open-problem

Status: partly resolved (2026-11-10). The defects below are fixed on the refactor branch, each with a test. What was found and not changed is listed at the end. Related: LP-023 Findings of the external reviews of 2026-11-04, Refactor 2026-11.

Sources: a review of dev-danielb at 28b1a315 by Claude (2026-11-09/10), which ran every suite and the scripts it cites; the full text is outside the repository (Lenticulum.jl-Review-2026-11-09/); code and tests as named in the tables.

Problem. The review read LenticulumCore and Mycelium in full, the Gaussian part of Lenticulum, implicit.jl and solve.jl, searched the rest, and rederived the mathematics behind the three paths (Refactor 2026-11). About half of the source was read; the three packages off the paths (ImplicitLayers, AdversarialFactors, AlgebraicFactors) were not reviewed.

Fixed

#findingwhat the code didfixtest
1a partly flat message into a GaussianFactora Kalman smoother with a two-dimensional state and a one-dimensional observation threw belief_mean needs a proper belief, or, when rounding let the Cholesky factorisation of the rank-deficient precision through, returned wrong marginals on a tree without an error (for the mean of was off by 0.049, the covariance by 0.019)the backward message is declared improper when the coefficient is rank-deficient; a partly flat input goes through the flat-direction limit of the constraint factors (_flat_message); the same for NonlinearFactorlib/Lenticulum.jl/test/gaussian_oracle.jl: 40 random trees with rectangular coefficients and the position-only smoother against the dense joint (means, covariances, free energy, gradient)
2freeze paths_trainable_indices(ps, (:second,)) froze coordinates from 1 when second began with an empty entry (a Lux container whose first layer has no parameters): the wrong parameters were frozen and the named ones trainedthe start of the range is accumulated while walking the pathLenticulumCore test/runtests.jl, against a brute-force mask on nested trees with empty entries
3composition of two GaussianFactorsthe collapsed factor’s noise failed the constructor’s exact issymmetric for 186 of 200 random parameter draws with a non-diagonal the constructor accepts symmetry up to rounding and uses the symmetric partgaussian_oracle.jl
4dampingdamp of two Gaussian messages dropped the improper declarationkept when both sides carry it (CPU and device methods)Mycelium test/runtests.jl
5parameter_step against free_energy_gradientfor exact or pathwise edges without a schedule, the second returned the envelope gradient (LP-021, option 3) and the first threwone rule, _uses_coupled_driver; ad defaults to nothing in bothgaussian_oracle.jl
6dimensionsa channel and its variable were never compared; a DataFactor of length 1 on a two-dimensional variable was broadcastvalidate refuses a mismatch where both sides declare a dimensiongaussian_oracle.jl
7element typesscalarise of a graded energy and _energy_vector promoted Float32 to Float64fixed in LenticulumCore. Mycelium’s graph total still accumulates in Float64 (open, below)LenticulumCore test/runtests.jl
8accelerated EMfit_em caught every exception of an extrapolated step, interrupts and method errors includedonly numerical failures are a rejected stepcovered by em.jl
9coupled gradient_graph_scalar evaluated the entropy of variables whose counting number is 0, which the forward free energy skips on purposeuses _variable_termcovered by lp021.jl
10two clamps to one valuecombine(DiracBelief(0.1 + 0.2), DiracBelief(0.3)) threw “contradictory hard clamps”equal up to rounding is equalMycelium test/runtests.jl
11Polarity== without hashhash addedLenticulumCore test/runtests.jl
12docstrings on the wrong object⊕, MessageStore, mean_field_message, variable_degree; infer! documented three return values and has fourmoved or correctedMycelium and LenticulumCore test/runtests.jl
13Unscentedthe default α = 1e-3 placed the sigma points a thousandth of a standard deviation outdefault α = 1 (nonlinear)the existing check on
14uncalibrated beliefsa DiffusionFactor with message = :gaussian or :mixture and the default λ = 1 sends covariances off by density_lambda (about 84 at the default levels)a warning, oncenone
15packagingDispatchDoctor was a dependency of five packages and used by none; no package bounded LuxCoreremoved; [compat] entries added

Corrected in the notes

notewhat was wrong
Implicit Diffusion Learners §5the smoothing-bias formula holds for noise-free nodes; nodes with noise average the logarithm over their shifts and smooth a second time, so the default nodes have about twice the bias (measured: lib/VariationalDiffusion.jl/examples/ring_bias.jl)
Universal Approximation for Implicit Diffusion Learners(R1) asked more than Lemma 1’s constant controls; restated
Acausal Composition is a Hypergraph Categorycombine(b, b) is not ; the special axiom holds for potentials, and what fails is representing a copied state by marginals
Channels and Polaritycompact closure lives among unnormalised kernels
Bethe Free Energyrewritten: the node rules that came after it, and the condition under which a tree is exact
The Linear Gaussian Chain”the smoother comes for free” needed the vector case
README.mdcounts, the package table, and a statement that conditional sampling is not implemented (it is: implicit_sample)

Found and not changed

findingwherewhy not
the cost of a sweep and of the free energy grows quadratically with the number of parameter keys (7.7, 25.8, 102 ms for 601, 1 201, 2 401 factors)passing.jl (_setsubtree), free_energy.jlthe backlog’s integer-slot store. Partly fixed 2026-11-11: the main cost of the free energy was _check_tied_state (O(keys × factors)), not _setsubtree; it is linear now and an unchanged state is not written back, so the free energy is 1.5–2.3 times faster (more for larger graphs) and a sweep unchanged, both still quadratic through the lookups by name (lib/Mycelium.jl/src/passing.md, “The state tree”)
excluded_marginal recombines the other messages for each edge: quadratic in a variable’s degreemessages.jlneeds prefix products or division; not a correctness problem
the graph free energy accumulates in Float64 whatever the element typefree_energy.jltouches every energy path; wants the element-type test first
convergence of Gaussian messages is an absolute bound on natural parametersgaussian.jl (Mycelium)a change of behaviour for every schedule; criterion = :marginals is the way round today
AndersonSchedule writes per edge, which unblocks a plate’s rowsanderson.jlby reading; not run
a mixture into a Gaussian or constraint factor is collapsed to its moments, after which a tree is not exactby designstated in Bethe Free Energy; cluster nodes restore exactness
root-solver tolerances of ImplicitLayers are absolute and written for Float64lib/ImplicitLayers.jl/src/solve.jlthe package is off the paths and not yet reviewed

Resolution. The numbered defects are closed by their tests. The open rows are in the Feature Backlog or belong to the shared solver layer proposed in Refactor 2026-11.