Ioana Dumitriu

Ioana Dumitriu is a Romanian-American mathematician whose research spans random matrix theory, numerical analysis, and the mathematics of networks.

Connect

Random matrix theory: The study of matrices with random entries or structure and the statistical behavior of their eigenvalues. It is one of Dumitriu’s central research areas.
Tridiagonal matrix: A square matrix whose nonzero entries lie on the main diagonal and the diagonals immediately beside it. Structured random tridiagonal matrices provide models for important eigenvalue distributions.
Alan Edelman: An applied mathematician known for research in random matrix theory, numerical analysis, and computation. Dumitriu’s doctoral research at MIT was supervised by Edelman.
Wigner semicircle law: A limiting distribution describing eigenvalues of large random symmetric matrices under broad conditions. It is a foundational universality result that complements explicit ensemble models.
Numerical analysis: The study of algorithms for approximating solutions to mathematical problems, especially those involving continuous quantities. Dumitriu’s work includes algorithms and computational questions in this field.
Gaussian ensemble: A family of random matrix distributions whose entries are Gaussian and whose symmetries define distinct ensembles. Dumitriu’s work includes tractable matrix models for classical random-matrix ensembles.
Mihai Putinar: A mathematician whose research includes operator theory, moment problems, and real algebraic geometry. He was among the mathematicians who advised Dumitriu during her doctoral training.
Marchenko–Pastur distribution: The limiting eigenvalue distribution for large sample covariance matrices under standard assumptions. It shows how random-matrix methods extend to covariance matrices and statistical data.
Spectral graph theory: The study of graphs through eigenvalues and eigenvectors of matrices associated with them. It connects her interests in networks, matrices, and spectral behavior.
β-ensemble: A family of random matrix eigenvalue distributions parameterized by a positive interaction parameter β. Her tridiagonal constructions helped extend classical eigenvalue models to general β.