added problem 5 to pset 3

This commit is contained in:
Steven G. Johnson 2023-03-08 14:43:27 -05:00
parent e24b37216b
commit 10da706709

View File

@ -134,12 +134,30 @@
"@show svdvals(S)[1:6]"
]
},
{
"cell_type": "markdown",
"id": "b9415ba3",
"metadata": {},
"source": [
"## Problem 5 (6 points)\n",
"\n",
"Suppose that $X$ is a random $m \\times n$ matrix (drawn from some probability distribution) and its mean (expectation value) is $E[X] = X_0$.\n",
"\n",
"Derive the following identity for the variance (which we used in class):\n",
"\n",
"$$\n",
"E[\\Vert X - X_0 \\Vert_F^2] = E[\\Vert X \\Vert_F^2] - \\Vert X_0 \\Vert_F^2\n",
"$$\n",
"\n",
"(If you just write out the Frobenius norm definition $\\Vert A \\Vert_F^2 = \\text{tr}[A^T A]$, the problem should be very straightforward. Remember that the expectation value is conceptually just a sum, so it can be moved inside any linear operation like the trace.)"
]
},
{
"cell_type": "markdown",
"id": "88ab5b5e",
"metadata": {},
"source": [
"## Problems 5, 6, etc: coming soon"
"## Problems 6, etc: coming soon"
]
},
{