Below is a list of topics and concepts that will be used throughout the course.
They should be part of your undergraduate curriculum, but it is frequently the c
ase that these concepts have not been used recently, especially by CS majors. So you may need to freshen up. This list provides a guideline to do so. 1. Linear algebra. - linear (vector) spaces: what they are, how they are represented, how they are transformed one onto the other - linear combinations: what it takes for a space to be a linear space. - linear independence: what it means, what happens when a set of vectors is not linearly independent; how an it be transformed into another set that is "equival ent" and yet is made of linearly independent elements - basis: what it is, how to transform from one to the other - inner products, orthogonality: what it means; how to transform from an inner p roduct to another; abstract definition, generalization to vector spaces beyond R ^n - orthogonal matrices: definitions, properties - Gram-Schmidt orthogonalization: procedure and its meaning - range space, null space, rank: definitions, interpretation, calculation: How d o you compute the (range/null/rank) of a matrix? - eigenvalues, eigenvectors: definitions, interpretation; spectrum of a matrix - symmetric matrices: definitions, properties - solving linear systems of equations of the form Ax = b: you must be able to so lve a linear system of equations by hand, provided it is solvable. You must also be able to ... - conditions under which the system above can be solved, and can be solved uniqu ely: determine whether a linear system of equation is solvable (has at least one solution), and whether it has multiple solutions; in the latter case, you must be able to represent the set of all possible solutions analytically - least-squares approximation to the solution of a linear system: If a linear sy stem does not admit a solution, how do you "relax" it so that it admits infinite ly many, and how can you determine a criterion to choose one particular solution , among the infinitely many, that is in some sense "sensible". - (optional: SVD) 2. Calculus and numerical analysis. - limits: you must understand the concept of limit; epsilon-delta and all that; be facile with computing limits and derivatives - differentiation - Riemann integral: how to compute the integral of a function that is represente d using samples, as opposed to analytically. - Gauss-Newton methods, conjugate gradient, Levemberg-Marquardt: how to minimize a scalar-valued smooth function iteratively; conditions under which you are gua ranteed to converge to the global minimum; necessary conditions for a local mini mum - Constrained optimization, Lagrangian multipliers: how to frame a constrained o ptimization problem into an unconstrained one. - trigonometric functions, Fourier series: series of functions (polynomials, sin /cos, exponentials) - solution of linear system of ordinary differential equations x-dot = A x; you must be able to solve a linear system of equations with constant coefficients; y ou must have a procedure to compute, or at least approximate, matrix powers and matrix exponentials 3. Basic probability and stochastic processes - probability space: what it is, examples - random variables: definition, interpretation, understand the difference betwee n a random variable, its probabilistic description (distribution, density, momen
ts etc.), and its realization (sample)
- expectation: definition, interpretation, examples - conditional expectation: definition (it is a random variable), interpretation - moments: definition, interpretation; second- and third-order moments - marginalization: how to "integrate out" a random variable. - Bayes' rule: definition, interpretation - law of large numbers, central limit theorem: definition, interpretation, impli cations - stochastic process: definition, interpretation, understand the difference betw een a random process, its probabilistic description, its realizations - correlation function: definition, interpretation