algebraic geometry - university of washington€¦ · algebraic geometry a personal view cse 590b...
TRANSCRIPT
Algebraic Geometry A Personal View
CSE 590B
James F. Blinn
Mailing List
Subscribe at
https://mailman.cs.washington.edu/
mailman/listinfo/cse590b
University of Michigan
1967-1974
University of Michigan
1967-1974
Gordon Romney (U Utah) 1969
+ Appendix
University of Utah 1974-1977
JPL/Caltech 1977-1995
JPL/Caltech 1977-1995
Voyager Cosmos
The Mechanical Universe Mathematics!
Render 3D Objects
Planar Polygons = First Order Surfaces
Render 3D Objects
Second Order Surfaces
Render 3D Objects
Third (and higher) Order Surfaces
UM, UU, JPL, Microsoft and Now
1962-present
Studying Algebraic Geometry
Algebraic Equations
Geometric Shapes
Making Algebraic Geometry
More Understandable
Jim Blinn’s Corner Articles 1987 - 2007
Many of them on Algebraic Geometry
Why Am I Here
• Share my enthusiasms
• Help me organize my ideas
I work better if I have an audience (M.B.)
Updates to old articles
Unpublished articles
Keep me from repeating myself
Publish on web site
• One Session every 2 weeks
• Later meetings may get more sketchy
• Discuss open questions
Why Are You Here
•Varied Audience
• Go slowly at first
• Prerequisites:
• vectors and matrices
• homogeneous coords
•See old stuff in new ways
•See new stuff
What I will talk about
•Real Algebraic Projective Geometry
• Real is more complex than Complex
• Projective is simpler than Euclidean
•Dimension 1,2,3
•Lowish Order Polynomials
•Notation, notation, notation
•Lots of Pictures
Pictures? Why no
can fool you
show only special cases
hard to generalize to high dimensions
hard to make
forces you to think (visualize internally)
Why yes
intuition
see patterns
I am visual thinker (see patterns)
pretty
Hartshorne vs. Abraham&Shaw
2 2 1X Y
Relation Between
Algebra and Geometry
2 24 1X Y
Relation Between
Algebra and Geometry
Relation Between
Algebra and Geometry
2 24 1X X Y
General Quadratic Curve
2 22
2 2
0
A X B XY CY
D X EY
F
Quadratic Curve
2 22
2 2 0
A X B XY CY
D X EY F
2 2 2... 2ACF BED D C E A B F D
Discriminant , , , , , 0A B C D E F D
Cubic Curve 3 2 2 3
2 2
3 3
3 6 3
3 3 0
AX BX Y CXY DY
EX FXY GY
HX JY K
1
Discriminant of Cubic
, , , , , , , , , 0A B C D E F G H J K D
G. Salmon (1879):
1
4 4 4 4 3 3
4 2 2 2 2 3 3 3 3
2 3 3 2 3 3 2 2
3 3 3 3
12
36 64
192 192
64 ...
A D K A D K GJ
A D K G J A D K F
A D K F BE AD K FB E
D K B E
D
D has over 10,000 terms
Discriminant of Cubic
3 264S T D
S: degree 4 in A…K
has 25 terms
T: degree 6 in A…K
has 103 terms
1
Want Better Notation
Notation = Creative Abbreviation
a c b e
f h d k
ab cd e
fb hd k
M v w
N M v NM v
Review of Typical Notation
And some snags
2D Euclidean Geometry
X
Y
X Y X YP
What Went Wrong?
Top View Front View (Post Perspective)
2D Projective Geometry
3D Algebraic Objects
X
Y
x y
w w
x y wP
x y w
Equation of a Line
L
x y wP
0ax by cw
0
a
x y w b
c
Equation of a Line
a
b
c
L
x y wP
0
a
x y w b
c
0 P L
Row/column standardization?
0ax by cw
Two Points Make A Line
S
L
P
L P S
0
0
P P P
S S S
ax y w
bx y w
c
P P P
S S S
a x y w
b
c x y w
, ,P S P S P S P S P S P Sa y w w y b w x x w c x y y x
Two Lines Make A Point
M
P
L
0 0
L M
L M
L M
a a
x y w b b
c c
P L M
L M
L M
L M
a a
x y w b b
c c
Transforming Points
ˆPT P
11 12 13
21 22 23
31 32 33
ˆ ˆ ˆ
T T T
x y w T T T x y w
T T T
Transforming Lines
0 P L
1 0 PT T L
1 T L L
1 0 P TT L
0 P L
PT P
Matrix Adjugate (fka Adjoint)
1
2
3
R
R
R
T *
2 3 3 1 1 2R R R R R R
T
*
det 0 0
0 det 0
0 0 det
T
TT T
T
1 0 0
det 0 1 0
0 0 1
T
?
