Beamer slides - University of Illinois at Urbana-Champaign
Beamer slides - University of Illinois at Urbana-Champaign
Beamer slides - University of Illinois at Urbana-Champaign
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Apolarity, steampunk canonical forms and the<br />
obvious inner product<br />
Bruce Reznick<br />
<strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong><br />
Geometric Commut<strong>at</strong>ive Algebra and Applic<strong>at</strong>ions<br />
2011 Spring Central Section Meeting <strong>of</strong> AMS<br />
Iowa City, March 19, 2011<br />
Algebra, Geometry and Combin<strong>at</strong>orics Seminar<br />
UIUC, April 20, 2011<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Wh<strong>at</strong> is steampunk? It is a style based on combining 19th century<br />
Victorian culture with a few elements <strong>of</strong> modern life, such as<br />
computers.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Wh<strong>at</strong> is steampunk? It is a style based on combining 19th century<br />
Victorian culture with a few elements <strong>of</strong> modern life, such as<br />
computers.<br />
Wh<strong>at</strong> are canonical forms? Before I give a definition, here’s the<br />
Basic Example for this talk.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Wh<strong>at</strong> is steampunk? It is a style based on combining 19th century<br />
Victorian culture with a few elements <strong>of</strong> modern life, such as<br />
computers.<br />
Wh<strong>at</strong> are canonical forms? Before I give a definition, here’s the<br />
Basic Example for this talk.<br />
Any binary quadr<strong>at</strong>ic form over C can be written as<br />
There also exist αj ∈ C so th<strong>at</strong><br />
<strong>at</strong> least unless a = 0 and b �= 0.<br />
p(x, y) = ax 2 + 2bxy + cy 2 . (1)<br />
p(x, y) = (α1x + α2y) 2 + (α3y) 2 , (2)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Wh<strong>at</strong> is steampunk? It is a style based on combining 19th century<br />
Victorian culture with a few elements <strong>of</strong> modern life, such as<br />
computers.<br />
Wh<strong>at</strong> are canonical forms? Before I give a definition, here’s the<br />
Basic Example for this talk.<br />
Any binary quadr<strong>at</strong>ic form over C can be written as<br />
There also exist αj ∈ C so th<strong>at</strong><br />
p(x, y) = ax 2 + 2bxy + cy 2 . (1)<br />
p(x, y) = (α1x + α2y) 2 + (α3y) 2 , (2)<br />
<strong>at</strong> least unless a = 0 and b �= 0.<br />
Th<strong>at</strong> is, a “general” binary quadr<strong>at</strong>ic form can be written with the<br />
square completed.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
1. Introduction<br />
Webpage for this talk: www.m<strong>at</strong>h.uiuc.edu/∼reznick/iowa.html<br />
Wh<strong>at</strong> is steampunk? It is a style based on combining 19th century<br />
Victorian culture with a few elements <strong>of</strong> modern life, such as<br />
computers.<br />
Wh<strong>at</strong> are canonical forms? Before I give a definition, here’s the<br />
Basic Example for this talk.<br />
Any binary quadr<strong>at</strong>ic form over C can be written as<br />
There also exist αj ∈ C so th<strong>at</strong><br />
p(x, y) = ax 2 + 2bxy + cy 2 . (1)<br />
p(x, y) = (α1x + α2y) 2 + (α3y) 2 , (2)<br />
<strong>at</strong> least unless a = 0 and b �= 0.<br />
Th<strong>at</strong> is, a “general” binary quadr<strong>at</strong>ic form can be written with the<br />
square completed.<br />
You don’t need a pro<strong>of</strong> <strong>of</strong> this, but I’ll give you several anyway.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Note also th<strong>at</strong> there are three parameters in each <strong>of</strong> these<br />
represent<strong>at</strong>ions.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Note also th<strong>at</strong> there are three parameters in each <strong>of</strong> these<br />
represent<strong>at</strong>ions.<br />
This equality is necessary, but not sufficient. Obviously,<br />
� 3<br />
i=1 (αix) 2 won’t work. There are three parameters in<br />
(α1x + α2y) 2 + (iα1x + α3y) 2 , (3)<br />
but the coefficient <strong>of</strong> x 2 is 0, so the sum (3) must have a factor <strong>of</strong><br />
y, which a general binary quadr<strong>at</strong>ic form does not. This example<br />
may look silly, but it isn’t.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Note also th<strong>at</strong> there are three parameters in each <strong>of</strong> these<br />
represent<strong>at</strong>ions.<br />
This equality is necessary, but not sufficient. Obviously,<br />
� 3<br />
i=1 (αix) 2 won’t work. There are three parameters in<br />
(α1x + α2y) 2 + (iα1x + α3y) 2 , (3)<br />
but the coefficient <strong>of</strong> x 2 is 0, so the sum (3) must have a factor <strong>of</strong><br />
y, which a general binary quadr<strong>at</strong>ic form does not. This example<br />
may look silly, but it isn’t.<br />
Wh<strong>at</strong> is the as-yet-nonexistent subject <strong>of</strong> steampunk canonical<br />
forms? First approxim<strong>at</strong>ion: 19th century algebra plus the concept<br />
<strong>of</strong> vector spaces plus the belief th<strong>at</strong> there is still something <strong>of</strong><br />
interest in the algebraic geometry <strong>of</strong> binary forms.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I freely acknowledge th<strong>at</strong> any result not specifically credited to<br />
someone else may nevetheless be old. (References from the<br />
audience welcome.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I freely acknowledge th<strong>at</strong> any result not specifically credited to<br />
someone else may nevetheless be old. (References from the<br />
audience welcome.)<br />
There is very serious m<strong>at</strong>hem<strong>at</strong>ics in this area being done today.<br />
There is a strong connection between this talk and the wonderful<br />
1993 paper <strong>of</strong> Richard Ehrenborg and Gian-Carlo Rota connecting<br />
apolarity and canonical forms using m<strong>at</strong>roids. There is also recent<br />
work on these questions from a more sophistic<strong>at</strong>ed point <strong>of</strong> view by<br />
Tony Geramita, Tony Iarrobino, J. M. Landsberg, Zach Teitler,<br />
Enrico Carlini, Giorgio Ottaviani and others I might have missed.<br />
And <strong>of</strong> course the deep theorem <strong>of</strong> Alexander-Hirschowitz hovers<br />
over all.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I freely acknowledge th<strong>at</strong> any result not specifically credited to<br />
someone else may nevetheless be old. (References from the<br />
audience welcome.)<br />
There is very serious m<strong>at</strong>hem<strong>at</strong>ics in this area being done today.<br />
There is a strong connection between this talk and the wonderful<br />
1993 paper <strong>of</strong> Richard Ehrenborg and Gian-Carlo Rota connecting<br />
apolarity and canonical forms using m<strong>at</strong>roids. There is also recent<br />
work on these questions from a more sophistic<strong>at</strong>ed point <strong>of</strong> view by<br />
Tony Geramita, Tony Iarrobino, J. M. Landsberg, Zach Teitler,<br />
Enrico Carlini, Giorgio Ottaviani and others I might have missed.<br />
And <strong>of</strong> course the deep theorem <strong>of</strong> Alexander-Hirschowitz hovers<br />
over all.<br />
This talk has been constructed to avoid pr<strong>of</strong>undity, especially in<br />
the pro<strong>of</strong>s.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here are two examples <strong>of</strong> wh<strong>at</strong> I’ll be talking about and proving<br />
(the pro<strong>of</strong>s are quite simple):<br />
Theorem<br />
A general binary sextic form is the sum <strong>of</strong> the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form and the square <strong>of</strong> a cubic form.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here are two examples <strong>of</strong> wh<strong>at</strong> I’ll be talking about and proving<br />
(the pro<strong>of</strong>s are quite simple):<br />
Theorem<br />
A general binary sextic form is the sum <strong>of</strong> the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form and the square <strong>of</strong> a cubic form.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here are two examples <strong>of</strong> wh<strong>at</strong> I’ll be talking about and proving<br />
(the pro<strong>of</strong>s are quite simple):<br />
Theorem<br />
A general binary sextic form is the sum <strong>of</strong> the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form and the square <strong>of</strong> a cubic form.<br />
Theorem<br />
A general cubic form p(x1, . . . , xn) has a unique represent<strong>at</strong>ion in<br />
the form<br />
p(x1, . . . , xn) = �<br />
(α i,j<br />
1≤i≤j≤n<br />
i xi + · · · + α i,j<br />
j xj) 3 . (4)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here are two examples <strong>of</strong> wh<strong>at</strong> I’ll be talking about and proving<br />
(the pro<strong>of</strong>s are quite simple):<br />
Theorem<br />
A general binary sextic form is the sum <strong>of</strong> the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form and the square <strong>of</strong> a cubic form.<br />
Theorem<br />
A general cubic form p(x1, . . . , xn) has a unique represent<strong>at</strong>ion in<br />
the form<br />
p(x1, . . . , xn) = �<br />
(α i,j<br />
1≤i≤j≤n<br />
i xi + · · · + α i,j<br />
j xj) 3 . (4)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here are two examples <strong>of</strong> wh<strong>at</strong> I’ll be talking about and proving<br />
(the pro<strong>of</strong>s are quite simple):<br />
Theorem<br />
A general binary sextic form is the sum <strong>of</strong> the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form and the square <strong>of</strong> a cubic form.<br />
Theorem<br />
A general cubic form p(x1, . . . , xn) has a unique represent<strong>at</strong>ion in<br />
the form<br />
p(x1, . . . , xn) = �<br />
(α i,j<br />
1≤i≤j≤n<br />
i xi + · · · + α i,j<br />
j xj) 3 . (4)<br />
Even though (4) is actually a canonical form, it is not a<br />
