Professor Alan Edelman has lead a somewhat unusual life proving
"pure math" theorems in random matrix
theory, developing numerical algorithms, and improving software for
high performance computing.
He has found that the theory and the practice go hand in hand.
While many suppose that the mathematics might suggest how to compute, the
is that the computing suggests how to do the
Edelman's current research passions:
What is it about Hermite, Laguerre, Jacobi that binds together so much
mathematics and computing? (There are hundreds of perfectly valid answers
to this question, none of which are yet completely satisfying.)
Clues are to be found in random matrix theory and in matrix computations
(eig, svd, gsvd).
Julia: A Fresh Approach to Technical Computing:
Technical computing languages let data scientists explore their
data sets, engineers model their structures, and scientists explore
We are building a technical computing language that is
Until now, most researchers thought all of the above were impossible.
- easy to use
- open source (science needs to be open source!)
- high performance (who says interpreted must mean low performance??)
- easy to parallelize (and no per processor fee!)
- lets you fairly readily do things that you did not even know were
possible (this is my favorite! for example, I can target parts of LAPACK
to take advantage of special matrix structures or you can explore
your own parallel algorithm without so much trouble)