Transforming Points and Lines
* T L L
* * *11 12 13
* * *21 22 23
* * *31 32 33
ˆ
ˆ
ˆ
aT T T a
T T T b b
T T T c c
PT P 11 12 13
21 22 23
31 32 33
ˆ ˆ ˆ
T T T
x y w T T T x y w
T T T
Point on Quadratic Curve 2
2
2
2 2
2
0
Ax Bxy Cxw
Dy Eyw
Fw
0
A B C x
x y w B D E y
C E F w
0T P Q P
P
Q
Transforming a Quadratic
0T PQP
* * 0T
T P TT Q TT P
* *T T QT Q
* * 0TT PT T QT PT
0T PQP
PT P
Given Point, Find Tangent
P
Q
T P QP
0 T PQP
P L
Given Point, Find Tangent
P
Q
L = Q PT T P QP
0 T PQP
P L
Is a Line Tangent to Q
L
Q
*0 T L Q L
Given Tangent, Find Point
L
Q
P L
*T L Q L
*0 TL Q L
Given Tangent, Find Point
L
Q
LT Q
* =P
*
*
0 T
T
L Q L
L Q L
P L
Three Kinds of Matrix
point point T
point lineT
Q
*line pointT Q
Point on Cubic Curve
3 2 2 3
2 2
2 2
3
3 3
3 6 3
3 3
0
Ax Bx y Cxy Dy
Ex w Fxyw Gy w
Hxw Jyw
Kw
P=[x,y,w]
C
Forms of Cubic Curve Equation 3 2 2 3
2 2
2 2
3
3 3
3 6 3
3 3
0
Ax Bx y Cxy Dy
Ex w Fxyw Gy w
Hxw Jyw
Kw
0
A B E B C F E F H x x
x y w B C F C D G F G J y y
E F H F G J H J K w w
0T T PCP P
Forms of Cubic Curve Equation 3 2 2 3
2 2
2 2
3
3 3
3 6 3
3 3
0
Ax Bx y Cxy Dy
Ex w Fxyw Gy w
Hxw Jyw
Kw
0
A B E B C F E F H x x
x y w B C F C D G F G J y y
E F H F G J H J K w w
, ,
, ,
0i j k i j k
i j k
PP P C
Two Problems With Notation
A B E B C F E F H
B C F C D G F G J
E F H F G J H J K
C
point lineT
Q
Row vs. Column Confusion
Handing More Than Two Indices
The Solution
• Steal Notational Tricks from Physics
– General Relativity
– Quantum Mechanics
• Tuned to Algberaic Geometry
Old Index Types
1 2 3P P PP
1
2
3
L
L
L
L
Column
Row
New Index Types
1 2 3P P P P
1 2 3L L LL
CoVariant
ContraVariant
The Multiplication Machine
1 2 3
1 2 3
i
i
i
P L P L P L
P L
P L
Einstein
Index
Notation
1
1 2 3 2
3
L
P P P L
L
P L
Three Kinds of Matrix
ijQPure covariant
*ij
Q
Pure contravariant
i
jTMixed
Three Kinds of Matrix
i
ij jP Q L
*ij
j
iL Q P
j i i
jP T P
General Tensor Contraction
k lu j l
ij km u imF H R S W
Taking More Ideas from Physics
Writing Tensor Contraction in
Diagram Form
P L
P L
Three Kinds of Matrix
i
ij jP Q L
*ij
j
iL Q P
ˆj i i
jP T P Pij
=~ i
P T
ji=
jP Q L
ji=
jL Q* P
General Tensor Contraction
k lu j l
ij km u imF H R S W
=k
i
lF
R j
H
u
S
m
W
il
m
Don’t need index labels
=F
R
H
S
W
Just be careful about matching dangling arcs
Sum of Terms
= +P R T S
P RT S
i j i i
jP R T S
Consistent type evident
Scalar Product
P R S
R S
P R S
Only Connectivity Matters
=P T L
L P
T
Rearranging internal arcs/nodes doesn’t change value
Now Back To Geometry
Point on a Line
0ax by cw 0
a
x y w b
c
0 P L
0i
iP L P L = 0
L
P
Point on a Quadratic Curve
2
2
2
2 2
2
0
Ax Bxy Cxw
Dy Eyw
Fw
0
A B C x
x y w B D E y
C E F w
0T P Q P
0i j
ijP Q P P Q P = 0
P
Q
Point on a Cubic Curve 3 2 2 3
2 2
2 2
3
3 3
3 6 3
2 3
0
Ax Bx y Cxy Dy
Ex w Fxyw Gyw
Hxw Jyw
Kw
0
A B E B C F E F H x x
x y w B C F C D G F G J y y
E F H F G J H J K w w
0i j k
ijkP P P C P = 0C
P
P
P
C
Transforming a Point
PT P
i j j
iP T P
P T = P~
Transforming a Line
* T L L
*i
i jjT L L
=LT*
L~
Transforming A Quadratic Curve
* *T
T Q T Q
* *i j
ij klk lT Q T Q
= QQT*
T* ~
Transforming A Transformation
* T MT M
*i lj l
i kjkT M T M
= MMT* T ~
Transforming a Cubic Curve
* * *i j k
ijk lmnl m nT T T C C
=CT*
T*
T*
C~
General Transformation Rule
DT*
T*
T
TÞ D ~
D
Dot and Cross Product
L
P
S
?