represent<strong>at</strong>ion <strong>of</strong> p as a minimal number <strong>of</strong> cubes, so its<br />
applic<strong>at</strong>ion to, for example, the rank <strong>of</strong> tensors is unclear.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The canonical results given l<strong>at</strong>er for binary forms generalize a<br />
couple <strong>of</strong> classical theorems <strong>of</strong> Sylvester from 1851. Sylvester also<br />
gave algorithms in the process <strong>of</strong> giving his pro<strong>of</strong>s.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The canonical results given l<strong>at</strong>er for binary forms generalize a<br />
couple <strong>of</strong> classical theorems <strong>of</strong> Sylvester from 1851. Sylvester also<br />
gave algorithms in the process <strong>of</strong> giving his pro<strong>of</strong>s.<br />
The cubic theorem was suggested by a much more pr<strong>of</strong>ound result<br />
<strong>of</strong> Boris Reichstein from 1987:<br />
Theorem (Reichstein)<br />
A general cubic p(x1, . . . , xn) can be written as<br />
n�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 3 + q(x1, . . . , xn−2).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The canonical results given l<strong>at</strong>er for binary forms generalize a<br />
couple <strong>of</strong> classical theorems <strong>of</strong> Sylvester from 1851. Sylvester also<br />
gave algorithms in the process <strong>of</strong> giving his pro<strong>of</strong>s.<br />
The cubic theorem was suggested by a much more pr<strong>of</strong>ound result<br />
<strong>of</strong> Boris Reichstein from 1987:<br />
Theorem (Reichstein)<br />
A general cubic p(x1, . . . , xn) can be written as<br />
n�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 3 + q(x1, . . . , xn−2).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The canonical results given l<strong>at</strong>er for binary forms generalize a<br />
couple <strong>of</strong> classical theorems <strong>of</strong> Sylvester from 1851. Sylvester also<br />
gave algorithms in the process <strong>of</strong> giving his pro<strong>of</strong>s.<br />
The cubic theorem was suggested by a much more pr<strong>of</strong>ound result<br />
<strong>of</strong> Boris Reichstein from 1987:<br />
Theorem (Reichstein)<br />
A general cubic p(x1, . . . , xn) can be written as<br />
n�<br />
k=1<br />
Anybody here know this?<br />
(αk1x1 + · · · + αknxn) 3 + q(x1, . . . , xn−2).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
2. Preliminaries<br />
Let Hd(Cn ) denote the set <strong>of</strong> forms p(x1, . . . , xn) <strong>of</strong> degree d with<br />
coefficients in C. This is a vector space <strong>of</strong> dimension<br />
N(n, d) := � � n+d−1<br />
d . Let I(n, d) denote the index set <strong>of</strong><br />
monomials:<br />
�<br />
I(n, d) = (i1, . . . , in) : 0 ≤ ik ∈ Z, �<br />
�<br />
ik = d .<br />
Let x i = x i1<br />
1 denote the multinomial<br />
coefficient. If p ∈ Hd(Cn ), then we can write<br />
· · · x in<br />
n and c(i) = d!<br />
Q ik!<br />
p(x1, . . . , xn) = �<br />
i∈I(n,d)<br />
k<br />
c(i)a(p; i)x i .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here is an altern<strong>at</strong>ive th<strong>at</strong> is amazing to non-algebraic geometers,<br />
but not, apparently, to the algebraic geometers I’ve talked to. The<br />
only accessible pro<strong>of</strong> I know is in Ehrenborg-Rota.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here is an altern<strong>at</strong>ive th<strong>at</strong> is amazing to non-algebraic geometers,<br />
but not, apparently, to the algebraic geometers I’ve talked to. The<br />
only accessible pro<strong>of</strong> I know is in Ehrenborg-Rota.<br />
Theorem<br />
Suppose F : C N → C N is a polynomial map; th<strong>at</strong> is,<br />
F (t1, . . . , tN) = (f1(t1, . . . , tN), . . . , fN(t1, . . . , tN))<br />
where each fj ∈ C[t1, . . . , tN]. Then either (i) or (ii) holds:<br />
(i) The N polynomials {fj : 1 ≤ j ≤ N} are algebraically dependent<br />
and F (C N ) lies in some non-trivial {P = 0} in C N .<br />
(ii) The N polynomials {fj : 1 ≤ j ≤ N} are algebraically<br />
independent and F (C N ) is (<strong>at</strong> least) dense in C N .<br />
Furthermore, the second case occurs if and only if there is a point<br />
a ∈ CN �<br />
<strong>at</strong> which the Jacobian m<strong>at</strong>rix has full rank.<br />
� ∂fi<br />
∂tj (a)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
When N = N(n, d), we may interpret such an F as a map from<br />
C N to Hd(C n ) by indexing I(n, d) as {ik : 1 ≤ k ≤ N} and<br />
making the interpret<strong>at</strong>ion in an abuse <strong>of</strong> not<strong>at</strong>ion th<strong>at</strong><br />
F (t1, . . . , tN) =<br />
N�<br />
c(ik)fk(t1 . . . , tN)x ik<br />
k=1<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
When N = N(n, d), we may interpret such an F as a map from<br />
C N to Hd(C n ) by indexing I(n, d) as {ik : 1 ≤ k ≤ N} and<br />
making the interpret<strong>at</strong>ion in an abuse <strong>of</strong> not<strong>at</strong>ion th<strong>at</strong><br />
F (t1, . . . , tN) =<br />
N�<br />
c(ik)fk(t1 . . . , tN)x ik<br />
k=1<br />
To retrieve our Main Example, let i1 = (2, 0), i2 = (1, 1),<br />
i3 = (0, 2), and suppose the original polynomial map is<br />
F (t1, t2, t3) = (t2 1 , t1t2, t2 2 + t2 3 ). Then<br />
F (t1, t2, t3) = t 2 1x 2 1 + 2t1t2x1x2 + (t 2 2 + t 2 3)x 2 2<br />
= (t1x1 + t2x2) 2 + (t3x2) 2 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Definition<br />
A canonical form for Hd(C n ) is any polynomial map F from C N<br />
to Hd(C n ) so th<strong>at</strong> almost every p ∈ Hd(C n ) is in the range <strong>of</strong> F .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Definition<br />
A canonical form for Hd(C n ) is any polynomial map F from C N<br />
to Hd(C n ) so th<strong>at</strong> almost every p ∈ Hd(C n ) is in the range <strong>of</strong> F .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Definition<br />
A canonical form for Hd(C n ) is any polynomial map F from C N<br />
to Hd(C n ) so th<strong>at</strong> almost every p ∈ Hd(C n ) is in the range <strong>of</strong> F .<br />
I am somewh<strong>at</strong> puzzled th<strong>at</strong> the simplest canonical forms were<br />
<strong>of</strong>ten not considered so by nineteenth century m<strong>at</strong>hem<strong>at</strong>icians.<br />
Suppose {pj : 1 ≤ j ≤ N} is a basis for Hd(C n ). Then for every<br />
p ∈ Hd(C n ), there exist tj ∈ C so th<strong>at</strong> p = � tjpj. This canonical<br />
form must not have been considered “sporting” to the Victorians.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Definition<br />
A canonical form for Hd(C n ) is any polynomial map F from C N<br />
to Hd(C n ) so th<strong>at</strong> almost every p ∈ Hd(C n ) is in the range <strong>of</strong> F .<br />
I am somewh<strong>at</strong> puzzled th<strong>at</strong> the simplest canonical forms were<br />
<strong>of</strong>ten not considered so by nineteenth century m<strong>at</strong>hem<strong>at</strong>icians.<br />
Suppose {pj : 1 ≤ j ≤ N} is a basis for Hd(C n ). Then for every<br />
p ∈ Hd(C n ), there exist tj ∈ C so th<strong>at</strong> p = � tjpj. This canonical<br />
form must not have been considered “sporting” to the Victorians.<br />
The interpret<strong>at</strong>ion <strong>of</strong> the Jacobian having full rank is th<strong>at</strong>, for<br />
some a ∈ C n , the forms { ∂F<br />
∂tj (a)} span Hd(C n ).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Let’s see how this works in the Main Example.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Let’s see how this works in the Main Example.<br />
The partials <strong>of</strong> (t1x1 + t2x2) 2 + (t3x2) 2 with respect to the tj’s are:<br />
2x1(t1x1 + t2x2), 2x2(t1x1 + t2x2), 2x2(t3x2). (5)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Let’s see how this works in the Main Example.<br />
The partials <strong>of</strong> (t1x1 + t2x2) 2 + (t3x2) 2 with respect to the tj’s are:<br />
2x1(t1x1 + t2x2), 2x2(t1x1 + t2x2), 2x2(t3x2). (5)<br />
If we specialize <strong>at</strong> (t1, t2, t3) = (1, 0, 1), so t1x1 + t2x2 = x1 and<br />
t3x2 = x2, then (5) becomes 2x 2 1 , 2x1x2, 2x 2 2 and these do span<br />
H2(C2 ), providing an abstract existential pro<strong>of</strong> th<strong>at</strong> you can<br />
complete the square for binary quadr<strong>at</strong>ic forms!<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
We can now prove the first theorem. Consider the represent<strong>at</strong>ion<br />
p(x, y) = f 2 (x, y) + g 3 (x, y) =<br />
(t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ) 2 + (t5x 2 + t6xy + t7y 2 ) 3 :<br />
f (x, y) = t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ,<br />
g(x, y) = t5x 2 + t6xy + t7y 2 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
We can now prove the first theorem. Consider the represent<strong>at</strong>ion<br />
p(x, y) = f 2 (x, y) + g 3 (x, y) =<br />
(t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ) 2 + (t5x 2 + t6xy + t7y 2 ) 3 :<br />
f (x, y) = t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ,<br />
g(x, y) = t5x 2 + t6xy + t7y 2 .<br />
Then the partials with respect to the tj’s are:<br />
2x 3 f , 2x 2 yf , 2xy 2 f , 2y 3 f ; 3x 2 g 2 , 3xyg 2 , 3y 2 g 2 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
We can now prove the first theorem. Consider the represent<strong>at</strong>ion<br />
p(x, y) = f 2 (x, y) + g 3 (x, y) =<br />
(t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ) 2 + (t5x 2 + t6xy + t7y 2 ) 3 :<br />
f (x, y) = t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ,<br />
g(x, y) = t5x 2 + t6xy + t7y 2 .<br />
Then the partials with respect to the tj’s are:<br />
2x 3 f , 2x 2 yf , 2xy 2 f , 2y 3 f ; 3x 2 g 2 , 3xyg 2 , 3y 2 g 2 .<br />
If we specialize <strong>at</strong> distinct linear powers for {f , g}, say<br />
f = x 3 , g = y 2 , then these partials become:<br />
2x 6 , 2x 5 y, 2x 4 y 2 , 2x 3 y 3 ; 3x 2 y 4 , 3xy 5 , 3y 6 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
We can now prove the first theorem. Consider the represent<strong>at</strong>ion<br />
p(x, y) = f 2 (x, y) + g 3 (x, y) =<br />
