pymc-devs · twiecki · Nov 8, 2025 · Nov 8, 2025 · Nov 8, 2025 · Nov 8, 2025
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -12,7 +12,7 @@ repos:
     - id: nbqa-isort
       additional_dependencies: [isort==5.6.4]
     - id: nbqa-pyupgrade
-      additional_dependencies: [pyupgrade==2.7.4]
+      additional_dependencies: [pyupgrade==3.19.0]
       args: [--py37-plus]
 - repo: https://github.com/MarcoGorelli/madforhooks
   rev: 0.4.1

diff --git a/examples/case_studies/factor_analysis.ipynb b/examples/case_studies/factor_analysis.ipynb
@@ -307,7 +307,7 @@
    "source": [
     "for i in trace.posterior.chain.values:\n",
     "    samples = trace.posterior[\"W\"].sel(chain=i, observed_columns=3, latent_columns=1)\n",
-    "    plt.plot(samples, label=\"Chain {}\".format(i + 1))\n",
+    "    plt.plot(samples, label=f\"Chain {i + 1}\")\n",
     "    plt.axhline(samples.mean(), color=f\"C{i}\")\n",
     "plt.legend(ncol=4, loc=\"upper center\", fontsize=12, frameon=True), plt.xlabel(\"Sample\");"
    ]
@@ -484,7 +484,7 @@
     "\n",
     "for i in range(4):\n",
     "    samples = trace.posterior[\"W\"].sel(chain=i, observed_columns=3, latent_columns=1)\n",
-    "    plt.plot(samples, label=\"Chain {}\".format(i + 1))\n",
+    "    plt.plot(samples, label=f\"Chain {i + 1}\")\n",
     "\n",
     "plt.legend(ncol=4, loc=\"lower center\", fontsize=8), plt.xlabel(\"Sample\");"
    ]
@@ -704,7 +704,7 @@
     "\n",
     "ax = az.plot_kde(\n",
     "    trace.posterior[\"W\"].sel(**col_selection).values,\n",
-    "    label=\"MCMC posterior for the explicit model\".format(0),\n",
+    "    label=f\"MCMC posterior for the explicit model\",\n",
     "    plot_kwargs={\"color\": f\"C{1}\"},\n",
     ")\n",
     "\n",

diff --git a/examples/case_studies/factor_analysis.myst.md b/examples/case_studies/factor_analysis.myst.md
@@ -130,7 +130,7 @@ At this point, there are already several warnings regarding failed convergence c
 ```{code-cell} ipython3
 for i in trace.posterior.chain.values:
     samples = trace.posterior["W"].sel(chain=i, observed_columns=3, latent_columns=1)
-    plt.plot(samples, label="Chain {}".format(i + 1))
+    plt.plot(samples, label=f"Chain {i + 1}")
     plt.axhline(samples.mean(), color=f"C{i}")
 plt.legend(ncol=4, loc="upper center", fontsize=12, frameon=True), plt.xlabel("Sample");
 ```
@@ -198,7 +198,7 @@ with pm.Model(coords=coords) as PPCA_identified:
 
 for i in range(4):
     samples = trace.posterior["W"].sel(chain=i, observed_columns=3, latent_columns=1)
-    plt.plot(samples, label="Chain {}".format(i + 1))
+    plt.plot(samples, label=f"Chain {i + 1}")
 
