replaced L1 with Linf, because I misnamed the norm, embarrassingly.

Author: pavelkomarov

notebooks/chebyshev.ipynb

@@ -678,7 +678,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.13.1"
+   "version": "3.13.2"
   }
  },
  "nbformat": 4,

notebooks/chebyshev_antiderivatives.ipynb

@@ -98,8 +98,8 @@
      "text": [
       "mean iy_n -0.007520096393034718\n",
       "mean iy 0.04580123098784975\n",
-      "L2 3.1176992465971132e-18\n",
-      "L1 6.675665797928332e-09\n"
+      "L2 3.1176986577396767e-18\n",
+      "L∞ 6.675664909749912e-09\n"
      ]
     },
     {
@@ -126,7 +126,7 @@
     "print(\"mean iy\", np.mean(iy))\n",
     "y_n_recalc = cheb_deriv(iy_n, x_n, 1)\n",
     "print(\"L2\", np.mean((y_n - y_n_recalc)**2))\n",
-    "print(\"L1\", np.max(np.abs(y_n - y_n_recalc)))\n",
+    "print(\"L∞\", np.max(np.abs(y_n - y_n_recalc)))\n",
     "\n",
     "pyplot.plot(x_n, y_n, 'k+')\n",
     "pyplot.plot(x_n, iy_n, 'b+')\n",
@@ -160,7 +160,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.13.1"
+   "version": "3.13.2"
   }
  },
  "nbformat": 4,

notebooks/dct_types_extension_experiments.ipynb

@@ -30,12 +30,12 @@ def test_cheb_deriv_accurate_to_6th(dct_type, x_n):
 						8*np.exp(x_n) * (2035*np.cos(5*x_n) - 828*np.sin(5*x_n))]	# 6th
 	# Things get less accurate for higher derivatives, so check < 10^f(nu)
 	L2_powers = [[-19, -14, -10, -6, -3, 0], [-17, -13, -9, -6, -2, 1]]
-	L1_powers = [[-9, -6, -4, -2, -1, 1], [-8, -6, -4, -2, 0, 1]]
+	Linf_powers = [[-9, -6, -4, -2, -1, 1], [-8, -6, -4, -2, 0, 1]]
 
 	for nu in range(1,7):
 		computed = cheb_deriv(y_n, x_n, nu, dct_type=dct_type) # strangely, this can be slightly different (but close) in CI vs local, despite same package versions
 		assert np.nanmean((analytic_truth[nu-1] - computed)**2) < 10**L2_powers[dct_type-1][nu-1]
-		assert np.nanmax(np.abs(analytic_truth[nu-1] - computed)) < 10**L1_powers[dct_type-1][nu-1]
+		assert np.nanmax(np.abs(analytic_truth[nu-1] - computed)) < 10**Linf_powers[dct_type-1][nu-1]
 
 def test_cheb_endpoints():
 	"""A test of the endpoints code, specifically. Endpoints should be found accurately up to the
@@ -131,7 +131,7 @@ def test_cheb_arbitrary_domains_to_3rd():
 	"""A test that we can take the derivative on domains that aren't the canonical [1, -1]
 	"""
 	L2_powers = [[-10, -6, -2],[-25, -24, -20]] # The function does *much* better on shorter 
-	L1_powers = [[-4, -2, 1],[-14, -11, -9]] # domains with an N this low and y this wobbly.
+	Linf_powers = [[-4, -2, 1],[-14, -11, -9]] # domains with an N this low and y this wobbly.
 
 	for i,(t_0,t_N) in enumerate([(3.5, 0.5), (-4,-5)]):
 		t_n = np.cos(np.arange(N+1) * np.pi / N) * (t_0 - t_N)/2 + (t_0 + t_N)/2
@@ -144,7 +144,7 @@ def test_cheb_arbitrary_domains_to_3rd():
 		for nu in range(1,4):
 			computed = cheb_deriv(y_n, t_n, nu)
 			assert np.mean((analytic_truth[nu-1] - computed)**2) < 10**L2_powers[i][nu-1]
-			assert np.max(np.abs(analytic_truth[nu-1] - computed)) < 10**L1_powers[i][nu-1]
+			assert np.max(np.abs(analytic_truth[nu-1] - computed)) < 10**Linf_powers[i][nu-1]
 
 def test_fourier_arbitrary_domains_to_3rd():
 	"""A test that we can take the derivative on domains that aren't the canonical [0, 2pi)