(t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ) 2 + (t5x 2 + t6xy + t7y 2 ) 3 :<br />
f (x, y) = t1x 3 + t2x 2 y + t3xy 2 + t4y 3 ,<br />
g(x, y) = t5x 2 + t6xy + t7y 2 .<br />
Then the partials with respect to the tj’s are:<br />
2x 3 f , 2x 2 yf , 2xy 2 f , 2y 3 f ; 3x 2 g 2 , 3xyg 2 , 3y 2 g 2 .<br />
If we specialize <strong>at</strong> distinct linear powers for {f , g}, say<br />
f = x 3 , g = y 2 , then these partials become:<br />
2x 6 , 2x 5 y, 2x 4 y 2 , 2x 3 y 3 ; 3x 2 y 4 , 3xy 5 , 3y 6 .<br />
These trivially span H6(C 2 ). No algorithm, though.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The classical version <strong>of</strong> the Jacobian argument is called the<br />
Lasker-Wakeford Theorem. It’s worthwhile to spend a few minutes<br />
on who these people are.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The classical version <strong>of</strong> the Jacobian argument is called the<br />
Lasker-Wakeford Theorem. It’s worthwhile to spend a few minutes<br />
on who these people are.<br />
Emanuel Lasker (1868-1941) received his Ph.D. under Max<br />
Noether <strong>at</strong> Göttingen in 1902. He first developed the concept <strong>of</strong> a<br />
primary ideal and proved the primary decomposition theorem for an<br />
ideal <strong>of</strong> a polynomial ring in terms <strong>of</strong> primary ideals.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The classical version <strong>of</strong> the Jacobian argument is called the<br />
Lasker-Wakeford Theorem. It’s worthwhile to spend a few minutes<br />
on who these people are.<br />
Emanuel Lasker (1868-1941) received his Ph.D. under Max<br />
Noether <strong>at</strong> Göttingen in 1902. He first developed the concept <strong>of</strong> a<br />
primary ideal and proved the primary decomposition theorem for an<br />
ideal <strong>of</strong> a polynomial ring in terms <strong>of</strong> primary ideals.<br />
He is probably better known for being the world chess champion<br />
for 27 years (1894-1921), which spanned the life <strong>of</strong> ...<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Edward Kingsley Wakeford (1894-1916). I quote from the<br />
memorial article by J. H. Grace in the Proceedings <strong>of</strong> the London<br />
M<strong>at</strong>hem<strong>at</strong>ical Society. It may be the angriest obituary I’ve ever<br />
read, and it can be found in its entirety on the Iowa webpage.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Edward Kingsley Wakeford (1894-1916). I quote from the<br />
memorial article by J. H. Grace in the Proceedings <strong>of</strong> the London<br />
M<strong>at</strong>hem<strong>at</strong>ical Society. It may be the angriest obituary I’ve ever<br />
read, and it can be found in its entirety on the Iowa webpage.<br />
“He [EKW] was slightly wounded early in 1916, and soon after<br />
coming home was busy again with Canonical Forms.... [H]e<br />
discovered a paper <strong>of</strong> Hilbert’s which contained the very theorem<br />
he had long been in want <strong>of</strong> – first vaguely, and l<strong>at</strong>er quite<br />
definitely. This was in March; April found him, full <strong>of</strong> the most<br />
joyous and reverential admir<strong>at</strong>ion for the gre<strong>at</strong> German master,<br />
working away in fearful haste to finish the dissert<strong>at</strong>ion ...<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Edward Kingsley Wakeford (1894-1916). I quote from the<br />
memorial article by J. H. Grace in the Proceedings <strong>of</strong> the London<br />
M<strong>at</strong>hem<strong>at</strong>ical Society. It may be the angriest obituary I’ve ever<br />
read, and it can be found in its entirety on the Iowa webpage.<br />
“He [EKW] was slightly wounded early in 1916, and soon after<br />
coming home was busy again with Canonical Forms.... [H]e<br />
discovered a paper <strong>of</strong> Hilbert’s which contained the very theorem<br />
he had long been in want <strong>of</strong> – first vaguely, and l<strong>at</strong>er quite<br />
definitely. This was in March; April found him, full <strong>of</strong> the most<br />
joyous and reverential admir<strong>at</strong>ion for the gre<strong>at</strong> German master,<br />
working away in fearful haste to finish the dissert<strong>at</strong>ion ...<br />
He returned to the front in June and was killed in July.... He only<br />
needed a chance, and he never got it.”<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I found the Lasker-Wakeford Theorem in The theory <strong>of</strong><br />
determinants, m<strong>at</strong>rices and invariants by H. W. Turnbull, who was<br />
by some accounts the last old-style invariant theorist. Turnbull was<br />
a good m<strong>at</strong>hem<strong>at</strong>ician and his book is a real Rosetta Stone for<br />
understanding 19th century algebra. He described the theorem as<br />
“paradoxical and very curious”. I will rephrase it for simplicity.<br />
Theorem (Lasker-Wakeford)<br />
If F : C M → Hd(C n ) (M ≥ N(n, d)) then F is a canonical form if<br />
and only if there is a point a ∈ C M so th<strong>at</strong> there is no non-zero<br />
form q which is apolar to all N forms { ∂F<br />
∂F (a), . . . , ∂t1 ∂tN (a)}.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I found the Lasker-Wakeford Theorem in The theory <strong>of</strong><br />
determinants, m<strong>at</strong>rices and invariants by H. W. Turnbull, who was<br />
by some accounts the last old-style invariant theorist. Turnbull was<br />
a good m<strong>at</strong>hem<strong>at</strong>ician and his book is a real Rosetta Stone for<br />
understanding 19th century algebra. He described the theorem as<br />
“paradoxical and very curious”. I will rephrase it for simplicity.<br />
Theorem (Lasker-Wakeford)<br />
If F : C M → Hd(C n ) (M ≥ N(n, d)) then F is a canonical form if<br />
and only if there is a point a ∈ C M so th<strong>at</strong> there is no non-zero<br />
form q which is apolar to all N forms { ∂F<br />
∂F (a), . . . , ∂t1 ∂tN (a)}.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
I found the Lasker-Wakeford Theorem in The theory <strong>of</strong><br />
determinants, m<strong>at</strong>rices and invariants by H. W. Turnbull, who was<br />
by some accounts the last old-style invariant theorist. Turnbull was<br />
a good m<strong>at</strong>hem<strong>at</strong>ician and his book is a real Rosetta Stone for<br />
understanding 19th century algebra. He described the theorem as<br />
“paradoxical and very curious”. I will rephrase it for simplicity.<br />
Theorem (Lasker-Wakeford)<br />
If F : C M → Hd(C n ) (M ≥ N(n, d)) then F is a canonical form if<br />
and only if there is a point a ∈ C M so th<strong>at</strong> there is no non-zero<br />
form q which is apolar to all N forms { ∂F<br />
∂F (a), . . . , ∂t1 ∂tN (a)}.<br />
Most writers restrict themselves to M = N(n, d) as do I.<br />
The question for us now is, wh<strong>at</strong> does “apolarity” mean in this<br />
context?<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Apolarity<br />
I call the following the “obvious inner product”; it can be found in<br />
ancient invariant theory, and analysts call it the “Fisher inner<br />
product”. Either way, it’s <strong>at</strong> least 120 years old.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Apolarity<br />
I call the following the “obvious inner product”; it can be found in<br />
ancient invariant theory, and analysts call it the “Fisher inner<br />
product”. Either way, it’s <strong>at</strong> least 120 years old.<br />
Recall th<strong>at</strong><br />
p(x1 . . . , xn) = �<br />
c(i)a(p; i)x i .<br />
We now define, for p, q ∈ Hd(C n ):<br />
[p, q] = �<br />
i∈I(n,d)<br />
i∈I(n,d)<br />
c(i)a(p; i)a(q; i).<br />
(Ehrenborg-Rota use this extensively, but never explicitly.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Apolarity<br />
I call the following the “obvious inner product”; it can be found in<br />
ancient invariant theory, and analysts call it the “Fisher inner<br />
product”. Either way, it’s <strong>at</strong> least 120 years old.<br />
Recall th<strong>at</strong><br />
p(x1 . . . , xn) = �<br />
c(i)a(p; i)x i .<br />
We now define, for p, q ∈ Hd(C n ):<br />
[p, q] = �<br />
i∈I(n,d)<br />
i∈I(n,d)<br />
c(i)a(p; i)a(q; i).<br />
(Ehrenborg-Rota use this extensively, but never explicitly.)<br />
This is only an inner product for real forms; for complex forms you<br />
need a(q; i). The conjug<strong>at</strong>e actually only makes our expressions<br />
more complic<strong>at</strong>ed, [p, q] is really just a bilinear form on Hd(C n ).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Definition<br />
p, q ∈ Hd(C n ) are apolar if [p, q] = 0.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
For α ∈ C n , define (α·) d ∈ Hd(C n ) by<br />
(α·) d (x) = (α · x) d =<br />
⎛<br />
⎝<br />
n�<br />
j=1<br />
αjxj<br />
⎞<br />
⎠<br />
d<br />
= �<br />
i∈I(n,d)<br />
c(i)α i x i ,<br />
where the usual multinomial conventions apply. We define the<br />
differential oper<strong>at</strong>or q(D) for q ∈ He(C n ) in the usual way by<br />
q(D) = �<br />
i∈I(n,e)<br />
� �i1 � �in ∂<br />
∂<br />
c(i)a(q; i) · · · .<br />
∂x1 ∂xn<br />
The reason the obvious inner product is so useful is th<strong>at</strong> it has<br />
many nice properties; all can be verified formally. Pro<strong>of</strong>s are<br />
available in many <strong>of</strong> the papers <strong>at</strong> Iowa.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
p(α) = [p, (α·) d ].<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
p(α) = [p, (α·) d ].<br />
If e = d, then p(D)q = d![p, q] = q(D)p.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
p(α) = [p, (α·) d ].<br />
If e = d, then p(D)q = d![p, q] = q(D)p.<br />
If f ∈ He(C n ) and g ∈ Hd−e(C n ), then<br />
d![fg, p] = (fg)(D)p = f (D)g(D)p = e![f , g(D)p]<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
p(α) = [p, (α·) d ].<br />
If e = d, then p(D)q = d![p, q] = q(D)p.<br />
If f ∈ He(C n ) and g ∈ Hd−e(C n ), then<br />
d![fg, p] = (fg)(D)p = f (D)g(D)p = e![f , g(D)p]<br />
1 ∂p<br />
d ∂xj (α) = [p, xj(α·) d−1 ], etc.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The inner product s<strong>at</strong>isfies these properties:<br />
