Undergraduate Summer Research Projects
Attention Students wanting
first-hand experience in mathematical research this
Spring/Summer in the field of Orthogonal Matrices*
(*Warning: exposure to research in
this field has been demonstrated to be highly addictive!)
I hope to take on students as RAs (research
assistants) this summer (2017) but that will be dependent on
grant funding, which will be pending until about April
2017. In the meantime students wanting to work should find
a professor with an NSERC grant and apply for an NSERC USRA
award -- the deadline is in December; watch for announcements
and/or speak to the Faculty of Science office. Note you
must have a grantholder (therefore not me this year) sign your
application. You are not bound to work for that
grant holder, but there must be a credible project you can do
for them. If your intention is to work with me or anyone
other than the grant holder you should discuss this with them so
there are no misunderstandings.
There are a few awards at UM which do not depend
on grant funding. You should also apply for those, and I would
be able to take on a student with such an award regardless of
whether I have a grant in 2017.
Seminar Series: Introduction to
orthogonal matrices
If I will be taking on students I will announce a
once-weekly seminar beginning in the first week of April, open
to all interested students. We will cover the background
to several problems to be explored by undergrad RA’s this
term. Watch this space and my office door for
announcements.
Those who end up employed as RAs will be reimbursed for their
time attending and doing the (required for them) weekly reading,
in preparation for the summer. Other students may participate as
they see fit.
If you do not receive an award I still may consider hiring you
part time if you have the right background. For this I
will want to see:
- your most recent grade report
(unofficial/photocopy ok)
- a list of your current courses (if they don’t
appear on the report)
- a 1-2 page letter explaining why you are
interested in mathematical research and listing any prior
experience, if any.
- mention anything that will affect your
availability and any additional employment you will have
during the summer.
If either the seminars or the RA positions interest you, let me
know your schedule by Tuesday March 14 and it will be
taken into account in planning the seminars. Students interested
in the seminar and not sure whether to apply for the RA position
should attend; applications will be accepted until positions are
filled.
Some recent projects (some of which are ongoing):
- Circulant partial Hadamard matrices.
This subject began over 10 years ago arising out of a question
about stream-cipher cryptography. As it turns out the
objects we introduced are relevant to some key questions
concerning Hadamard matrices. Students working for me
over several subjects have produced some lovely new approaches
and many computational results, and we have developed the
theory for these interesting objects. An r-H(kxn) is a
kxn matrix obtained from a row of 1s and -1s of length n, with
subsequent rows obtained by circulating entries of prior rows
to the right by one position. The key property we want
is that any pair of rows are orthogonal, which means they
match in exactly half the positions. r is the sum of the
entries along that row. In the cryptographic setting we
always insisted that r=0, whereas in the general setting we
allow all values, and understanding the relationship between
r, k and n is critical. If we can refine this
relationship sufficiently we have a good chance of solving one
of the big unsolved problems in design theory, the circulant
Hadamard matrix conjecture (sometimes also called "Ryser's
Conjecture").
- Negacyclic weighing matrices.
This topic grew as a side-project in 2013 when a team of four
students were working on Circulant partial Hadamard
matrices. We kept happening upon questions about these
little-studied objects. Hampered by a too-thin
literature on the subject we embarked on our own investigation
into the existence of these objects, and two of the students
produced a significant body of new information on the
subject. Their work is summarized on a poster on the 5th
floor.
- 2x2 block decomposition conjecture for
Hadamard matrices. A decade ago a summer RA
working for me showed that, in order 24, there exists a
Hadamard matrix which cannot be decomposed entirely into 2x2
submatrices which are all rank 1. This was the first
instance of a class of objects of which we now have a number
of examples. A number of mathematicians have been
putting effort into the 2x2 block decompositions of Hadamard
matrices to learn something of their hidden structure. A
related conjecture -- which remains unsolved -- asks whether
every Hadamard matrix can be decomposed into 2x2 submatrices
which are all rank 2. Most people in the field appear to
believe this is true, but nobody has proven it. This is
a lovely project for student exploration.
- Sequences with zero autocorrelation.
A number of sets of sequences with a propery we call "zero
autocorrelation" are useful for building powerful
constructions for important combinatorial objects. About
15 years ago we pushed a class called ternary complementary
pairs past what was known, and an RA working for me found the
very first case having a weight of 29. We have some idea
where to look for a weight-37 case (the current smallest
unknown weight) but this will require some very clever
computer-based searching. A good project for those
interested in the computational side of combinatorics!
- Synthetic Orthogonality Theory.
This is a new and completely different project, which pushes
inroads into an unexplored field -- what exactly is
"orthogonality"? That is, what do we mean in
combinatorics when we speak of this? It has been a
concept that has grown without direction as we consider
discrete variants on the notion, which are consistent with
certain kinds of configurations that, in some abstract sense,
embody "orthogonality" but do not necessarily bear much
resemblance to the usual geometric notion of perpendicularity
as in the usual linear algebra. A few years ago
associates of mine in Ireland (Dr. Dane Flannery) and
California (Dr. Warwick de Launey) wrote a book, Algebraic
Design Theory, introducing an axiomatic notion of
"pairwise orthogonality" that embraces the gamut of these
meanings ... but they leave the elementary theory largely
undeveloped ... and most of the sweeping terrain thus called
into existence simply unexplored. Their groundbreaking
book opened up an entirely new field, but left another field,
which I call Synthetic Orthogonality Theory, more-or-less
virgin territory. I plan, over the next few years, to
map it out with some aspiring new explorers.
- Lots more ... I'll add some when I have time.
Further info:
R. Craigen
MH 523
474-7489
craigenr@umanitoba.ca
server.maths.umanitoba/~craigen/PWP/