becomes
P L = s
Levi-Civita Epsilon
123 231 312
321 132 213
1
1
0 otherwiseijk
0 0 0 0 0 1 0 1 0
0 0 1 0 0 0 1 0 0
0 1 0 1 0 0 0 0 0
Cross Product
0 0 0 0 0 1 0 1 0
0 0 1 0 0 0 1 0 0
0 1 0 1 0 0 0 0 0
S
P P P S
S
P S P S P S P S P S P S
x
x y w y
w
y w w y w x x w y x x y
i j
ijk kP S L
Levi-Civita Epsilon Diagram
ijk
jk
i
Levi-Civita Epsilon Diagram
ijk
jk
i
Cross Product
1
1 2 3 1 2 3
2
3
L
P P P S S S L
L
P S L
i j
ijk kP S L
P
S
= L
i
j k k
Anti-Symmetry and Epsilon
P S S P
P S S P
Mirror Reflections flip sign
AxA=0
A A A A
A A
0
Two Types of Epsilon
ijk
jk
i
ijkjk
i
CONTRAvariant
COvariant
The Other Cross Product
1 1
1 2 3
2 2
3 3
L M
L M P P P
L M
L M P
ijk k
i jL M P
L
M
= P
i
j k k
Triple Product
L M N M N L N L M
L
M
N
P R S R S P S P R
P
R
S
P
R
S
L
N
M
PRS
LMN
Generating Algebraic
Relations Between Diagrams
A
B C
D
A
B C
A
B C
D D
? ?
Linear Combinations of Points
B
BR
R
B
R
BR gG
BR
G
g
Linear Combinations of Points
g
g
B
W R
G
det
,
det
W
R
G
B
R
G
det
,
det
B
W
G
B
R
G
det
,
det
g
B
R
W
B
R
G
Cramer’s Rule
Basic Linear Relationship
det
,
det
W
R
G
B
R
G
det
,
det
B
W
G
B
R
G
det
,
det
g
B
R
W
B
R
G
det
B
R
G
det
W
R
G
det
B
W
G
det
B
R
W
g
Note Symmetry
Grassman-Plucker relation
Arc Swapping Identity
Arc Swapping Identity - Variations
Swap heads
Swap tails
Swap heads (dual)
An Application of Arc Swapping
M
M
M
An Application of Arc Swapping
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
M
An Application of Arc Swapping
M
M
M
M
M
M3
* detMM M I
M
M
M
M
M
M
M
M
M
M
M
M
Compare with:
Relation of Diagram to Adjugate
M
Mn
i
k
m
l
jij kl
mn ikn ljmD M M
M* 2
M
M
33 22 33 22321 231
22 33 22 33
231 321 22 33 32 23
11 23 32 23 32
231 231
32 2332 23
321 321
2
M M M M
M M M MD M M M M
M M M M
M MM M
Example element:
Constant factor
Actual adjugate
Adjugate and Determinant
M
M
M MM =det M
M* 2
Another Arc Swap Application
Another Arc Swap Application
2
Another Arc Swap Application
2 2
Epsilon-Delta Rule
jklm
jl
km
jm
kl
m
j
j j
l
k k k
l lm m
j
ii
j
Algebraic Interpretation
A B C A C B A B C a f a f a f
AA
B BC C
A
B C
Projection from L thru C onto G
L
S
C
G
Projection from L thru C onto G
L
S
C
GR
G
L C
=S
G R S
L C R
Projection from L thru C onto G
L
S
C
S L C
Projection from L thru C onto G
L
S
C
G
G
L C
G
L C
S
0 S G
S C G L L G C
C G
L G
S L C
0 L G C G
Shadow Projection Matrix
L
S
C
G
GL
C
= S
GL
T
TC S
An Important Identity
T
TT
What is transformed Epsilon?