[p, q] = [q, p].<br />
p(α) = [p, (α·) d ].<br />
If e = d, then p(D)q = d![p, q] = q(D)p.<br />
If f ∈ He(C n ) and g ∈ Hd−e(C n ), then<br />
d![fg, p] = (fg)(D)p = f (D)g(D)p = e![f , g(D)p]<br />
1 ∂p<br />
d ∂xj (α) = [p, xj(α·) d−1 ], etc.<br />
If e ≤ d and g ∈ Hd−e(Cn ), then<br />
g(D)(α·) d = d!<br />
e! g(α)(α·)e .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The consequences <strong>of</strong> this definition include:<br />
p is apolar to (α·) d iff p(α) = 0.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The consequences <strong>of</strong> this definition include:<br />
p is apolar to (α·) d iff p(α) = 0.<br />
p is apolar to g(x)(α·) d−e for every g ∈ He(C n ) iff all e-th<br />
order deriv<strong>at</strong>ives <strong>of</strong> p vanish <strong>at</strong> α iff p vanishes to e-th order<br />
<strong>at</strong> α.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The consequences <strong>of</strong> this definition include:<br />
p is apolar to (α·) d iff p(α) = 0.<br />
p is apolar to g(x)(α·) d−e for every g ∈ He(C n ) iff all e-th<br />
order deriv<strong>at</strong>ives <strong>of</strong> p vanish <strong>at</strong> α iff p vanishes to e-th order<br />
<strong>at</strong> α.<br />
If deg q ≤ deg p and q(D)p = 0, then all multiples <strong>of</strong> q in<br />
Hd(C n ) are apolar to p.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The consequences <strong>of</strong> this definition include:<br />
p is apolar to (α·) d iff p(α) = 0.<br />
p is apolar to g(x)(α·) d−e for every g ∈ He(C n ) iff all e-th<br />
order deriv<strong>at</strong>ives <strong>of</strong> p vanish <strong>at</strong> α iff p vanishes to e-th order<br />
<strong>at</strong> α.<br />
If deg q ≤ deg p and q(D)p = 0, then all multiples <strong>of</strong> q in<br />
Hd(C n ) are apolar to p.<br />
In Ehrenborg-Rota, if p and q are forms <strong>of</strong> possibly different<br />
degree, p is apolar to q if p(D)q = 0. The definitions coincide<br />
when the degrees are equal, but not otherwise. In th<strong>at</strong> case,<br />
the definition is not symmetric: if deg p > deg q, then p(D)q<br />
will always equal 0.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
The classical “Fundamental Theorem <strong>of</strong> Apolarity” can now be<br />
easily st<strong>at</strong>ed and proved. I don’t know how it was understood<br />
before the Nullstellens<strong>at</strong>z. The theorem applies even when e > d.<br />
Theorem (FTA)<br />
Suppose q ∈ He(C n ) is irreducible and p ∈ Hd(C n ). Then<br />
q(D)p = 0 iff there exist αk ⊂ {α : q(α) = 0} and λk ∈ C such<br />
th<strong>at</strong><br />
p(x) =<br />
n�<br />
λk(αk · x) d .<br />
k=1<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
Fix q. Define the two subspaces<br />
A = {p ∈ Hd(C n ) : q(D)p = 0},<br />
B = {p ∈ Hd(C n ) : p = � λk(αk·) d , q(αk) = 0}.<br />
We want to show th<strong>at</strong> B ⊆ A and B ⊥ ⊆ A ⊥ ; if so, then since<br />
Hd(C n ) is finite dimensional, it will follow th<strong>at</strong> A = B.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
Fix q. Define the two subspaces<br />
A = {p ∈ Hd(C n ) : q(D)p = 0},<br />
B = {p ∈ Hd(C n ) : p = � λk(αk·) d , q(αk) = 0}.<br />
We want to show th<strong>at</strong> B ⊆ A and B ⊥ ⊆ A ⊥ ; if so, then since<br />
Hd(C n ) is finite dimensional, it will follow th<strong>at</strong> A = B.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
Fix q. Define the two subspaces<br />
A = {p ∈ Hd(C n ) : q(D)p = 0},<br />
B = {p ∈ Hd(C n ) : p = � λk(αk·) d , q(αk) = 0}.<br />
We want to show th<strong>at</strong> B ⊆ A and B ⊥ ⊆ A ⊥ ; if so, then since<br />
Hd(C n ) is finite dimensional, it will follow th<strong>at</strong> A = B.<br />
First, if e > d, then A = Hd(Cn �<br />
), so B ⊆ A. If e ≤ d and p ∈ B,<br />
λkq(αk)(αk·) d−e = 0, so p ∈ A.<br />
then q(D)p = d!<br />
(d−e)!<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
Fix q. Define the two subspaces<br />
A = {p ∈ Hd(C n ) : q(D)p = 0},<br />
B = {p ∈ Hd(C n ) : p = � λk(αk·) d , q(αk) = 0}.<br />
We want to show th<strong>at</strong> B ⊆ A and B ⊥ ⊆ A ⊥ ; if so, then since<br />
Hd(C n ) is finite dimensional, it will follow th<strong>at</strong> A = B.<br />
First, if e > d, then A = Hd(Cn �<br />
), so B ⊆ A. If e ≤ d and p ∈ B,<br />
λkq(αk)(αk·) d−e = 0, so p ∈ A.<br />
then q(D)p = d!<br />
(d−e)!<br />
Observe th<strong>at</strong> f ∈ B ⊥ iff q(α) = 0 implies [f , (α·) d ] = f (α) = 0.<br />
Since q is irreducible, the Nullstellens<strong>at</strong>z implies th<strong>at</strong> q | f . If<br />
e > d, this is impossible unless f = 0, so B ⊥ = {0} ⊆ A ⊥ . If<br />
e ≤ d, then f = gq where g ∈ Hd−e(Cn ). But p ∈ A =⇒<br />
[q(D)p, g] = 0. It follows<br />
q(D)p = 0 =⇒ [p, f ] = [p, gq] = d!<br />
(d−e)!<br />
th<strong>at</strong> f ∈ A ⊥ , completing the pro<strong>of</strong>.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
About 15 years ago, I proved a fairly obvious generaliz<strong>at</strong>ion <strong>of</strong> this<br />
theorem when q is not irreducible. The pro<strong>of</strong> is, in spirit, the same<br />
as the one given above.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
About 15 years ago, I proved a fairly obvious generaliz<strong>at</strong>ion <strong>of</strong> this<br />
theorem when q is not irreducible. The pro<strong>of</strong> is, in spirit, the same<br />
as the one given above.<br />
Theorem<br />
Suppose q ∈ He(Cn ) factors as �r j=1 qmj<br />
j into a product <strong>of</strong> distinct<br />
irreducible factors and suppose p ∈ Hd(Cn ). Then q(D)p = 0 iff<br />
there exist αjk ⊂ {qj(α) = 0}, and φjk ∈ Hmj −1(Cn ) such th<strong>at</strong><br />
p(x) =<br />
r�<br />
j=1<br />
� nj<br />
�<br />
k=1<br />
φjk(x)(αkj · x) d−(mj −1)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
�<br />
.
About 15 years ago, I proved a fairly obvious generaliz<strong>at</strong>ion <strong>of</strong> this<br />
theorem when q is not irreducible. The pro<strong>of</strong> is, in spirit, the same<br />
as the one given above.<br />
Theorem<br />
Suppose q ∈ He(Cn ) factors as �r j=1 qmj<br />
j into a product <strong>of</strong> distinct<br />
irreducible factors and suppose p ∈ Hd(Cn ). Then q(D)p = 0 iff<br />
there exist αjk ⊂ {qj(α) = 0}, and φjk ∈ Hmj −1(Cn ) such th<strong>at</strong><br />
p(x) =<br />
r�<br />
j=1<br />
� nj<br />
�<br />
k=1<br />
φjk(x)(αkj · x) d−(mj −1)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
�<br />
.
About 15 years ago, I proved a fairly obvious generaliz<strong>at</strong>ion <strong>of</strong> this<br />
theorem when q is not irreducible. The pro<strong>of</strong> is, in spirit, the same<br />
as the one given above.<br />
Theorem<br />
Suppose q ∈ He(Cn ) factors as �r j=1 qmj<br />
j into a product <strong>of</strong> distinct<br />
irreducible factors and suppose p ∈ Hd(Cn ). Then q(D)p = 0 iff<br />
there exist αjk ⊂ {qj(α) = 0}, and φjk ∈ Hmj −1(Cn ) such th<strong>at</strong><br />
p(x) =<br />
r�<br />
j=1<br />
� nj<br />
�<br />
k=1<br />
φjk(x)(αkj · x) d−(mj −1)<br />
In the case <strong>of</strong> binary forms, where zeros correspond to linear<br />
factors, the theorem is even simpler, and can be made equivalent<br />
to Gundelfinger’s generaliz<strong>at</strong>ion <strong>of</strong> Sylvester’s canonical forms.<br />
This next theorem also appears in Ehrenborg-Rota.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
�<br />
.
Theorem<br />
Suppose � mj = d + 1 and let ℓi(x, y) = αix + βiy. Suppose<br />
further th<strong>at</strong> ℓi and ℓj are pairwise linearly independent for i �= j.<br />
Then every p ∈ Hd(C 2 ) can be written as<br />
�<br />
j φj(x, y)(βjx − αjy) d−(mj −1) , where φj ∈ H d−(mj −1)(C 2 ).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Theorem<br />
Suppose � mj = d + 1 and let ℓi(x, y) = αix + βiy. Suppose<br />
further th<strong>at</strong> ℓi and ℓj are pairwise linearly independent for i �= j.<br />
Then every p ∈ Hd(C 2 ) can be written as<br />
�<br />
j φj(x, y)(βjx − αjy) d−(mj −1) , where φj ∈ H d−(mj −1)(C 2 ).<br />
Pro<strong>of</strong>.<br />
This is an easy corollary <strong>of</strong> the last theorem. More directly,<br />
�<br />
�<br />
A = φj(x, y)(βjx − αjy) d−(mj −1)<br />
: φj ∈ Hmj −1(C 2 �<br />
) .<br />
j<br />
Observe th<strong>at</strong> p ∈ A ⊥ ⇐⇒ p is apolar to each possible<br />
x r y mj −1−r (βjx − αjy) d−(mj −1) ⇐⇒ p vanishes to mj-th order <strong>at</strong><br />
(βj, −αj) ⇐⇒ p is divisible by each ℓ mj<br />
j . Since �<br />
j mj = d + 1,<br />
this implies p = 0, hence A = {0} ⊥ = Hd(C 2 ).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Remember canonical forms? Suppose α1, . . . , αn appear in a<br />
canonical form as<br />
(α1x1 + · · · + αnxn) d = ℓ d .<br />
Then ∂F<br />
∂αj = dxjℓ d−1 , and in applying Lasker-Wakeford, note th<strong>at</strong> a<br />
form is apolar to each <strong>of</strong> these if and only if it is singular <strong>at</strong><br />
(α1, . . . , αn). Start thinking about general forms which are singular<br />
<strong>at</strong> general sets <strong>of</strong> points and you enter the context in which<br />
Alexander-Hirschowitz comes in, and we change topics.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Some canonical forms for binary forms<br />
In 1869, J. J. Sylvester (1814-1897) reflected on the discovery <strong>of</strong><br />
some <strong>of</strong> his most famous research in 1851, done while he was<br />
working as an actuary.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Some canonical forms for binary forms<br />
In 1869, J. J. Sylvester (1814-1897) reflected on the discovery <strong>of</strong><br />
some <strong>of</strong> his most famous research in 1851, done while he was<br />
working as an actuary.<br />
“I discovered and developed the whole theory <strong>of</strong> canonical binary<br />
forms for odd degrees, and, as far as yet made out, for even<br />
degrees too, <strong>at</strong> one evening sitting, with a decanter <strong>of</strong> port wine to<br />
sustain n<strong>at</strong>ure’s flagging energies, in a back <strong>of</strong>fice in Lincoln’s Inn<br />
Fields. The work was done, and well done, but <strong>at</strong> the usual cost <strong>of</strong><br />