 plt.legend(ncol=4, loc="lower center", fontsize=8), plt.xlabel("Sample");
 ```
@@ -253,7 +253,7 @@ col_selection = dict(observed_columns=3, latent_columns=1)
 
 ax = az.plot_kde(
     trace.posterior["W"].sel(**col_selection).values,
-    label="MCMC posterior for the explicit model".format(0),
+    label=f"MCMC posterior for the explicit model",
     plot_kwargs={"color": f"C{1}"},
 )
 

diff --git a/examples/case_studies/probabilistic_matrix_factorization.ipynb b/examples/case_studies/probabilistic_matrix_factorization.ipynb
@@ -1135,7 +1135,7 @@
     "    Method = baseline_methods[name]\n",
     "    method = Method(train)\n",
     "    baselines[name] = method.rmse(test)\n",
-    "    print(\"{} RMSE:\\t{:.5f}\".format(method, baselines[name]))"
+    "    print(f\"{method} RMSE:\\t{baselines[name]:.5f}\")"
    ]
   },
   {

diff --git a/examples/case_studies/probabilistic_matrix_factorization.myst.md b/examples/case_studies/probabilistic_matrix_factorization.myst.md
@@ -537,7 +537,7 @@ for name in baseline_methods:
     Method = baseline_methods[name]
     method = Method(train)
     baselines[name] = method.rmse(test)
-    print("{} RMSE:\t{:.5f}".format(method, baselines[name]))
+    print(f"{method} RMSE:\t{baselines[name]:.5f}")
 ```
 
 As expected: the uniform random baseline is the worst by far, the global mean baseline is next best, and the mean of means method is our best baseline. Now let's see how PMF stacks up.

diff --git a/examples/gaussian_processes/GP-Heteroskedastic.ipynb b/examples/gaussian_processes/GP-Heteroskedastic.ipynb
@@ -147,7 +147,7 @@
     "    return ℓ_μ, ℓ_σ\n",
     "\n",
     "\n",
-    "ℓ_μ, ℓ_σ = [stat for stat in get_ℓ_prior(X_)]"
+    "ℓ_μ, ℓ_σ = (stat for stat in get_ℓ_prior(X_))"
    ]
   },
   {
@@ -1036,7 +1036,7 @@
     "    return np.hstack([np.tile(x, (2, 1)), np.vstack([np.zeros(x.shape), np.ones(x.shape)])])\n",
     "\n",
     "\n",
-    "Xu_c, X_obs_c, Xnew_c = [add_coreg_idx(x) for x in [Xu, X_obs, Xnew]]\n",
+    "Xu_c, X_obs_c, Xnew_c = (add_coreg_idx(x) for x in [Xu, X_obs, Xnew])\n",
     "\n",
     "with pm.Model() as model_htsc:\n",
     "    ℓ = pm.InverseGamma(\"ℓ\", mu=ℓ_μ, sigma=ℓ_σ)\n",

diff --git a/examples/gaussian_processes/GP-Heteroskedastic.myst.md b/examples/gaussian_processes/GP-Heteroskedastic.myst.md
@@ -87,7 +87,7 @@ def get_ℓ_prior(points):
     return ℓ_μ, ℓ_σ
 
 
-ℓ_μ, ℓ_σ = [stat for stat in get_ℓ_prior(X_)]
+ℓ_μ, ℓ_σ = (stat for stat in get_ℓ_prior(X_))
 ```
 
 ```{code-cell} ipython3
@@ -388,7 +388,7 @@ def add_coreg_idx(x):
     return np.hstack([np.tile(x, (2, 1)), np.vstack([np.zeros(x.shape), np.ones(x.shape)])])
 
 
-Xu_c, X_obs_c, Xnew_c = [add_coreg_idx(x) for x in [Xu, X_obs, Xnew]]
+Xu_c, X_obs_c, Xnew_c = (add_coreg_idx(x) for x in [Xu, X_obs, Xnew])
 
 with pm.Model() as model_htsc:
     ℓ = pm.InverseGamma("ℓ", mu=ℓ_μ, sigma=ℓ_σ)