Use Modification of EpsDel
Apply to Transformed Epsilon
T
TT
T
TT
T
TT
Mirror Reflection = Change Sign
T
TT
T
TT
T
TT
T
TT
T
TT
T
TT
Move over = sign
T
TT
T
TT2
T
TT
T
TT
T
TT
Recall previous diagram
T
TT
T
TT2
2
T
TT det T
T
TT 2
2det T
}
An Important Identity
T
TT
=
det T
Diagram of
Transformed Quadratic Determinant
Q
Q
QT
T
T
T
T
T Q
Q
Q
Q
Q
Q
(det T)2
T
T
T
T
T
T Q
Q
Q
Weight
MAJOR PUNCHLINE
Of all the Gazillion possible
polynomials in the
coefficients
Tensor Diagrams express
only those that represent
Invariant Properties Q
Q
Q
Trace of Matrix
trace i
i
i
TT =T
=Q
?trace ii
i
Discriminant of Cubic
1
C C C 0
C
C
Discriminant of Cubic
1
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C C
CC C
C
Discriminant of Cubic
C
C
C
C
124S
3 264S T D
S: degree 4 in A…K
has 25 terms
T: degree 6 in A…K
has 103 terms
C C
CC C
C
16T
1
“Phase space” of cubics
C C
CC C
C
C
C
C
C
C C
CC C
C
C
C
C
C
C
C
C
C
D =
C C
CC C
C
C
C
C
C
, C C
CC C
C
C
C
C
C
C
C
C
C
{ ,}
3 264S T D
History of Diagrammatic
Invariant Notation
1878 Sylvester & Clifford
1885 Kempe
.
.
1989 Olver & Shakiban
1990 Stedman
1992-2007 Blinn
2011 Richter-Gebert
Effect of Changes
Q
Q
Q
Geometric Transformation
Homogeneous Scale
Q
Q
QsQ
sQ
sQ
= s3
Q
Q
Q
(det T)2
What Stays Constant?
Q
Q
L L
Q
Q
Q
Zeroness
Sign
0
0
0
0
0
2 2 22ACF BED D C E A B F
Odd number
of nodes
Even number
of nodes
Where do we go from here
Other Dimensions
2 2, , ...f x y w Dx Eyw Fw Curves in P2
3D algebra
Surfaces in P3
4D algebra 2, , , ...f x y z w G x H yw J zw
Polynomials in P1
2D algebra 2 2,f x w Ax Bxw Cw
The Grid 2D=P1
Point sets on
line
3D=P2
Curves in
plane
4D=P3
Surfaces in
space
LINEAR
QUADRATIC
CUBIC
QUARTIC
etc
1
Order of traversal?
Other Dimensions
P L
2 :D ax bw
3 :D ax by cw
4 :D ax by cz dw
Same Across Dimensionality
P L = 0
P Q P = 0
P = 0C
P
P
= C'CT*
T*
T*
F
P
PP
P
P Q = L
Dimensionality and Epsilon
ijk
i
k
j
l
ijklijk
i
kj
ij
i j
3D algebra
2D geometry
2D algebra
1D geometry
4D algebra
3D geometry
Previews of Coming Attractions
Discriminant Surface
2 2 3 3 2 26 4 4 3 0A D ABCD AC B D B C
Resultants
Q
Q
C CQ C Q C QQ 8 ,Q C 4 3
C
Q
Theorem of Pascal
B
A
FE
D
CP1
P2
P3
B D
A
D
F
E
C
B
EA
C F
= 0
5 Points Determine a Quadratic
B
A C
D
CA
B D
E
E
= Q
D
B A
C
AB
D C
E
E
B
A
D
C
E
Intersecting Two Quadratic Curves
Q R
Þ
QR QR
QR QR
P1
P2
P3
P4
QR
QR
QR
QRRank
Q R
Analyzing Cubic Curves
KL
M
N
C
LM
N
KK
K
Parametric Curves
= x,y,w
p
Q
p
x,y,wQx,y,w x,y,wx,y,w
Q Q
Q Q
=
Q
Q
Q = 0
Q
Q
Implicit
Degeneracy:
Base Point if
Parametric
Group Structure of Cubic
=
a b
a b
C
C
C
C
c
ab
c
u
a+b
=
c u
c u
C
C
C
C
a+b
Three Dimensional Projective
Geometry
= E
P
S
T
=S
T
L
Q
Q
Q
Q
3 Points = A Plane
2 Points = A Line
Discriminant of
Quadric
Three Skew Lines
M
L
K
Q
K
L
M
Q KL
M ML
K
Steiner Surfaces
f
f
f
F1
p
p
fParametric
F1F1
F1 F1
F4
Tangent
Implicit
Tensor Diagrams
• Keep Track of CoVariant/ContraVariant
Pairings
• Represent Higher Order Curves Nicely
• Express Only Invariant Quantities
• Allow for Algebra on These Quantities
• Are coordinate free
• Allow us to feel really cool at sharing
notation with Einstein and Feynman