racking thought — a brain on fire, and feet feeling, or feelingless,<br />
as if plunged in an ice-pail. Th<strong>at</strong> night we slept no more.”<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Theorem (Sylvester)<br />
(i) A general binary form <strong>of</strong> degree d = 2k − 1 can be written as<br />
k�<br />
(αjx + βy) 2k−1 .<br />
j=1<br />
(ii) A general binary form <strong>of</strong> degree d = 2k can be written as<br />
λx 2k +<br />
k�<br />
(αjx + βy) 2k .<br />
j=1<br />
Note th<strong>at</strong> the constant counts work out: 2 · k = (2k − 1) + 1 and<br />
1 + 2 · k = 2k + 1.<br />
These are immedi<strong>at</strong>e consequences <strong>of</strong> the results I just described,<br />
but Sylvester did even better, giving an algorithmic construction.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Theorem (Sylvester)<br />
Suppose p(x, y) = �d � � d<br />
j=0 j ajx d−j y j and h(x, y) =<br />
�r t=0 ctx r−t y t = �r j=1 (βjx − αjy) is a product <strong>of</strong> pairwise<br />
distinct linear factors. Then there exist λk ∈ C so th<strong>at</strong><br />
if and only if<br />
⎛<br />
⎜<br />
⎝<br />
p(x, y) =<br />
r�<br />
λk(αkx + βky) d<br />
k=1<br />
a0 a1 · · · ar<br />
a1 a2 · · · ar+1<br />
.<br />
.<br />
. ..<br />
ad−r ad−r+1 · · · ad<br />
.<br />
⎞<br />
⎟<br />
⎠ ·<br />
⎛ ⎞<br />
c0<br />
⎜<br />
⎜c1<br />
⎟<br />
⎜ ⎟<br />
⎝ . ⎠ =<br />
⎛ ⎞<br />
0<br />
⎜<br />
⎜0<br />
⎟<br />
⎜ ⎟<br />
⎝.<br />
⎠<br />
0<br />
.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
cr
Here is an example <strong>of</strong> Sylvester’s algorithm in action. Let<br />
p(x, y) = 3x 5 − 20x 3 y 2 + 10xy 4 =<br />
� �<br />
5<br />
· 3 x<br />
0<br />
5 +<br />
� �<br />
5<br />
+<br />
3<br />
� 5<br />
1<br />
· 0 x 2 y 3 +<br />
�<br />
· 0 x 4 � �<br />
5<br />
y +<br />
2<br />
� �<br />
5<br />
· 2 xy<br />
4<br />
4 +<br />
· (−2) x 3 y 2<br />
� 5<br />
5<br />
�<br />
· 0 y 5 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here is an example <strong>of</strong> Sylvester’s algorithm in action. Let<br />
Since<br />
p(x, y) = 3x 5 − 20x 3 y 2 + 10xy 4 =<br />
� �<br />
5<br />
· 3 x<br />
0<br />
5 +<br />
� �<br />
5<br />
+<br />
3<br />
� 5<br />
1<br />
· 0 x 2 y 3 +<br />
�<br />
· 0 x 4 � �<br />
5<br />
y +<br />
2<br />
� �<br />
5<br />
· 2 xy<br />
4<br />
4 +<br />
· (−2) x 3 y 2<br />
� 5<br />
5<br />
�<br />
· 0 y 5 .<br />
⎛ ⎞<br />
⎛<br />
⎞ 0<br />
3 0 −2 0 ⎜<br />
⎝ 0 −2 0 2⎠<br />
· ⎜1<br />
⎟<br />
⎝0⎠<br />
−2 0 2 0<br />
1<br />
=<br />
⎛ ⎞<br />
0<br />
⎝0⎠<br />
,<br />
0<br />
we have h(x, y) = y(x 2 + y 2 ) = y(y − ix)(y + ix).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Here is an example <strong>of</strong> Sylvester’s algorithm in action. Let<br />
Since<br />
p(x, y) = 3x 5 − 20x 3 y 2 + 10xy 4 =<br />
� �<br />
5<br />
· 3 x<br />
0<br />
5 +<br />
� �<br />
5<br />
+<br />
3<br />
� 5<br />
1<br />
· 0 x 2 y 3 +<br />
�<br />
· 0 x 4 � �<br />
5<br />
y +<br />
2<br />
� �<br />
5<br />
· 2 xy<br />
4<br />
4 +<br />
· (−2) x 3 y 2<br />
� 5<br />
5<br />
�<br />
· 0 y 5 .<br />
⎛ ⎞<br />
⎛<br />
⎞ 0<br />
3 0 −2 0 ⎜<br />
⎝ 0 −2 0 2⎠<br />
· ⎜1<br />
⎟<br />
⎝0⎠<br />
−2 0 2 0<br />
1<br />
=<br />
⎛ ⎞<br />
0<br />
⎝0⎠<br />
,<br />
0<br />
we have h(x, y) = y(x 2 + y 2 ) = y(y − ix)(y + ix).<br />
Accordingly, there exist λk ∈ C so th<strong>at</strong><br />
p(x, y) = λ1x 5 + λ2(x + iy) 5 + λ3(x − iy) 5 .<br />
Indeed, λ1 = λ2 = λ3 = 1, as may be checked.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks about Sylvester’s algorithm<br />
If h(D) = �r ∂ ∂<br />
j=1 (βj ∂x − αj ∂y ) = �r t=0 ct<br />
h(D)p =<br />
�d−r<br />
m=0<br />
d!<br />
(d − r − m)!m!<br />
� d−r<br />
�<br />
i=0<br />
∂r ∂x r−t ∂y t , then<br />
ai+mci<br />
�<br />
x d−r−m y m<br />
The coefficients <strong>of</strong> h(D)p are, up to multiple, the rows in the<br />
m<strong>at</strong>rix product, so the m<strong>at</strong>rix condition is h(D)p = 0. This is<br />
FTA configured for products <strong>of</strong> linear factors.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks about Sylvester’s algorithm<br />
If h(D) = �r ∂ ∂<br />
j=1 (βj ∂x − αj ∂y ) = �r t=0 ct<br />
h(D)p =<br />
�d−r<br />
m=0<br />
d!<br />
(d − r − m)!m!<br />
� d−r<br />
�<br />
i=0<br />
∂r ∂x r−t ∂y t , then<br />
ai+mci<br />
�<br />
x d−r−m y m<br />
The coefficients <strong>of</strong> h(D)p are, up to multiple, the rows in the<br />
m<strong>at</strong>rix product, so the m<strong>at</strong>rix condition is h(D)p = 0. This is<br />
FTA configured for products <strong>of</strong> linear factors.<br />
An altern<strong>at</strong>e pro<strong>of</strong> <strong>of</strong> Sylvester’s Theorem is basically<br />
equivalent to computing the solution <strong>of</strong> constant-coefficient<br />
linear recurrence sequences.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks about Sylvester’s algorithm<br />
If h(D) = �r ∂ ∂<br />
j=1 (βj ∂x − αj ∂y ) = �r t=0 ct<br />
h(D)p =<br />
�d−r<br />
m=0<br />
d!<br />
(d − r − m)!m!<br />
� d−r<br />
�<br />
i=0<br />
∂r ∂x r−t ∂y t , then<br />
ai+mci<br />
�<br />
x d−r−m y m<br />
The coefficients <strong>of</strong> h(D)p are, up to multiple, the rows in the<br />
m<strong>at</strong>rix product, so the m<strong>at</strong>rix condition is h(D)p = 0. This is<br />
FTA configured for products <strong>of</strong> linear factors.<br />
An altern<strong>at</strong>e pro<strong>of</strong> <strong>of</strong> Sylvester’s Theorem is basically<br />
equivalent to computing the solution <strong>of</strong> constant-coefficient<br />
linear recurrence sequences.<br />
If d = 2s − 1 and r = s, then the m<strong>at</strong>rix is s × (s + 1) and<br />
has a non-trivial null-vector. The corresponding h (which can<br />
be given in terms <strong>of</strong> the coefficients <strong>of</strong> p) has distinct factors<br />
unless its discriminant vanishes. This gives the canonical form<br />
in odd degree.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If d = 2s and r = s, then the m<strong>at</strong>rix is square, and in general,<br />
there exists λ so th<strong>at</strong> p(x, y) − λx 2s has a m<strong>at</strong>rix with a<br />
non-trivial null-vector as above. This gives the canonical form<br />
in even degree. Th<strong>at</strong> extra wobble is wh<strong>at</strong> Sylvester must<br />
have meant by “as far as yet made out”.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If d = 2s and r = s, then the m<strong>at</strong>rix is square, and in general,<br />
there exists λ so th<strong>at</strong> p(x, y) − λx 2s has a m<strong>at</strong>rix with a<br />
non-trivial null-vector as above. This gives the canonical form<br />
in even degree. Th<strong>at</strong> extra wobble is wh<strong>at</strong> Sylvester must<br />
have meant by “as far as yet made out”.<br />
In this even case, the determinant <strong>of</strong> the square m<strong>at</strong>rix is the<br />
c<strong>at</strong>alecticant. Sylvester apologized for introducing this term:<br />
“Meic<strong>at</strong>alecticizant would more completely express the<br />
meaning <strong>of</strong> th<strong>at</strong> which, for the sake <strong>of</strong> brevity, I denomin<strong>at</strong>e<br />
the c<strong>at</strong>alecticant.” Sylvester was very interested in the<br />
technical aspects <strong>of</strong> poetry and a “c<strong>at</strong>alectic” verse is one in<br />
which the last line is missing a foot.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If d = 2s and r = s, then the m<strong>at</strong>rix is square, and in general,<br />
there exists λ so th<strong>at</strong> p(x, y) − λx 2s has a m<strong>at</strong>rix with a<br />
non-trivial null-vector as above. This gives the canonical form<br />
in even degree. Th<strong>at</strong> extra wobble is wh<strong>at</strong> Sylvester must<br />
have meant by “as far as yet made out”.<br />
In this even case, the determinant <strong>of</strong> the square m<strong>at</strong>rix is the<br />
c<strong>at</strong>alecticant. Sylvester apologized for introducing this term:<br />
“Meic<strong>at</strong>alecticizant would more completely express the<br />
meaning <strong>of</strong> th<strong>at</strong> which, for the sake <strong>of</strong> brevity, I denomin<strong>at</strong>e<br />
the c<strong>at</strong>alecticant.” Sylvester was very interested in the<br />
technical aspects <strong>of</strong> poetry and a “c<strong>at</strong>alectic” verse is one in<br />
which the last line is missing a foot.<br />
To his credit, in the same paper, Sylvester introduced the<br />
term “unimodular” in its current meaning.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Possibly new results<br />
It is easy to see th<strong>at</strong> if ℓj(x, y) = αjx + βjy, 1 ≤ j ≤ d + 1 is a set<br />
<strong>of</strong> d + 1 pairwise distinct linear forms, then {ℓd j } is a basis for<br />
Hd(C2 ): the represent<strong>at</strong>ion <strong>of</strong> {ℓd j } with respect to the basis<br />
{ � d<br />
j<br />
� x d−j y j } has Vandermonde determinant<br />
�<br />
1≤i
Possibly new results<br />
It is easy to see th<strong>at</strong> if ℓj(x, y) = αjx + βjy, 1 ≤ j ≤ d + 1 is a set<br />
<strong>of</strong> d + 1 pairwise distinct linear forms, then {ℓd j } is a basis for<br />
Hd(C2 ): the represent<strong>at</strong>ion <strong>of</strong> {ℓd j } with respect to the basis<br />
{ � d<br />
j<br />
� x d−j y j } has Vandermonde determinant<br />
�<br />
1≤i
Theorem<br />
Suppose m, n ≥ 0 and d ≥ 1 are integers so th<strong>at</strong> m + n = d + 1.<br />
Suppose ℓj(x, y) = βjx + γjy, 1 ≤ j ≤ m, are fixed pairwise<br />
linearly independent linear forms and suppose ek | d, 1 ≤ k ≤ r<br />
and �r k=1 (ek + 1) = n. Then a general binary form <strong>of</strong> degree d<br />
can be written as<br />
p(x, y) =<br />
m�<br />
j=1<br />
cjℓ d j (x, y) +<br />
r�<br />
k=1<br />
where cj ∈ C and fk is a form <strong>of</strong> degree ek.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
f d/ek<br />
k (x, y), (6)
Theorem<br />
Suppose m, n ≥ 0 and d ≥ 1 are integers so th<strong>at</strong> m + n = d + 1.<br />
Suppose ℓj(x, y) = βjx + γjy, 1 ≤ j ≤ m, are fixed pairwise<br />
linearly independent linear forms and suppose ek | d, 1 ≤ k ≤ r<br />
