Terms and notation that you might encounter in your future studies that we did not cover: • Regular matrix (common use) = invertible matrix • Singular matrix (common use) = not invertible matrix • “∀x . . . “ = “for all values of x. . . ” • “∃x . . . ” = “there exists a value of x. . . ” • Sn = all permutations on n elements • Quadratic form = a function f : Rn → R defined by f (x) = xT Ax for some A symmetric matrix. • Positive definite matrix = n × n symmetric matrix such that xT Ax > 0 for all x 6= 0. • Positive semidefinite matrix = n×n symmetric matrix such that xT Ax ≥ 0 for all x ∈ Rn . • Negative definite matrix = n × n symmetric matrix such that xT Ax < 0 for all x 6= 0. • Negative semidefinite matrix = n×n symmetric matrix such that xT Ax ≤ 0 for all x ∈ Rn . • Indefinite matrix = n × n matrix A that is not positive semidefinite, nor negative semidefinite. • Unitary matrix = a complex n×n matrix U such that U −1 = U T (complex analogue of orthogonal matrices). • GL(n) = all invertible n×n matrices together with the operation of matrix product (an example of a mathematical structure called a group). • |A| = det(A) • kAk is a norm of a matrix. Roughly speaking, a norm is a nonnegative number that tells us how much A differs from the zero matrix. There are several possible norms one can define. One example is the norm defined by kAk = max{kAxk : kxk = 1}. • Linear form = any linear transformation f : V → R. • “Vector space over the field Zp ” = Instead of calculating in R (or C) we will use integers {0, 1, . . . , p − 1} and do arithmetics modulo p, taking remainders after dividing by p (ie. “4 + 2 = 1 (mod 5)”). If p is a prime number, things work more or less like in R. The field Z2 is often used in computer science. • Singular value/Cholesky/LU/. . . decomposition = Decompositions often show up in computer algorithms. I’m omitting the details here, but the general idea is similar to QR factorization: Given a matrix A, we want to write A as a product of “nice” matrices. • Spectrum of A = the list of all eigenvalues of A with their algebraic multiplicities. • Dominant eigenvalue of A = the eigenvalue of A with the largest modulus (absolute value). • Spectral radius of A = the maximum modulus (absolute value) of an eigenvalue of A. • Exponential of a matrix = Given A square matrix, we define exp(A) = ∞ X Ak k=0 k! . This series converges for all matrices and has nice properties that are useful when solving differential equations. For example: d exp(At) = A exp(At). dt • “Conjugate transpose”/“Hermitian conjugate”/“Hermitian adjoint”/ “bedaggered matrix”/“adjoint matrix”/“transjugate” of A = the matrix AT (denoted by AH , A† or even A? ). • Homomorphism of vector spaces f : V → U = Here U, V are vector spaces and f is a mapping that satisfies f (0) = 0, f (g + h) = f (g) + f (h) for any g, h ∈ V and f (ch) = cf (h) for any c ∈ R (or C).