diff --git a/examples/generalized_linear_models/GLM-hierarchical-binomial-model.ipynb b/examples/generalized_linear_models/GLM-hierarchical-binomial-model.ipynb
@@ -8,7 +8,7 @@
     ":::{post} Jan 10, 2023\n",
     ":tags: generalized linear model, hierarchical model \n",
     ":category: intermediate\n",
-    ":author: Demetri Pananos, Junpeng Lao, Raúl Maldonado, Farhan Reynaldo\n",
+    ":author: Demetri Pananos, Junpeng Lao, Ra\u00fal Maldonado, Farhan Reynaldo\n",
     ":::"
    ]
   },
@@ -203,7 +203,7 @@
     "ix_z, ix_x = np.unravel_index(np.argmax(surface, axis=None), surface.shape)\n",
     "ax.scatter([X[0, ix_x]], [Z[ix_z, 0]], color=\"red\")\n",
     "\n",
-    "text = r\"$({a},{b})$\".format(a=np.round(X[0, ix_x], 2), b=np.round(Z[ix_z, 0], 2))\n",
+    "text = rf\"$({np.round(X[0, ix_x], 2)},{np.round(Z[ix_z, 0], 2)})$\"\n",
     "\n",
     "ax.annotate(\n",
     "    text,\n",
@@ -602,7 +602,7 @@
        "\n",
        ".xr-section-summary-in + label:before {\n",
        "  display: inline-block;\n",
-       "  content: '►';\n",
+       "  content: '\u25ba';\n",
        "  font-size: 11px;\n",
        "  width: 15px;\n",
        "  text-align: center;\n",
@@ -613,7 +613,7 @@
        "}\n",
        "\n",
        ".xr-section-summary-in:checked + label:before {\n",
-       "  content: '▼';\n",
+       "  content: '\u25bc';\n",
        "}\n",
        "\n",
        ".xr-section-summary-in:checked + label > span {\n",
@@ -892,7 +892,7 @@
     "## Authors\n",
     "\n",
     "* Adapted from chapter 5 of Bayesian Data Analysis 3rd Edition {cite:p}`gelman2013bayesian` by Demetri Pananos and Junpeng Lao ([pymc#3054](https://github.com/pymc-devs/pymc/pull/3054))\n",
-    "* Updated and reexecuted by Raúl Maldonado ([pymc-examples#24](https://github.com/pymc-devs/pymc-examples/pull/24), [pymc-examples#45](https://github.com/pymc-devs/pymc-examples/pull/45) and [pymc-examples#147](https://github.com/pymc-devs/pymc-examples/pull/147))\n",
+    "* Updated and reexecuted by Ra\u00fal Maldonado ([pymc-examples#24](https://github.com/pymc-devs/pymc-examples/pull/24), [pymc-examples#45](https://github.com/pymc-devs/pymc-examples/pull/45) and [pymc-examples#147](https://github.com/pymc-devs/pymc-examples/pull/147))\n",
     "* Updated and reexecuted by Farhan Reynaldo in November 2021 ([pymc-examples#248](https://github.com/pymc-devs/pymc-examples/pull/248))\n",
     "* Rerun with PyMC v5, by Reshama Shaikh, January 2023"
    ]
@@ -969,4 +969,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}
diff --git a/examples/generalized_linear_models/GLM-hierarchical-binomial-model.myst.md b/examples/generalized_linear_models/GLM-hierarchical-binomial-model.myst.md
@@ -159,7 +159,7 @@ ax.set_ylabel(r"$\log(\alpha+\beta)$", fontsize=16)
 ix_z, ix_x = np.unravel_index(np.argmax(surface, axis=None), surface.shape)
 ax.scatter([X[0, ix_x]], [Z[ix_z, 0]], color="red")
 
