feat(ac-sweep): add port_aliases, holomorphic mode, and per-port z0 support - #37
Merged
Conversation
Extends AC sweep analysis to support complex-valued (photonic) circuits via: - assemble_gc_complex(): Extracts complex G/C matrices from 4-block Jacobian using Wirtinger formula for frequency-domain admittance - setup_ac_sweep(): Added is_complex parameter to select real or complex assembly path; handles 2N block-format DC point for complex circuits - Circuit.ac(): Removed ValueError guard blocking complex circuits; now passes is_complex through to setup_ac_sweep for seamless integration - Tests: 5 comprehensive tests covering shapes, finiteness, passivity, JIT, and Circuit.ac() integration with photonic waveguide models - Benchmarks: Import reformatting for consistency Addresses GitHub issue #29.
Demonstrates that linearized AC analysis (one matrix solve per frequency) achieves the same EO bandwidth curve as transient sweep from Part 2, but ~168× faster. Adds three new cells explaining the 2N block approach for non-holomorphic EO circuits, extracting G/C Jacobian blocks at DC operating point, and comparing AC/transient/analytic curves. Updates table of contents and Part 2 introduction to reference Part 2b.
…ck system Add nonholomorphic=True parameter to setup_ac_sweep() and Circuit.ac() to enable full 2N×2N real-block AC analysis. Required for circuits with non-holomorphic operations (jnp.real(), jnp.abs()) that couple the field and conjugate, making the Wirtinger N×N system incomplete. New functions: - assemble_gc_complex_2n(): returns four separate Jacobian blocks [RR, RI, IR, II] for both G and C matrices, instead of collapsing via Wirtinger formula - _setup_ac_sweep_2n(): AC sweep using 2N×2N real-block system with proper port terminations and S-parameter extraction at real-part indices Updated: - Circuit.ac(): added nonholomorphic parameter wired to setup_ac_sweep() - setup_ac_sweep(): added nonholomorphic parameter, dispatches to _setup_ac_sweep_2n() - tests: 3 new tests verifying equivalence with holomorphic for waveguides, Circuit.ac() interface, and JIT compatibility
…efault Rename parameter from `nonholomorphic=False` to `holomorphic=True` across `setup_ac_sweep()`, `Circuit.ac()`, and all tests. The positive default reads more naturally — most photonic components are holomorphic; users opt out with `holomorphic=False` when the circuit contains non-holomorphic operations like `jnp.real()` or `jnp.abs()`. - Update circuit.py and ac_sweep.py parameter and docstrings - Flip conditional logic: `if nonholomorphic:` → `if not holomorphic:` - Rename test functions: test_nonholomorphic_* → test_non_holomorphic_* - Update all test calls to use holomorphic=False Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Ti1GWaiLy8ndzJ5maMVYYk
… ring modulator Add test_non_holomorphic_ring_modulator_matches_analytic() and helper _ring_modulator_circuit() to verify that the 2N block system (holomorphic=False) correctly handles non-holomorphic EO power modulation in ring modulators. The test demonstrates that Wirtinger (holomorphic=True) produces >1 dB error on power-domain quantities when jnp.real() breaks holomorphicity, while the 2N formulation matches analytic behavior to <0.01 dB.
Renamed ac() method to sp() for consistency with S-parameter naming. The old ac() method now emits a DeprecationWarning (removed in 0.3) and delegates to sp(). Updated all tests to use sp() and added test_ac_deprecation_warning to verify backward compatibility. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Ti1GWaiLy8ndzJ5maMVYYk
Add flexible impedance parameter to setup_ac_sweep() and circuit.sp()/ac(): - Accept scalar z0 (uniform across ports and frequencies) - Accept (N_ports,) array for per-port reference impedance - Accept (N_freqs, N_ports) array for per-frequency, per-port impedance Implement _normalize_z0() helper to validate and broadcast z0 shapes. Update both N×N holomorphic and 2N×2N real-block AC sweep paths to receive z0 per-frequency via vmap to _solve_one_freq(). Add three tests: per-port scalar equivalence, per-port analytical match at 75Ω, and per-frequency with varying z0 across frequencies. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Ti1GWaiLy8ndzJ5maMVYYk
…fy z0 Remove per-frequency z0 from ac_sweep solve path (z0 is now scalar or per-port only, constant across frequencies). Add renormalize(S, z0_from, z0_to) function for post-hoc impedance renormalization. Simplify vmap loop by removing z0 as a vmap'd argument. Update docstrings to document the change and guide users to use renormalize for frequency-dependent impedance needs.
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
Add holomorphic parameter to @component and @source decorators to declare whether components use non-holomorphic operations (jnp.real, jnp.abs, etc). circuit.sp() now auto-infers holomorphic mode from component flags (holomorphic="auto" by default), eliminating the need for users to manually pass holomorphic=False. - ComponentGroup.holomorphic aggregates per-component flags (defaults to True) - _infer_holomorphic() checks if all groups are holomorphic - RingEO marked as holomorphic=False - Add test_holomorphic_auto_detection() to verify auto-detection path Backward compatible: explicit holomorphic={True,False} overrides auto mode.
…ex circuits Add runtime validation that warns when a component marked holomorphic=True uses non-holomorphic JAX primitives (real, imag, conj, abs) on traced complex variables. Runs at compile_circuit time for complex circuits only, avoiding false positives on electronic-only circuits. Filters out operations on constants (e.g. conj(z0)) by only flagging primitives with traced complex inputs. - Added _jaxpr_has_non_holomorphic() in base_component.py to walk jaxpr and detect non-holomorphic primitives on traced complex variables - Added _validate_holomorphic_flags() in circuit.py called from compile_circuit when is_complex=True - Added test_holomorphic_jaxpr_validation_warns to verify warnings on bad components - Added test_holomorphic_jaxpr_validation_no_false_positive to ensure electronic circuits don't trigger false warnings Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Ti1GWaiLy8ndzJ5maMVYYk
…olomorphic types Make the 2N×2N real-block system the default for AC analysis, ensuring correctness for non-holomorphic components (diodes, transistors, nonlinear devices). Explicitly mark as holomorphic=True only components guaranteed to preserve analyticity (resistors, capacitors, inductors, ideal op-amps, optical waveguides, etc.). Also updates test fixtures and adds documentation on holomorphic components.
cdaunt
marked this pull request as ready for review
July 27, 2026 11:48
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
Collaborator
Author
|
fixes #29 |
cdaunt
force-pushed
the
fix/issue-30-port-aliases
branch
from
July 27, 2026 21:07
f93a4c8 to
1332ab3
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
port_aliasesparameter to component decorators for port aliasing supportcircuit.ac()withcircuit.sp()(AC method deprecated, removal in 0.3)holomorphic=Falseflag to enable 2N×2N real-block system for non-holomorphic circuits (required for EO power modulation)z0in AC sweep (scalar or array)renormalize(S, z0_from, z0_to)function for frequency-dependent impedance transformationassemble_gc_complex_2n()for 4-block G/C extraction in complex circuitsTest plan
🤖 Generated with Claude Code
https://claude.ai/code/session_01Ti1GWaiLy8ndzJ5maMVYYk