GDL - Typo fixes

* typo corrected

* second typo corrected

Author: psteinb

docs/tutorial_notebooks/DL2/Geometric_deep_learning/tutorial2_steerable_cnns.ipynb

@@ -537,7 +537,7 @@
     "\\rho(g) = \\rho(g \\cdot e) = \\rho(g) \\rho(e)\n",
     "$$\n",
     "\n",
-    "Because $\\rho(g)$ is invertible, we can left multiply by $\\rho(g)^{-1}$ to find that $\\rho(e)$ is the identify.\n",
+    "Because $\\rho(g)$ is invertible, we can left multiply by $\\rho(g)^{-1}$ to find that $\\rho(e)$ is the identity.\n",
     "\n",
     "Second question.\n",
     "Note that\n",
@@ -1133,7 +1133,7 @@
    "source": [
     "#### Irreducible Representations (or *Irreps*)\n",
     "\n",
-    "Under minor conditions, any representation can be decomposed in this way, that is, any representation $\\rho$ of a compact group $G$ can be written as a *direc sum* of a number of smaller representations, up to a *change of basis*.\n",
+    "Under minor conditions, any representation can be decomposed in this way, that is, any representation $\\rho$ of a compact group $G$ can be written as a *direct sum* of a number of smaller representations, up to a *change of basis*.\n",
     "These \"smaller representations\" can not be decomposed further and play a very important role in the theory of group representations and steerable CNNs and are called **irreducible representations**, or simply **irreps**.\n",
     "\n",
     "The set of *irreducible representations* of a group $G$ is generally denoted as $\\hat{G}$.\n",