and �r k=1 (ek + 1) = n. Then a general binary form <strong>of</strong> degree d<br />
can be written as<br />
p(x, y) =<br />
m�<br />
j=1<br />
cjℓ d j (x, y) +<br />
r�<br />
k=1<br />
where cj ∈ C and fk is a form <strong>of</strong> degree ek.<br />
f d/ek<br />
k (x, y), (6)<br />
The novelty here is the existence <strong>of</strong> forms <strong>of</strong> intermedi<strong>at</strong>e degree<br />
taken to higher powers.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
The parameters are the m constants cj and, for each k, the ek + 1<br />
coefficients <strong>of</strong><br />
ek�<br />
fk(x, y) = αkux ek−u u<br />
y<br />
u=0<br />
The partials with respect to the cj’s are simply {ℓ d 1 , · · · , ℓd m}, and<br />
the partial with respect to αku is<br />
(d/ek)x ek−u u d/ek−1<br />
y fk .<br />
Now evalu<strong>at</strong>e the Jacobian <strong>at</strong> a choice <strong>of</strong> parameters so th<strong>at</strong><br />
fk(x, y) = ˜ℓ ek<br />
k , where the linear forms ˜ℓk are chosen so th<strong>at</strong> the<br />
combined set {ℓj, ˜ℓk} is pairwise linearly independent. Then<br />
f d/ek−1<br />
k = ˜ℓ d−ek<br />
k , and it is taken times a basis <strong>of</strong> Hek (C2 ). By<br />
earlier theorems, this set, taken all together, is a basis for Hd(C2 )<br />
and so (6) is a canonical form.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks:<br />
This result follows easily from the end <strong>of</strong> the Ehrenborg-Rota<br />
paper, but does not seem to appear there explicitly.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks:<br />
This result follows easily from the end <strong>of</strong> the Ehrenborg-Rota<br />
paper, but does not seem to appear there explicitly.<br />
If ek ≡ 1, m = d + 1 and n = 0, this is the “Vandermonde<br />
basis” canonical form.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks:<br />
This result follows easily from the end <strong>of</strong> the Ehrenborg-Rota<br />
paper, but does not seem to appear there explicitly.<br />
If ek ≡ 1, m = d + 1 and n = 0, this is the “Vandermonde<br />
basis” canonical form.<br />
If ek ≡ 1, d = 2r − 1, m = 0 and n = r, this recovers<br />
Sylvester’s canonical form for binary forms <strong>of</strong> odd degree.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks:<br />
This result follows easily from the end <strong>of</strong> the Ehrenborg-Rota<br />
paper, but does not seem to appear there explicitly.<br />
If ek ≡ 1, m = d + 1 and n = 0, this is the “Vandermonde<br />
basis” canonical form.<br />
If ek ≡ 1, d = 2r − 1, m = 0 and n = r, this recovers<br />
Sylvester’s canonical form for binary forms <strong>of</strong> odd degree.<br />
If ek ≡ 1, d = 2r, m = 1, ℓ1(x, y) = x and n = r, this<br />
recovers Sylvester’s canonical form for binary forms <strong>of</strong> even<br />
degree.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
A few remarks:<br />
This result follows easily from the end <strong>of</strong> the Ehrenborg-Rota<br />
paper, but does not seem to appear there explicitly.<br />
If ek ≡ 1, m = d + 1 and n = 0, this is the “Vandermonde<br />
basis” canonical form.<br />
If ek ≡ 1, d = 2r − 1, m = 0 and n = r, this recovers<br />
Sylvester’s canonical form for binary forms <strong>of</strong> odd degree.<br />
If ek ≡ 1, d = 2r, m = 1, ℓ1(x, y) = x and n = r, this<br />
recovers Sylvester’s canonical form for binary forms <strong>of</strong> even<br />
degree.<br />
If ek ≡ 1, so th<strong>at</strong> n = 2r and m = n − 2r, this interpol<strong>at</strong>es<br />
between the first two examples, and Sylvester’s algorithm can<br />
still be used. Let q = �m j=1 (βjx − αjy). Then q(D) kills the<br />
fixed d-th powers and q(D)p has degree d − m = 2r − 1.<br />
Write q(D)p as a sum <strong>of</strong> r 2r − 1-st powers and then<br />
“integr<strong>at</strong>e” to find an explicit represent<strong>at</strong>ion for p.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If ek ≡ 2, then an analogue to Sylvester’s canonical forms<br />
occurs for general forms <strong>of</strong> even degree d = 2k: they are the<br />
sum <strong>of</strong> the k-th power <strong>of</strong> ⌊(d + 1)/3⌋ quadr<strong>at</strong>ics plus a linear<br />
combin<strong>at</strong>ion <strong>of</strong> the d − 3⌊(d + 1)/3⌋ ∈ {0, 1, 2} 2k-th powers<br />
<strong>of</strong> specified linear forms, say, chosen from {x 2k , y 2k }. We<br />
don’t have an algorithm for this. We want one.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If ek ≡ 2, then an analogue to Sylvester’s canonical forms<br />
occurs for general forms <strong>of</strong> even degree d = 2k: they are the<br />
sum <strong>of</strong> the k-th power <strong>of</strong> ⌊(d + 1)/3⌋ quadr<strong>at</strong>ics plus a linear<br />
combin<strong>at</strong>ion <strong>of</strong> the d − 3⌊(d + 1)/3⌋ ∈ {0, 1, 2} 2k-th powers<br />
<strong>of</strong> specified linear forms, say, chosen from {x 2k , y 2k }. We<br />
don’t have an algorithm for this. We want one.<br />
If d = 4, m = 0, e1 = 2 and e2 = 1, a general binary quartic<br />
can be written as the sum <strong>of</strong> the square <strong>of</strong> a quadr<strong>at</strong>ic and<br />
the fourth power <strong>of</strong> a linear form. (We have an algorithm for<br />
this which shows th<strong>at</strong> it can be done in six different ways.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
If ek ≡ 2, then an analogue to Sylvester’s canonical forms<br />
occurs for general forms <strong>of</strong> even degree d = 2k: they are the<br />
sum <strong>of</strong> the k-th power <strong>of</strong> ⌊(d + 1)/3⌋ quadr<strong>at</strong>ics plus a linear<br />
combin<strong>at</strong>ion <strong>of</strong> the d − 3⌊(d + 1)/3⌋ ∈ {0, 1, 2} 2k-th powers<br />
<strong>of</strong> specified linear forms, say, chosen from {x 2k , y 2k }. We<br />
don’t have an algorithm for this. We want one.<br />
If d = 4, m = 0, e1 = 2 and e2 = 1, a general binary quartic<br />
can be written as the sum <strong>of</strong> the square <strong>of</strong> a quadr<strong>at</strong>ic and<br />
the fourth power <strong>of</strong> a linear form. (We have an algorithm for<br />
this which shows th<strong>at</strong> it can be done in six different ways.)<br />
If d = 6, m = 0, e1 = 3 and e2 = 2, then ( 6<br />
6<br />
3 + 1) + ( 2 + 1) =<br />
6 + 1 implies th<strong>at</strong> a general binary sextic can be written as the<br />
sum <strong>of</strong> the square <strong>of</strong> a cubic and the cube <strong>of</strong> a quadr<strong>at</strong>ic<br />
form. We don’t have an algorithm for this. We want one.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
More variables<br />
Why are canonical forms even an issue? The main reason is th<strong>at</strong><br />
maps which one would think have full range don’t. Apart from<br />
sums <strong>of</strong> squares, where the orthogonal group plays a role, the<br />
simplest example occurs in H4(C3 ). Since N(3, 4) = � � 6<br />
2 = 15, one<br />
expects th<strong>at</strong> a general ternary quartic could be written as<br />
p(x1, x2, x3) =<br />
5�<br />
k=1<br />
(αk1x1 + αk2x2 + αk3x3) 4<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
(7)
More variables<br />
Why are canonical forms even an issue? The main reason is th<strong>at</strong><br />
maps which one would think have full range don’t. Apart from<br />
sums <strong>of</strong> squares, where the orthogonal group plays a role, the<br />
simplest example occurs in H4(C3 ). Since N(3, 4) = � � 6<br />
2 = 15, one<br />
expects th<strong>at</strong> a general ternary quartic could be written as<br />
p(x1, x2, x3) =<br />
5�<br />
k=1<br />
(αk1x1 + αk2x2 + αk3x3) 4<br />
This would be a canonical form, if the partials with respect to the<br />
αkj’s <strong>at</strong> some chosen value would span H4(C 3 ). By apolarity, this<br />
means th<strong>at</strong> there should be no non-singular quartic which is<br />
singular <strong>at</strong> the five points αk = (αk1, αk2, αk3).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
(7)
More variables<br />
Why are canonical forms even an issue? The main reason is th<strong>at</strong><br />
maps which one would think have full range don’t. Apart from<br />
sums <strong>of</strong> squares, where the orthogonal group plays a role, the<br />
simplest example occurs in H4(C3 ). Since N(3, 4) = � � 6<br />
2 = 15, one<br />
expects th<strong>at</strong> a general ternary quartic could be written as<br />
p(x1, x2, x3) =<br />
5�<br />
k=1<br />
(αk1x1 + αk2x2 + αk3x3) 4<br />
This would be a canonical form, if the partials with respect to the<br />
αkj’s <strong>at</strong> some chosen value would span H4(C 3 ). By apolarity, this<br />
means th<strong>at</strong> there should be no non-singular quartic which is<br />
singular <strong>at</strong> the five points αk = (αk1, αk2, αk3).<br />
However, as Clebsch argued in the 1860’s, since N(3.2) = 6, any<br />
choice <strong>of</strong> five αk’s pass through a non-zero quadr<strong>at</strong>ic h(x1, x2, x3),<br />
and so h 2 will be apolar to all the partials and so (7) is not a<br />
canonical form.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
(7)
A few years l<strong>at</strong>er, Sylvester gave another pro<strong>of</strong>. Given<br />
p(x1, x2, x3) = �<br />
r+s+t=4<br />
4!<br />
r!s!t! arstx r 1x s 2x t 3,<br />
define the c<strong>at</strong>alecticant Hp as a quadr<strong>at</strong>ic form in 6 variables (or a<br />
6 × 6 symmetric m<strong>at</strong>rix defined linearly in terms <strong>of</strong> p.<br />
⎛<br />
⎜<br />
Hp = ⎜<br />
⎝<br />
a400 a220 a202 a310 a301 a211<br />
a220 a040 a022 a130 a121 a031<br />
a202 a022 a004 a112 a103 a013<br />
a310 a130 a112 a220 a211 a121<br />
a301 a121 a103 a211 a202 a112<br />
a211 a031 a013 a121 a112 a022<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
⎞<br />
⎟<br />
⎠
Under this definition, H (α·) 4 is a perfect square. Thus if p is a sum<br />
<strong>of</strong> five fourth powers, then rank(Hp) ≤ 5, so Hp is singular. This<br />
can’t happen for a general ternary quartic, for which the<br />
determinant is non-zero. This gives the algebraic rel<strong>at</strong>ion <strong>of</strong> the<br />
coefficients in (7).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Under this definition, H (α·) 4 is a perfect square. Thus if p is a sum<br />
