docs: append full ACL-format References sections to remaining doc pages

aaronstevenwhite · aaronstevenwhite · commit 9c6e0e461ae4 · 2026-05-22T14:30:45.000-04:00
For every doc page that cites at least one paper through a
DOI / arXiv / ACL / JSTOR / Cambridge link but lacked a
References section, append a ## References section with the
canonical author-year-title-venue-volume-pages entry pulled
directly from the publisher's page during this audit.

Citation strings are drawn from a verified mapping built across
the session covering ~90 unique sources (Giry 1982; Kock 1972;
Fritz 2020; Cho &amp; Jacobs 2019; Joyal Street Verity 1996; Booth
&amp; Thompson 1973; Robbins &amp; Monro 1951; Varin Reid &amp; Firth 2011;
Lafferty McCallum Pereira 2001; Lambek 1958; Carlin Gelfand Smith
1992; Lambert 1992; Spearman 1904 — at the corrected JSTOR ID
1412107; Nelder &amp; Wedderburn 1972; Tierney &amp; Kadane 1986; Casella
&amp; Robert 1996; Carvalho Polson Scott 2010; Kalman 1960; Rabiner
1989; Tipping &amp; Bishop 1999; Stephens 2000; Hochreiter &amp;
Schmidhuber 1997; MacKay 1992; Rauch Tung Striebel 1965; Kim
Shephard Chib 1998; Krishnan Shalit Sontag 2017; Blei Ng Jordan
2003; Hoffman Blei Wang Paisley 2013; Hoffman &amp; Gelman 2014;
Lewandowski Kurowicka Joe 2009; Bartholomew Knott Moustaki 2011;
Klein &amp; Moeschberger 2003; Ferrari &amp; Cribari-Neto 2004;
Papaspiliopoulos Roberts Sköld 2007; Yao et al. 2018; Vehtari
et al. 2021; Neal 2003; Neal 2011; Gelman &amp; Shirley 2011; Gelman
et al. 2013; Vehtari Gelman Gabry 2017; Vehtari Simpson Gelman
Yao Gabry 2024; the various arXiv preprints for normalizing
flows / HMC / annealing / GRU / LSTM / Transformer / BERT;
Montague 1973; Moggi 1991; Hughes 2000; Plotkin &amp; Power 2003;
Bauer &amp; Pretnar 2015; Lawvere 1973; Cartmell 1986; Hutton 1999;
Shieber Schabes Pereira 1995; Pereira &amp; Warren 1983; Klein &amp;
Manning 2001; Knuth 1977; Nederhof 2003; McAllester 2002;
Goodman 1999; Huang &amp; Chiang 2005; Bumford &amp; Charlow 2026).

14 files updated; combined with the earlier round, every doc
page under docs/guides/, docs/examples/, and docs/tutorials/qvr/
that uses a bibliographic citation now has a References section.
diff --git a/docs/examples/ar1.md b/docs/examples/ar1.md
@@ -108,3 +108,8 @@ for name, samples in result.samples.items():
 ## Categorical Perspective
 
 The model is a Kleisli morphism in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category whose denotation is the standard AR(1) joint $p(\alpha, \phi, \sigma) \prod_t \mathcal{N}(y_t \mid \alpha + \phi y_{t-1}, \sigma)$. The `Step`-indexed plate is the [right Kan extension](https://ncatlab.org/nlab/show/Kan+extension) of the per-step Normal kernel along the trivial projection $\mathrm{Step} \to \mathbf{1}$.
+
+
+## References
+
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/bidirectional-rnn-lm.md b/docs/examples/bidirectional-rnn-lm.md
@@ -159,3 +159,9 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The model denotes a Kleisli morphism $\mathrm{Token} \to \mathcal{G}(\mathrm{Token})$ assembled by `fan`-composing two independent scan-folds and following with a merge. The `fan` combinator is the diagonal-pair construction $f \times g \circ \Delta$ in the Kleisli category that delivers a common input to both branches; `combine` is the merge $\mathrm{Hidden}^2 \to \mathcal{G}(\mathrm{Combined})$ that pulls the bilinear pairing back onto a single object. The Categorical head closes with the masked-token likelihood as a sub-probability kernel.
+
+
+## References
+
+- Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In *Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT)*, pages 4171–4186. ACL.
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/changepoint.md b/docs/examples/changepoint.md
@@ -113,3 +113,10 @@ for name, samples in result.samples.items():
 ## Categorical Perspective
 