-text = r"$({a},{b})$".format(a=np.round(X[0, ix_x], 2), b=np.round(Z[ix_z, 0], 2))
+text = rf"$({np.round(X[0, ix_x], 2)},{np.round(Z[ix_z, 0], 2)})$"
 
 ax.annotate(
     text,

diff --git a/examples/generalized_linear_models/GLM-robust-with-outlier-detection.ipynb b/examples/generalized_linear_models/GLM-robust-with-outlier-detection.ipynb
@@ -245,7 +245,7 @@
     "    columns=[\"id\", \"x\", \"y\", \"sigma_y\", \"sigma_x\", \"rho_xy\"],\n",
     ")\n",
     "\n",
-    "dfhogg[\"id\"] = dfhogg[\"id\"].apply(lambda x: \"p{}\".format(int(x)))\n",
+    "dfhogg[\"id\"] = dfhogg[\"id\"].apply(lambda x: f\"p{int(x)}\")\n",
     "dfhogg.set_index(\"id\", inplace=True)\n",
     "dfhogg.head()"
    ]

diff --git a/examples/generalized_linear_models/GLM-robust-with-outlier-detection.myst.md b/examples/generalized_linear_models/GLM-robust-with-outlier-detection.myst.md
@@ -113,7 +113,7 @@ dfhogg = pd.DataFrame(
     columns=["id", "x", "y", "sigma_y", "sigma_x", "rho_xy"],
 )
 
-dfhogg["id"] = dfhogg["id"].apply(lambda x: "p{}".format(int(x)))
+dfhogg["id"] = dfhogg["id"].apply(lambda x: f"p{int(x)}")
 dfhogg.set_index("id", inplace=True)
 dfhogg.head()
 ```

diff --git a/examples/ode_models/ODE_with_manual_gradients.ipynb b/examples/ode_models/ODE_with_manual_gradients.ipynb
@@ -100,7 +100,7 @@
     "        self._y0 = y0\n",
     "\n",
     "    def simulate(self, parameters, times):\n",
-    "        alpha, beta, gamma, delta, Xt0, Yt0 = [x for x in parameters]\n",
+    "        alpha, beta, gamma, delta, Xt0, Yt0 = (x for x in parameters)\n",
     "\n",
     "        def rhs(y, t, p):\n",
     "            X, Y = y\n",
@@ -165,7 +165,7 @@
     "        return self._simulate(parameters, times, True)\n",
     "\n",
     "    def _simulate(self, parameters, times, sensitivities):\n",
-    "        alpha, beta, gamma, delta, Xt0, Yt0 = [x for x in parameters]\n",
+    "        alpha, beta, gamma, delta, Xt0, Yt0 = (x for x in parameters)\n",
     "\n",
     "        def r(y, t, p):\n",
     "            X, Y = y\n",
@@ -951,7 +951,7 @@
     "        self._times = times\n",
     "\n",
     "    def _simulate(self, parameters, times):\n",
-    "        a, b, c = [float(x) for x in parameters]\n",
+    "        a, b, c = (float(x) for x in parameters)\n",
     "\n",
     "        def rhs(y, t, p):\n",
     "            V, R = y\n",

diff --git a/examples/ode_models/ODE_with_manual_gradients.myst.md b/examples/ode_models/ODE_with_manual_gradients.myst.md
@@ -78,7 +78,7 @@ class LotkaVolterraModel:
         self._y0 = y0
 
     def simulate(self, parameters, times):
-        alpha, beta, gamma, delta, Xt0, Yt0 = [x for x in parameters]
+        alpha, beta, gamma, delta, Xt0, Yt0 = (x for x in parameters)
 
         def rhs(y, t, p):
             X, Y = y
@@ -133,7 +133,7 @@ class LotkaVolterraModel:
         return self._simulate(parameters, times, True)
 
     def _simulate(self, parameters, times, sensitivities):
-        alpha, beta, gamma, delta, Xt0, Yt0 = [x for x in parameters]
+        alpha, beta, gamma, delta, Xt0, Yt0 = (x for x in parameters)
 
         def r(y, t, p):
             X, Y = y
@@ -453,7 +453,7 @@ class FitzhughNagumoModel:
         self._times = times
 
     def _simulate(self, parameters, times):
-        a, b, c = [float(x) for x in parameters]
+        a, b, c = (float(x) for x in parameters)
 
         def rhs(y, t, p):
             V, R = y