build_parameter_dict(), _normalize_shape() helper,
  Parameter branches in matmul/multiply converters, param_dict threading
  through convert_expr/convert_expressions (now returns 4-tuple)
- c_problem.py: use convert_expressions(), register params with C problem,
  add update_params() method for fast parameter value refresh

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@Transurgeon @claude
Propagate initial values to reduced variables in CvxAttr2Constr
6f167a0 
The refactor in 33fe669 lost the inline code that copied initial
values from dimension-reduced variables (diag, sparse, symmetric) to
their reduced counterparts. This caused NLP solvers to find value=None
on the reduced variable, breaking initialization. Mirror the existing
parameter value-propagation pattern by calling lower_value(var) after
creating the reduced variable.

Also fix lower_value to handle sparse diagonal matrices by using
.diagonal() instead of np.diag(), which does not accept sparse inputs.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@github-actions
Copy link

github-actions bot commented Feb 17, 2026 •
edited
Loading

Benchmarks that have improved:

   before           after         ratio
 [0c4f10e2]       [6bac4ca2]

  •     548±0ms          493±0ms     0.90  semidefinite_programming.SemidefiniteProgramming.time_compile_problem

  •     1.22±0s          1.02±0s     0.84  gini_portfolio.Cajas.time_compile_problem

Benchmarks that have stayed the same:

   before           after         ratio
 [0c4f10e2]       [6bac4ca2]
     39.1±0ms         39.9±0ms     1.02  matrix_stuffing.SmallMatrixStuffing.time_compile_problem
      12.1±0s          12.3±0s     1.02  finance.CVaRBenchmark.time_compile_problem
     14.5±0ms         14.7±0ms     1.02  simple_LP_benchmarks.SimpleFullyParametrizedLPBenchmark.time_compile_problem
      1.56±0s          1.58±0s     1.01  tv_inpainting.TvInpainting.time_compile_problem
      986±0ms          999±0ms     1.01  finance.FactorCovarianceModel.time_compile_problem
      886±0ms          893±0ms     1.01  simple_LP_benchmarks.SimpleScalarParametrizedLPBenchmark.time_compile_problem
      671±0ms          675±0ms     1.01  matrix_stuffing.ConeMatrixStuffingBench.time_compile_problem
      20.5±0s          20.6±0s     1.01  sdp_segfault_1132_benchmark.SDPSegfault1132Benchmark.time_compile_problem
      737±0ms          740±0ms     1.00  simple_QP_benchmarks.LeastSquares.time_compile_problem
     15.1±0ms         15.2±0ms     1.00  simple_QP_benchmarks.ParametrizedQPBenchmark.time_compile_problem
      1.78±0s          1.78±0s     1.00  simple_QP_benchmarks.UnconstrainedQP.time_compile_problem
      226±0ms          226±0ms     1.00  gini_portfolio.Murray.time_compile_problem
      134±0ms          133±0ms     1.00  high_dim_convex_plasticity.ConvexPlasticity.time_compile_problem
      278±0ms          277±0ms     1.00  slow_pruning_1668_benchmark.SlowPruningBenchmark.time_compile_problem
      282±0ms          281±0ms     1.00  matrix_stuffing.ParamSmallMatrixStuffing.time_compile_problem
      3.92±0s          3.90±0s     1.00  huber_regression.HuberRegression.time_compile_problem
      1.41±0s          1.40±0s     1.00  matrix_stuffing.ParamConeMatrixStuffing.time_compile_problem
      4.46±0s          4.44±0s     1.00  svm_l1_regularization.SVMWithL1Regularization.time_compile_problem
      10.1±0s          10.1±0s     0.99  simple_LP_benchmarks.SimpleLPBenchmark.time_compile_problem
      239±0ms          238±0ms     0.99  simple_QP_benchmarks.SimpleQPBenchmark.time_compile_problem
      5.06±0s          5.02±0s     0.99  optimal_advertising.OptimalAdvertising.time_compile_problem
      317±0ms          313±0ms     0.98  gini_portfolio.Yitzhaki.time_compile_problem
      2.98±0s          2.79±0s     0.94  quantum_hilbert_matrix.QuantumHilbertMatrix.time_compile_problem

Transurgeon and others added 4 commits February 17, 2026 20:45
@Transurgeon @claude
Fuse sparse parameter matmul in diff engine converter
c2ab706 
Detect the reshape(sparse_coeff @ reduced_param, shape, 'F') @ x pattern
produced by CvxAttr2Constr for sparse parameters and fuse it into a single
make_left_matmul/make_right_matmul with native CSR sparsity, instead of
two separate matmuls. Values in update_params are converted to CSR order
via COO→CSR for fused sparse parameters.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@Transurgeon @claude
Add parameter derivative tests for NLP diff engine
082611b 
Test scalar, vector, dense matrix, and sparse matrix parameter
multiplication with derivative checking and parameter value updates.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@Transurgeon @claude
Remove dead parameter-to-constant fallthrough in converter
2bda8ae 
Parameters are always registered via build_parameter_dict, so the
fallthrough path that created them as make_constant was dead code.
Replace with an explicit error if a parameter is missing from param_dict.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@Transurgeon @claude
Add right matmul derivative test for sparse parameters
10419b9 
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
dance858
dance858 reviewed Feb 18, 2026
View reviewed changes
cvxpy/reductions/solvers/nlp_solvers/diff_engine/converters.py
# Recursive case: atoms
atom_name = type(expr).__name__

# Try to fuse sparse-parameter reconstruction matmuls before normal dispatch
Copy link
Collaborator

@dance858 dance858 Feb 18, 2026

Choose a reason for hiding this comment

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

is it possible to move this logic into convert_matmul?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@dance858 dance858 dance858 left review comments

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@Transurgeon @dance858