 The model is a Kleisli morphism in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category whose denotation is the joint $p(\tau, r_1, r_2) \prod_t \mathrm{Poisson}(y_t \mid r(t; \tau, r_1, r_2))$. The soft indicator is the [reparameterisation gradient](https://doi.org/10.48550/arXiv.1312.6114)-friendly relaxation of an indicator function; in the hard-change-point variant, the discrete `tau_idx` is integrated out by `marginalize`, realizing the right Kan extension along the trivial projection from the time axis to a singleton.
+
+
+## References
+
+- Bradley P. Carlin, Alan E. Gelfand, and Adrian F. M. Smith. 1992. Hierarchical Bayesian analysis of changepoint problems. *Journal of the Royal Statistical Society Series C: Applied Statistics*, 41(2):389–405.
+- Diederik P. Kingma and Max Welling. 2013. Auto-Encoding Variational Bayes. arXiv preprint arXiv:1312.6114.
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/deep-markov.md b/docs/examples/deep-markov.md
@@ -140,3 +140,9 @@ flowchart LR
     h_emit["h_emit"] --> emit_mlp_2["emit_mlp_2"]
     emit_mlp_2["emit_mlp_2"] --> o_t["o_t"]
 ```
+
+
+## References
+
+- Diederik P. Kingma and Max Welling. 2013. Auto-Encoding Variational Bayes. arXiv preprint arXiv:1312.6114.
+- Rahul G. Krishnan, Uri Shalit, and David Sontag. 2017. Structured inference networks for nonlinear state space models. In *Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI '17)*, pages 2101–2109. AAAI Press.
diff --git a/docs/examples/gru-lm.md b/docs/examples/gru-lm.md
@@ -175,3 +175,9 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The GRU cell is a Kleisli morphism $\mathrm{Embedded} \times \mathrm{Hidden} \to \mathcal{G}(\mathrm{Hidden})$ in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category; `scan(gru_cell)` is its iterated composition along the sequence. The Categorical head and `observe` step close the composite into the LM likelihood by accumulating per-batch categorical log-probabilities.
+
+
+## References
+
+- Kyunghyun Cho, Bart van Merriënboer, Caglar Gulcehre, Dzmitry Bahdanau, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. 2014. Learning phrase representations using RNN encoder–decoder for statistical machine translation. In *Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)*, pages 1724–1734, Doha, Qatar. Association for Computational Linguistics.
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/hmm.md b/docs/examples/hmm.md
@@ -148,3 +148,8 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 The forward algorithm and the Viterbi algorithm are the same composed morphism evaluated in different [algebras](https://ncatlab.org/nlab/show/algebra): under product, composition multiplies probabilities and summation marginalizes; under tropical, composition adds log-probabilities and summation maximizes. Quivers makes this explicit: switching `algebra` changes the V-enriched composition rule without touching the program text.
 
 The row-wise Dirichlet prior is the standard conjugate prior for a categorical kernel; declaring it via `over cod iid over dom` resolves the axis-role ambiguity that distinguishes a flat Dirichlet on $|State|\cdot|State|$ entries (wrong: not row-stochastic) from $|State|$ independent simplex draws (right). The categorical reading: each row of $T$ is a fiber of the dependent kernel $\prod_{c \,:\, \mathrm{State}} \mathcal{G}(\mathrm{State})$, so the prior factors as a product of independent simplex priors.
+
+
+## References
+
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/irt-2pl.md b/docs/examples/irt-2pl.md
@@ -117,3 +117,8 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The 2PL is a Kleisli morphism over the Person + Item plate structure in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category. The plate-gather operations are pullbacks of indexed kernels along the response-row index maps `person_idx : Resp -> Person` and `item_idx : Resp -> Item`.
+
+
+## References
+
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/lda.md b/docs/examples/lda.md
@@ -131,3 +131,8 @@ The discrete per-word topic $z : \mathsf{Topic}$ is integrated out by the [pushf
 ## See Also
 
 - [Bayesian Gaussian Mixture Model](mixture-model.md) for a simpler grouped `marginalize` over a discrete latent.
+
+
+## References
+
+- David M. Blei, Andrew Y. Ng, and Michael I. Jordan. 2003. Latent Dirichlet allocation. *Journal of Machine Learning Research*, 3:993–1022.
diff --git a/docs/examples/linear-gaussian-ssm.md b/docs/examples/linear-gaussian-ssm.md
@@ -137,3 +137,10 @@ flowchart LR
     u_2["u_2"] --> transition_cell_2["transition_cell_2"]
     transition_cell_2["transition_cell_2"] --> s_2["s_2"]
 ```
+
+
+## References
+
+- Herbert E. Rauch, F. Tung, and Charles T. Striebel. 1965. Maximum likelihood estimates of linear dynamic systems. *AIAA Journal*, 3(8):1445–1450.
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
+- Rudolf E. Kalman. 1960. A new approach to linear filtering and prediction problems. *Journal of Basic Engineering*, 82(1):35–45.
diff --git a/docs/examples/lstm-lm.md b/docs/examples/lstm-lm.md
@@ -184,3 +184,9 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The cell program denotes a Kleisli morphism $\mathrm{Embedded} \times \mathrm{Hidden} \to \mathcal{G}(\mathrm{Hidden})$ in the Kleisli category of the [Giry monad](https://doi.org/10.1007/BFb0092872); `scan(lstm_cell)` is its iterated composition over the sequence. The Categorical head closes the composite with a finite-set codomain, and `observe next_token` accumulates per-batch categorical log-likelihood through a [right Kan extension](https://ncatlab.org/nlab/show/Kan+extension).
+
+
+## References
+
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
+- Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. *Neural Computation*, 9(8):1735–1780.
diff --git a/docs/examples/transformer-lm.md b/docs/examples/transformer-lm.md
@@ -162,3 +162,9 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The model denotes a Kleisli morphism $\mathrm{Token} \to \mathcal{G}(\mathrm{Token})$ in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category, assembled by composition of replicated heads, an output projection, residual mixers, and a two-stage feed-forward block. [`stack`](../guides/dsl-declarations.md#stack-independent-multi-layer) is independent multi-layer deep composition; [`fan`](../guides/dsl-declarations.md#fan-out-diagonal-morphism) is the diagonal followed by parallel composition, the categorical realization of multi-head attention. The Categorical head accumulates per-position log-likelihood as a sub-probability kernel.
+
+
+## References
+
+- Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. arXiv preprint arXiv:1706.03762.
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/examples/vanilla-rnn-lm.md b/docs/examples/vanilla-rnn-lm.md
@@ -146,3 +146,8 @@ print(f"divergences: {int(result.divergence_counts.sum())}")
 ## Categorical Perspective
 