<strong>of</strong> five fourth powers, then rank(Hp) ≤ 5, so Hp is singular. This<br />
can’t happen for a general ternary quartic, for which the<br />
determinant is non-zero. This gives the algebraic rel<strong>at</strong>ion <strong>of</strong> the<br />
coefficients in (7).<br />
Clebsch’s pro<strong>of</strong> and Sylvester’s pro<strong>of</strong> are really the same, because<br />
Hp(t1, . . . , t6) =<br />
[(t1x 2 1 + t2x 2 2 + t3x 2 3 + t4x1x2 + t5x1x3 + t6x2x3) 2 (D)]p.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Under this definition, H (α·) 4 is a perfect square. Thus if p is a sum<br />
<strong>of</strong> five fourth powers, then rank(Hp) ≤ 5, so Hp is singular. This<br />
can’t happen for a general ternary quartic, for which the<br />
determinant is non-zero. This gives the algebraic rel<strong>at</strong>ion <strong>of</strong> the<br />
coefficients in (7).<br />
Clebsch’s pro<strong>of</strong> and Sylvester’s pro<strong>of</strong> are really the same, because<br />
Hp(t1, . . . , t6) =<br />
[(t1x 2 1 + t2x 2 2 + t3x 2 3 + t4x1x2 + t5x1x3 + t6x2x3) 2 (D)]p.<br />
This becomes a serious topic in algebraic geometry, so I won’t talk<br />
about it here. Our 19th century ancestors saw th<strong>at</strong> funny things<br />
happen when (n, d) = (3, 4), (4, 4), (5, 4), (5, 3). In the early<br />
1990s, Alexander and Hirschowitz proved th<strong>at</strong> these are the only<br />
cases these funny things can happen.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Sums <strong>of</strong> two squares<br />
Suppose p ∈ H2r (C2 ), has even degree. Then the FTA (algebra<br />
this time) says th<strong>at</strong> p can be written as a product <strong>of</strong> 2r linear<br />
forms, which in general are pairwise linearly independent, so by<br />
pairing them <strong>of</strong>f, we have p = gh, where f , g ∈ Hr (C2 ) in<br />
essentially � � 2r−1<br />
different ways. Further, by the old algebraist’s<br />
trick:<br />
r<br />
�<br />
g + h<br />
p = gh =<br />
2<br />
� 2<br />
� �2 g − h<br />
+ .<br />
2i<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Sums <strong>of</strong> two squares<br />
Suppose p ∈ H2r (C2 ), has even degree. Then the FTA (algebra<br />
this time) says th<strong>at</strong> p can be written as a product <strong>of</strong> 2r linear<br />
forms, which in general are pairwise linearly independent, so by<br />
pairing them <strong>of</strong>f, we have p = gh, where f , g ∈ Hr (C2 ) in<br />
essentially � � 2r−1<br />
different ways. Further, by the old algebraist’s<br />
trick:<br />
r<br />
�<br />
g + h<br />
p = gh =<br />
2<br />
� 2<br />
� �2 g − h<br />
+ .<br />
2i<br />
A sum <strong>of</strong> two squares from Hr (C 2 ) is not a canonical form because<br />
it uses 2(r + 1) = (2r + 1) + 1 parameters. But, if (u, v) ∈ C 2 and<br />
u 2 + v 2 = 1, then<br />
a 2 + b 2 = (ua + vb) 2 + (va − ub) 2<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
(8)
Sums <strong>of</strong> two squares<br />
Suppose p ∈ H2r (C2 ), has even degree. Then the FTA (algebra<br />
this time) says th<strong>at</strong> p can be written as a product <strong>of</strong> 2r linear<br />
forms, which in general are pairwise linearly independent, so by<br />
pairing them <strong>of</strong>f, we have p = gh, where f , g ∈ Hr (C2 ) in<br />
essentially � � 2r−1<br />
different ways. Further, by the old algebraist’s<br />
trick:<br />
r<br />
�<br />
g + h<br />
p = gh =<br />
2<br />
� 2<br />
� �2 g − h<br />
+ .<br />
2i<br />
A sum <strong>of</strong> two squares from Hr (C 2 ) is not a canonical form because<br />
it uses 2(r + 1) = (2r + 1) + 1 parameters. But, if (u, v) ∈ C 2 and<br />
u 2 + v 2 = 1, then<br />
a 2 + b 2 = (ua + vb) 2 + (va − ub) 2<br />
(These correspond to gh = (ζg)(ζ −1 h) where ζ = u + iv; in terms<br />
<strong>of</strong> sos forms, when u, v ∈ R, all the represent<strong>at</strong>ions in (8) have the<br />
same Gram m<strong>at</strong>rix.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
(8)
Accordingly, for a general p, we can choose (u, v) so th<strong>at</strong> the<br />
coefficient <strong>of</strong> x r , say, disappears in the second sum, and<br />
p(x, y) =<br />
� r�<br />
k=0<br />
αkx r−k y k<br />
� 2<br />
+<br />
� r�<br />
k=1<br />
βkx r−k y k<br />
is another old canonical form for binary forms <strong>of</strong> even degree.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
� 2<br />
(9)
Accordingly, for a general p, we can choose (u, v) so th<strong>at</strong> the<br />
coefficient <strong>of</strong> x r , say, disappears in the second sum, and<br />
p(x, y) =<br />
� r�<br />
k=0<br />
αkx r−k y k<br />
� 2<br />
+<br />
� r�<br />
k=1<br />
βkx r−k y k<br />
is another old canonical form for binary forms <strong>of</strong> even degree.<br />
This also gives another pro<strong>of</strong> <strong>of</strong> the Basic Example.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
� 2<br />
(9)
Accordingly, for a general p, we can choose (u, v) so th<strong>at</strong> the<br />
coefficient <strong>of</strong> x r , say, disappears in the second sum, and<br />
p(x, y) =<br />
� r�<br />
k=0<br />
αkx r−k y k<br />
� 2<br />
+<br />
� r�<br />
k=1<br />
βkx r−k y k<br />
is another old canonical form for binary forms <strong>of</strong> even degree.<br />
This also gives another pro<strong>of</strong> <strong>of</strong> the Basic Example.<br />
Suppose we don’t just look <strong>at</strong> maps C N → Hd(C n ), but also maps<br />
from N-dimensional subspaces <strong>of</strong> C M for M > N. Here is the<br />
simplest non-trivial case:<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
� 2<br />
(9)
For which 0 �= c = (c1, . . . , c4) ∈ C 4 is it true th<strong>at</strong><br />
(α1x + α2y) 2 + (α3x + α4y) 2 ,<br />
is a canonical form for binary quadr<strong>at</strong>ic forms?<br />
4�<br />
cjαj = 0 (10)<br />
j=1<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
For which 0 �= c = (c1, . . . , c4) ∈ C 4 is it true th<strong>at</strong><br />
(α1x + α2y) 2 + (α3x + α4y) 2 ,<br />
is a canonical form for binary quadr<strong>at</strong>ic forms?<br />
Theorem<br />
4�<br />
cjαj = 0 (10)<br />
A general binary quadr<strong>at</strong>ic form can be written as in (10), unless<br />
c3 = ɛc1 and c4 = ɛc2, where ɛ ∈ {i, −i}.<br />
j=1<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
For which 0 �= c = (c1, . . . , c4) ∈ C 4 is it true th<strong>at</strong><br />
(α1x + α2y) 2 + (α3x + α4y) 2 ,<br />
is a canonical form for binary quadr<strong>at</strong>ic forms?<br />
Theorem<br />
4�<br />
cjαj = 0 (10)<br />
A general binary quadr<strong>at</strong>ic form can be written as in (10), unless<br />
c3 = ɛc1 and c4 = ɛc2, where ɛ ∈ {i, −i}.<br />
In other words, (10) is canonically a canonical form. In the special<br />
case (c1, c2, c3, c4) = (1, 0, i, 0), �4 j=1 cjαj = 0 ⇐⇒ α3 = iα1,<br />
which is the “silly” example from the introduction.<br />
j=1<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
We conjecture th<strong>at</strong> for d = 2r and “most” c = (c1, . . . , c2r ) ∈ C 2r ,<br />
r�<br />
j=1<br />
(α2j−1x + α2jy) 2r ,<br />
�2r<br />
j=1<br />
cjαj = 0 (11)<br />
is a canonical form for binary forms <strong>of</strong> even degree 2r. We’ve<br />
proved this for 2r = 2, 4. In the special case where cj ≡ 1, it is<br />
true for 2r = 6, 8.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Completing the d-th power, all props to Boris Reichstein<br />
Let me give yet another ridiculous (but constructive) pro<strong>of</strong> <strong>of</strong><br />
completing the square for a quadr<strong>at</strong>ic form. Suppose p ∈ H2(Cn ).<br />
Then we have:<br />
∂p<br />
= α1x1 + · · · + αnxn.<br />
∂xn<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Completing the d-th power, all props to Boris Reichstein<br />
Let me give yet another ridiculous (but constructive) pro<strong>of</strong> <strong>of</strong><br />
completing the square for a quadr<strong>at</strong>ic form. Suppose p ∈ H2(Cn ).<br />
Then we have:<br />
∂p<br />
= α1x1 + · · · + αnxn.<br />
∂xn<br />
If we assume αn �= 0, and let<br />
q(x1, . . . , xn) = p(x1, . . . , xn) − 1<br />
2αn (α1x1 + · · · + αnxn) 2 ,<br />
then it is easy to see th<strong>at</strong><br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Completing the d-th power, all props to Boris Reichstein<br />
Let me give yet another ridiculous (but constructive) pro<strong>of</strong> <strong>of</strong><br />
completing the square for a quadr<strong>at</strong>ic form. Suppose p ∈ H2(Cn ).<br />
Then we have:<br />
∂p<br />
= α1x1 + · · · + αnxn.<br />
∂xn<br />
If we assume αn �= 0, and let<br />
q(x1, . . . , xn) = p(x1, . . . , xn) − 1<br />
2αn (α1x1 + · · · + αnxn) 2 ,<br />
then it is easy to see th<strong>at</strong><br />
∂q<br />
∂xn<br />
= ∂p<br />
−<br />
∂xn<br />
∂p<br />
∂xn<br />
= 0 =⇒ q = q(x1, . . . , xn−1)!<br />
Now just iter<strong>at</strong>e this, losing one variable <strong>at</strong> a time, to get the<br />
traditional lower diagonal sum <strong>of</strong> squares.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
This argument can be repe<strong>at</strong>ed. Suppose p ∈ H3(Cn ) is cubic;<br />
then ∂p<br />
is a quadr<strong>at</strong>ic form, so we can complete the square in<br />
∂xn<br />
general, but now lets do it diagonally from left to right:<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
(αjjxj + · · · + αjnxn) 2 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
This argument can be repe<strong>at</strong>ed. Suppose p ∈ H3(Cn ) is cubic;<br />
then ∂p<br />
is a quadr<strong>at</strong>ic form, so we can complete the square in<br />
∂xn<br />
general, but now lets do it diagonally from left to right:<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
You can guess where this is going. Let<br />
Then<br />
q(x1, . . . , xn) = p(x1, . . . , xn) −<br />
∂q<br />
∂xn<br />
= ∂p<br />
−<br />
∂xn<br />
∂p<br />
∂xn<br />
(αjjxj + · · · + αjnxn) 2 .<br />
n�<br />
j=1<br />
1<br />
3αjn (αjjxj + · · · + αjnxn) 3 .<br />
= 0 =⇒ q = q(x1, . . . , xn−1).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
This argument can be repe<strong>at</strong>ed. Suppose p ∈ H3(Cn ) is cubic;<br />
