Merge branch 'main' into maint-software

Author: mmcky

.github/workflows/ci.yml

@@ -1,5 +1,7 @@
-name: Build HTML [using jupyter-book]
-on: [push]
+name: CI - Build HTML [using jupyter-book]
+on:
+  pull_request:
+  workflow_dispatch:
 jobs:
   tests:
     runs-on: ubuntu-latest
@@ -77,7 +79,7 @@ jobs:
         uses: nwtgck/actions-netlify@v3
         with:
           publish-dir: '_build/html/'
-          production-branch: master
+          production-branch: main
           github-token: ${{ secrets.GITHUB_TOKEN }}
           deploy-message: "Preview Deploy from GitHub Actions"
         env:

.github/workflows/linkcheck.yml

@@ -1,8 +1,8 @@
 name: Link Checker [Anaconda, Linux]
 on:
   schedule:
-    # UTC 23:00 is early morning in Australia (9am)
-    - cron:  '0 23 * * *'
+    # UTC 23:00 is early morning in Australia (9am) every Sunday
+    - cron:  '0 23 * * 0'
   workflow_dispatch:
 jobs:
   link-checking:
@@ -21,7 +21,7 @@ jobs:
         uses: lycheeverse/lychee-action@v2
         with:
           fail: false
-          args: --accept 403,503 *.html
+          args: --accept 403,503 **/*.html
       - name: Create Issue From File
         if: steps.lychee.outputs.exit_code != 0
         uses: peter-evans/create-issue-from-file@v5

lectures/BCG_complete_mkts.md

@@ -985,7 +985,7 @@ class BCG_complete_markets:
         weights = self.weights
         integ = lambda 𝜖: np.exp(𝜖) * fk * q(𝜖, k)
 
-        return -k + np.sum(weights * integ(self.points_integral)) / np.sqrt(np.pi)
+        return -k + (weights @ integ(self.points_integral)) / np.sqrt(np.pi)
 
     #=========== Optimal c ===========#
     # Function: Compute optimal consumption choices c
@@ -1007,10 +1007,10 @@ class BCG_complete_markets:
         fk = self.f(k)
 
         c1 = lambda 𝜖: (w1(𝜖) + np.exp(𝜖)*fk)*q(𝜖,k)
-        denom = np.sum(weights * c1(self.points_integral)) / np.sqrt(np.pi) + (w0 - k)
+        denom = (weights @ c1(self.points_integral)) / np.sqrt(np.pi) + (w0 - k)
 
         w11q = lambda 𝜖: w11(𝜖)*q(𝜖,k)
-        num = w10 + 𝜃10 * V(k) + np.sum(weights * w11q(self.points_integral)) / np.sqrt(np.pi)
+        num = w10 + 𝜃10 * V(k) + (weights @ w11q(self.points_integral)) / np.sqrt(np.pi)
 
         𝜂 = num / denom
 
@@ -1048,7 +1048,7 @@ def k_foc_factory(model):
 
     @njit
     def k_foc(k, 𝜒1, 𝜒2):
-        int_k = np.sum(weights * integrand(points_integral, 𝜒1, 𝜒2, k=k)) / np.sqrt(np.pi)
+        int_k = (weights @ integrand(points_integral, 𝜒1, 𝜒2, k=k)) / np.sqrt(np.pi)
 
         mul = 𝛽 * 𝛼 * A * k ** (𝛼 - 1) / ((w0 - k) ** (-𝜓))
         val = mul * int_k - 1

lectures/arellano.md

@@ -432,7 +432,7 @@ def T_d(y_idx, v_c, v_d, params, arrays):
 
     current_utility = u(def_y[y_idx], γ)
     v = np.maximum(v_c[B0_idx, :], v_d)
-    cont_value = np.sum((θ * v + (1 - θ) * v_d) * P[y_idx, :])
+    cont_value = (θ * v + (1 - θ) * v_d) @ P[y_idx, :]
 
     return current_utility + β * cont_value
 
@@ -457,7 +457,7 @@ def T_c(B_idx, y_idx, v_c, v_d, q, params, arrays):
         c = y + B - q[Bp_idx, y_idx] * Bp
         if c > 0:
             v = np.maximum(v_c[Bp_idx, :], v_d)
-            val = u(c, γ) + β * np.sum(v * P[y_idx, :])
+            val = u(c, γ) + β * (v @ P[y_idx, :])
             if val > current_max:
                 current_max = val
                 Bp_star_idx = Bp_idx

lectures/calvo_abreu.md

@@ -555,7 +555,7 @@ def abreu_plan(clq, T=1000, T_A=10, μ_bar=0.1, T_Plot=20):
     # Calculate implied stick θ series
     discount = (clq.α / (1 + clq.α)) ** np.arange(T)
     clq.θ_A = np.array([
-        1 / (clq.α + 1) * np.sum(clq.μ_A[t:] * discount[:T - t])
+        1 / (clq.α + 1) * (clq.μ_A[t:] @ discount[:T - t])
         for t in range(T)
     ])
 

lectures/calvo_machine_learn.md

@@ -567,7 +567,7 @@ def compute_θ(μ, α=1):
     