 The model is a Kleisli morphism $\mathrm{Token} \to \mathcal{G}(\mathrm{Token})$ in the [Giry monad](https://doi.org/10.1007/BFb0092872)'s Kleisli category. [`scan(cell)`](../guides/dsl-declarations.md#scan-temporal-recurrence) is the recursive [fold](https://ncatlab.org/nlab/show/fold) along the sequence in the Kleisli category: each step composes the previous step's output kernel with the new cell. The closing Categorical head observes the next-token label as a sub-probability kernel in $\mathcal{G}_{\le 1}$.
+
+
+## References
+
+- Michèle Giry. 1982. A categorical approach to probability theory. In Bernhard Banaschewski, editor, *Categorical Aspects of Topology and Analysis*, volume 915 of *Lecture Notes in Mathematics*, pages 68–85. Springer, Berlin, Heidelberg.
diff --git a/docs/guides/continuous-spaces.md b/docs/guides/continuous-spaces.md
@@ -301,3 +301,11 @@ normalization layers, and LU-parameterized linear layers.
 - [Stochastic Morphisms](stochastic.md): the finite-state
   counterpart and the
   [Giry monad](https://ncatlab.org/nlab/show/Giry+monad) substrate.
+
+
+## References
+
+- Conor Durkan, Artur Bekasov, Iain Murray, and George Papamakarios. 2019. Neural spline flows. arXiv preprint arXiv:1906.04032.
+- Diederik P. Kingma, Tim Salimans, Rafal Jozefowicz, Xi Chen, Ilya Sutskever, and Max Welling. 2016. Improving variational inference with inverse autoregressive flow. arXiv preprint arXiv:1606.04934.
+- George Papamakarios, Theo Pavlakou, and Iain Murray. 2017. Masked autoregressive flow for density estimation. arXiv preprint arXiv:1705.07057.
+- Laurent Dinh, Jascha Sohl-Dickstein, and Samy Bengio. 2016. Density estimation using Real NVP. arXiv preprint arXiv:1605.08803.
diff --git a/docs/guides/dsl-overview.md b/docs/guides/dsl-overview.md
@@ -386,3 +386,9 @@ errors surface as `ParseError`.
 - [Contractions](dsl-contractions.md): operadic n-ary contractions,
   type-driven wiring inference, `share`, and the explicit `wiring`
   escape hatch.
+
+
+## References
+
+- Carlos M. Carvalho, Nicholas G. Polson, and James G. Scott. 2010. The horseshoe estimator for sparse signals. *Biometrika*, 97(2):465–480.
+- Daniel Lewandowski, Dorota Kurowicka, and Harry Joe. 2009. Generating random correlation matrices based on vines and extended onion method. *Journal of Multivariate Analysis*, 100(9):1989–2001.