then ∂p<br />
is a quadr<strong>at</strong>ic form, so we can complete the square in<br />
∂xn<br />
general, but now lets do it diagonally from left to right:<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
You can guess where this is going. Let<br />
Then<br />
q(x1, . . . , xn) = p(x1, . . . , xn) −<br />
∂q<br />
∂xn<br />
= ∂p<br />
−<br />
∂xn<br />
∂p<br />
∂xn<br />
(αjjxj + · · · + αjnxn) 2 .<br />
n�<br />
j=1<br />
1<br />
3αjn (αjjxj + · · · + αjnxn) 3 .<br />
= 0 =⇒ q = q(x1, . . . , xn−1).<br />
We repe<strong>at</strong> this running backwards inductively to get our second<br />
theorem from the introduction, which is an algebra prelim question<br />
in a 19th century gradu<strong>at</strong>e m<strong>at</strong>h program:<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Theorem<br />
A canonical form for the general cubic is<br />
p(x1, . . . , xn) = �<br />
(α i,j<br />
1≤i≤j≤n<br />
i xi + · · · + α i,j<br />
j<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
xj) 3
Theorem<br />
A canonical form for the general cubic is<br />
p(x1, . . . , xn) = �<br />
(α i,j<br />
1≤i≤j≤n<br />
i xi + · · · + α i,j<br />
j<br />
xj) 3<br />
Pro<strong>of</strong>.<br />
We can do this abstractly with Lasker-Wakeford, or directly by<br />
using the previous construction. We leave as an exercise the<br />
constant-counting via the combin<strong>at</strong>orial identity:<br />
�<br />
� �<br />
n + 2<br />
(j − i + 1) =<br />
3<br />
1≤i≤j≤n<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
There is a wonderful non-trivial way to complete the cube, but<br />
almost nobody knows it. It appears in a paper by Boris Reichstein<br />
from 1987 which (according to M<strong>at</strong>hSciNet) has had no cit<strong>at</strong>ions.<br />
It is a truly beautiful theorem, though it was not transparently<br />
presented and appears in the context <strong>of</strong> trilinear forms.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
There is a wonderful non-trivial way to complete the cube, but<br />
almost nobody knows it. It appears in a paper by Boris Reichstein<br />
from 1987 which (according to M<strong>at</strong>hSciNet) has had no cit<strong>at</strong>ions.<br />
It is a truly beautiful theorem, though it was not transparently<br />
presented and appears in the context <strong>of</strong> trilinear forms.<br />
Theorem (Reichstein)<br />
A general cubic p(x1, . . . , xn) can be written as<br />
n�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 3 + q(x1, . . . , xn−2). (12)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
There is a wonderful non-trivial way to complete the cube, but<br />
almost nobody knows it. It appears in a paper by Boris Reichstein<br />
from 1987 which (according to M<strong>at</strong>hSciNet) has had no cit<strong>at</strong>ions.<br />
It is a truly beautiful theorem, though it was not transparently<br />
presented and appears in the context <strong>of</strong> trilinear forms.<br />
Theorem (Reichstein)<br />
A general cubic p(x1, . . . , xn) can be written as<br />
n�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 3 + q(x1, . . . , xn−2). (12)<br />
We have N(n, 3) − N(n − 2, 3) = n3 +3n2 +2n<br />
6 − n3−3n2 +2n<br />
6 = n2 , so<br />
if (12) is generally possible, it is a canonical form. This can be<br />
verified by Lasker-Wakeford, specializing <strong>at</strong> x1, x2, x1 + kx2 + xk<br />
(for k ≥ 3), but Reichstein’s constructive pro<strong>of</strong> is prettier.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
A general pair <strong>of</strong> quadr<strong>at</strong>ic forms can be simultaneously<br />
diagonalized: for general p, there exist n linearly independent<br />
forms Lj(x) = � n<br />
k=1 αj,kxk and complex numbers cj so th<strong>at</strong><br />
∂p<br />
∂xn−1<br />
=<br />
n�<br />
j=1<br />
L 2 j ,<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
cjL 2 j ,<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Pro<strong>of</strong>.<br />
A general pair <strong>of</strong> quadr<strong>at</strong>ic forms can be simultaneously<br />
diagonalized: for general p, there exist n linearly independent<br />
forms Lj(x) = � n<br />
k=1 αj,kxk and complex numbers cj so th<strong>at</strong><br />
∂p<br />
∂xn−1<br />
=<br />
n�<br />
j=1<br />
L 2 j ,<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
cjL 2 j ,<br />
Mixed partials are equal, so 2 �n j=1 αj,nLj = 2 �n the Lj’s are linearly independent, so αj,n = cjαj,n−1. Now,<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms<br />
j=1 cjαj,n−1Lj and
Pro<strong>of</strong>.<br />
A general pair <strong>of</strong> quadr<strong>at</strong>ic forms can be simultaneously<br />
diagonalized: for general p, there exist n linearly independent<br />
forms Lj(x) = � n<br />
k=1 αj,kxk and complex numbers cj so th<strong>at</strong><br />
∂p<br />
∂xn−1<br />
=<br />
n�<br />
j=1<br />
L 2 j ,<br />
∂p<br />
∂xn<br />
=<br />
n�<br />
j=1<br />
cjL 2 j ,<br />
Mixed partials are equal, so 2 �n j=1 αj,nLj = 2 �n the Lj’s are linearly independent, so αj,n = cjαj,n−1. Now,<br />
q(x1, . . . , xn) = p(x1, . . . , xn) −<br />
=⇒ ∂q<br />
∂xn−1<br />
= ∂q<br />
∂xn<br />
n�<br />
j=1<br />
j=1 cjαj,n−1Lj and<br />
1<br />
3αj,n (αj,1x1 + · · · + αj,nxn) 3<br />
= 0 =⇒ q = q(x1, . . . , xn−2).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
We conclude the talk by veering <strong>of</strong>f in another direction.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
We conclude the talk by veering <strong>of</strong>f in another direction.<br />
Imagine a general canonical form for quartics <strong>of</strong> “Reichstein-type”<br />
p(x1, . . . , xn) =<br />
r�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 4 + q(x1, . . . , xm). (13)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
We conclude the talk by veering <strong>of</strong>f in another direction.<br />
Imagine a general canonical form for quartics <strong>of</strong> “Reichstein-type”<br />
p(x1, . . . , xn) =<br />
r�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 4 + q(x1, . . . , xm). (13)<br />
We would need N(n, 4) = r × n + N(m, 4).<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
We conclude the talk by veering <strong>of</strong>f in another direction.<br />
Imagine a general canonical form for quartics <strong>of</strong> “Reichstein-type”<br />
p(x1, . . . , xn) =<br />
r�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 4 + q(x1, . . . , xm). (13)<br />
We would need N(n, 4) = r × n + N(m, 4).<br />
It turns out th<strong>at</strong> if n = 12, there does not exist m < 12 so th<strong>at</strong><br />
� � � � �<br />
�<br />
12 �<br />
15 m + 3<br />
� − ,<br />
4 4<br />
so number theory rules out universal Reichstein-type canonical<br />
forms for quartics.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Reichstein’s construction gives a cubic as a sum <strong>of</strong> roughly n2<br />
4<br />
cubes, which is more than the actual minimum <strong>of</strong> roughly n2<br />
6 .<br />
We conclude the talk by veering <strong>of</strong>f in another direction.<br />
Imagine a general canonical form for quartics <strong>of</strong> “Reichstein-type”<br />
p(x1, . . . , xn) =<br />
r�<br />
k=1<br />
(αk1x1 + · · · + αknxn) 4 + q(x1, . . . , xm). (13)<br />
We would need N(n, 4) = r × n + N(m, 4).<br />
It turns out th<strong>at</strong> if n = 12, there does not exist m < 12 so th<strong>at</strong><br />
� � � � �<br />
�<br />
12 �<br />
15 m + 3<br />
� − ,<br />
4 4<br />
so number theory rules out universal Reichstein-type canonical<br />
forms for quartics.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Now define<br />
�<br />
Ad = n : 0 ≤ m < n =⇒ n � | � � � � n+d−1 m+d−1<br />
d − d<br />
�<br />
.<br />
We have a few partial results.<br />
If 3 � | k, then n = 2 2k · 3 ∈ A4.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Now define<br />
�<br />
Ad = n : 0 ≤ m < n =⇒ n � | � � � � n+d−1 m+d−1<br />
d − d<br />
�<br />
.<br />
We have a few partial results.<br />
If 3 � | k, then n = 2 2k · 3 ∈ A4.<br />
If p ≡ 1 (mod 144) is prime, then 12p ∈ A4.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Now define<br />
�<br />
Ad = n : 0 ≤ m < n =⇒ n � | � � � � n+d−1 m+d−1<br />
d − d<br />
�<br />
.<br />
We have a few partial results.<br />
If 3 � | k, then n = 2 2k · 3 ∈ A4.<br />
If p ≡ 1 (mod 144) is prime, then 12p ∈ A4.<br />
If p is prime, then p | � � � � n+p−1 n<br />
p − p , hence Ap is empty for<br />
prime p (such as p = 2, 3.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Now define<br />
�<br />
Ad = n : 0 ≤ m < n =⇒ n � | � � � � n+d−1 m+d−1<br />
d − d<br />
�<br />
.<br />
We have a few partial results.<br />
If 3 � | k, then n = 2 2k · 3 ∈ A4.<br />
If p ≡ 1 (mod 144) is prime, then 12p ∈ A4.<br />
If p is prime, then p | � � � � n+p−1 n<br />
p − p , hence Ap is empty for<br />
prime p (such as p = 2, 3.)<br />
The smallest elements <strong>of</strong> A6, A8, A10, A12, A14 and A15 are 10,<br />
1792, 6, 242, 338 and 273 respectively. If A9 or A16 are<br />
non-empty, then their smallest elements are <strong>at</strong> least 10 5 .<br />
(Fortun<strong>at</strong>ely, steampunk allows M<strong>at</strong>hem<strong>at</strong>ica.)<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms
Now define<br />
�<br />
Ad = n : 0 ≤ m < n =⇒ n � | � � � � n+d−1 m+d−1<br />
d − d<br />
�<br />
.<br />
We have a few partial results.<br />
If 3 � | k, then n = 2 2k · 3 ∈ A4.<br />
If p ≡ 1 (mod 144) is prime, then 12p ∈ A4.<br />
If p is prime, then p | � � � � n+p−1 n<br />
p − p , hence Ap is empty for<br />
prime p (such as p = 2, 3.)<br />
The smallest elements <strong>of</strong> A6, A8, A10, A12, A14 and A15 are 10,<br />
1792, 6, 242, 338 and 273 respectively. If A9 or A16 are<br />
non-empty, then their smallest elements are <strong>at</strong> least 10 5 .<br />
(Fortun<strong>at</strong>ely, steampunk allows M<strong>at</strong>hem<strong>at</strong>ica.)<br />
Thus the way is clear for a Reichstein canonical form for quintics. I<br />
hope one <strong>of</strong> you in the audience can find it.<br />
Bruce Reznick <strong>University</strong> <strong>of</strong> <strong>Illinois</strong> <strong>at</strong> <strong>Urbana</strong>-<strong>Champaign</strong> Steampunk canonical forms