     # Compute the weighted sums for all t
     weighted_sums = jnp.array(
-        [jnp.sum(λ_powers[:T-t] * μ[t:T]) for t in range(T)])
+        [(λ_powers[:T-t] @ μ[t:T]) for t in range(T)])
     
     # Compute θ values except for the last element
     θ = (1 - λ) * weighted_sums + λ**(T - jnp.arange(T)) * μbar
@@ -595,7 +595,7 @@ def compute_V(μ, β, c, α=1, u0=1, u1=0.5, u2=3):
     t = np.arange(T)
     
     # Compute sum except for the last element
-    V_sum = np.sum(β**t * (h0 + h1 * θ[:T] + h2 * θ[:T]**2 - 0.5 * c * μ[:T]**2))
+    V_sum = (β**t) @ (h0 + h1 * θ[:T] + h2 * θ[:T]**2 - 0.5 * c * μ[:T]**2)
     
     # Compute the final term
     V_final = (β**T / (1 - β)) * (h0 + h1 * μ[-1] + h2 * μ[-1]**2 - 0.5 * c * μ[-1]**2)
@@ -931,7 +931,7 @@ def compute_J(μ, β, c, α=1, u0=1, u1=0.5, u2=3):
                        (β**T/(1-β))])
     
     θ = B @ μ
-    βθ_sum = jnp.sum((β_vec * h1) * θ)
+    βθ_sum = (β_vec * h1) @ θ
     βθ_square_sum = β_vec * h2 * θ.T @ θ
     βμ_square_sum = 0.5 * c * β_vec * μ.T @ μ
     

lectures/dyn_stack.md

@@ -1043,7 +1043,7 @@ We'll compute it two ways and get the same answer.
 In addition to being a useful check on the accuracy of our coding, computing things in these two ways helps us think about the structure of the problem.  
 
 ```{code-cell} python3
-v_leader_forward = np.sum(βs * π_leader)
+v_leader_forward = βs @ π_leader
 v_leader_direct = -yt[:, 0].T @ P @ yt[:, 0]
 
 # Display values
@@ -1380,8 +1380,8 @@ u2 = (- F2 @ z).flatten()
 π_1 = p * q1 - γ * (u1) ** 2
 π_2 = p * q2 - γ * (u2) ** 2
 
-v1_forward = np.sum(βs * π_1)
-v2_forward = np.sum(βs * π_2)
+v1_forward = βs @ π_1
+v2_forward = βs @ π_2
 
 v1_direct = (- z[:, 0].T @ P1 @ z[:, 0])
 v2_direct = (- z[:, 0].T @ P2 @ z[:, 0])

lectures/five_preferences.md

@@ -100,7 +100,7 @@ def ent(π, π_hat):
     Compute the relative entropy of a probability vector `π_hat` with respect to `π`. JIT-compiled using Numba.
     
     """
-    ent_val = -np.sum(π_hat * (np.log(π) - np.log(π_hat)))
+    ent_val = -π_hat @ (np.log(π) - np.log(π_hat))
     
     return ent_val
 
@@ -115,7 +115,7 @@ def T_θ_factory(θ, π):
         Risk-sensitivity operator of Jacobson (1973) and Whittle (1981) taking a function `u` as argument.
         
         """
-        return lambda c: -θ * np.log(np.sum(π * np.exp(-u(c) / θ)))
+        return lambda c: -θ * np.log(π @ np.exp(-u(c) / θ))
     
     return T_θ
 
@@ -128,7 +128,7 @@ def compute_change_measure(u, c, θ, π):
     """
     
     m_unnormalized = np.exp(-u(c) / θ)
-    m =  m_unnormalized / (π * m_unnormalized).sum()
+    m =  m_unnormalized / (π @ m_unnormalized)
     return m
 
 

lectures/orth_proj.md

@@ -790,7 +790,7 @@ def gram_schmidt(X):
 
     # The first col of U is just the normalized first col of X
     v1 = X[:,0]
-    U[:, 0] = v1 / np.sqrt(np.sum(v1 * v1))
+    U[:, 0] = v1 / np.sqrt(v1 @ v1)
 
     for i in range(1, k):
         # Set up
@@ -802,7 +802,7 @@ def gram_schmidt(X):
         u = M @ b
 
         # Normalize
-        U[:, i] = u / np.sqrt(np.sum(u * u))
+        U[:, i] = u / np.sqrt(u @ u)
 
     return U
 ```