Cohomological
Field
Theories Based
from jointon work
with B. Kim
Gauged & with
Linear g. can-Fontanne -
G- uéré -Kim
Sigma - shoemaker
Models
I : GLSMS
II : COHFTS
III : History
II ! Construction
I : GLSMS ( Roughly )
A GLSM is a G- IT quotient
of affine space with a global
function .
c- → ¢
where
• 11/10-6 is a smooth
DM stack
• Zcaw) ≤ V%G is proper
Data : ( V , G, O , W )
• V Ctvector space
• G- ≤ Glad
• ⊖ : G- → character
• wE§ymVYG isfunctiona G-inv
G-ITQvo-t.int
V70 ) : =
• f-* Symv /
{REV / •• flyf- (v) -1-00-491
VK.ci. = @ "%. ]
( quotient stack)
Een [1/10]
Examples of affine Gltq what
• Projective space .
• Grassmann / ans
• semi projective toric varieties
• Quiver varieties
By varying ⊖ c- ( V. GO,w)
GLSMS specialize to
• Complete intersections in
all of the above
• Quantum singularities (Efate)
w :[%-) → ¢
To specialize to complete
intersections in G-
Choose
• W a G- representation
• s ←Gym IN w)G
Gives
• → w Not
• <s , -> : totw"→ v16 G-
Form the GLSM
(v⊕W!G , 0 , < si>)
2- (s ) ≤ G
GLSMS ( Precise]
Data :( 11,721, wir)
• 1) ¢ - vector space
• P≤ Gen) linear red. gp.
• Xt M → surjective character
• V a Q1 - character of
Define fr extends
• G- :=kerX G- by "R-cha.ge")
• D- : = VIG
Require
• Vˢˢ(v1 -_ 11%-2.us/o-1fYgo:ki-toff ")
• 2- (dw) is proper
Goat : Produce an
enumerative theory
for G-↳ Ms
• specializes to
Gromov - Witten theory
for complete intersections
in G-
• specializes to FJRW
theory for quantum singularities
(Ende) [VIG] -4¢
• Varying 0 interpolates
between the above
2¥ ①✗ @ an +2
= Spec Q [Xo , . . . , ✗n , p ]
weights : 1 , . _ ^ , 1 , - d
a) 0--1 : ①×
b) f- : a.✗ is
-2 ↳ t - '
• f-(x) ≈ homogeneous polynomial of degree of
• w ≈ = Pf (X)
a) ex = tot Opntd ¢
Gives G-W theory of 2-(f) ≤ 1pm b) an-%¢✗=[Aⁿ%d] ¢
Gives FJRW theory
TT-coht-TSDefn-AC-homologica-fieldtheory-i.si.
• H a graded a-vector space
• 4-,-7 a supercommutate pairing
• 7- EH lunit)
• Rg,:D: H←→H*(MgT)
• satisfying natural axioms
• permutation covariance
• tree
• loop
• forgetting tails
• metric
EI 1GW theory of Z)
Let 2- be a smooth variety Mg,!aZ ) = Moduli of maps
e. → z
• 741=1-1*12-7 (stale space)
• LY ,V2> _' = SV , u V2
2- n[m jir
ev¥H¥Mg,r,dlZY→H*Mg%↓for*
1-1*1%7
/IS PD
1*-1-1*174 - - if
?ᵈ µ*Ñsid
EI CFJRW theory]
• H : =get④ Jae (Wtvg ) dwg
14 * ? HHMJ.ir )
Observe :
* =
g-Q-o-J-aclwtpgldwg-g-Q.o.HN?Hldw1ygY )
= 1-1*1=11%-7*1duty
In general , given * -90
there is a twisted Hodge complex:
@*%r:i-% . . -7k¥ ]
if W = 0
[ 0*-3 R'✗ → . . . -5s:)
1-1-1*1541,01=+0+1*4*1
If W 13 an isolated singularity in 11th
Idw is regular
& (R•* , Idw) = * (dw)"-5Jackdaw
H*( Rj , rdw) = Jacko)dw
In general , given A, Gaw)
define
H-tt.HR,=%o-• Mdw)
14%-11-1*19Kiimeth ?-µqᵈw* )
Formula
Observation
If ✗ c- HIM
.si#,dw)Bc-H1*lRj-,dv)=arBc-1H*CA*,d(w+vD-dS
%= -?^dcw+vandWAB -1 ✗) ABAD )
Idea : Let
Marie __MgridᵗᵗlV,FQw) be the
moduli space of LG quasi maps
to he → 2-law)G
• Embedd Mg,r,d→Ug,r;d¥¥¥?
↓ en er '
1%6-7 ¢¥> Most
• Construct a virtual cycle in
twisted Hodge cohomology of Vg,r,d
[ M],, c- It/Mgird*(RugBe .me ,rdlÑ◦w))
Supported on Mg,rid
( which is proper by FJR 2018)
G- LSM Invariants
1[mjrir9144
Hf*Ngµd,AdlÑ◦w☒Y) -711-1*14.ro/Mgird
↑ñ* ↓f%*
11-1*6.
ygy.gr ±%EMᵈw☒%?-4%1-41%1;)
# History d- enumerative
invariantsofGLS.MS
• Fan-Jarvis-Ruan 12013) G- finite
• Polishchub-Vainttib (20/6) Purely algebraic
version
usingfactorizationsmatrix
• Kiem -Li 12018) using cosection localization
• Fan-Jarvis -Ruan 12018) "Narrowsectors"
for general
GLSMS
• G-•can-Fontanne-F- Gvéré - convex hybrid
Kim -Shoemaker 120181 models
• F- - Kim 12020) General
case
Thin ( CF -1=-0--1-1--5,2018)
Enumerative invariants for
convex hybrid models specialize
to FJRW theory
and to GW ' theory
as defined using the co section
localized virtual cycle .
Thm (Kim-04,2018) The cosection
localized virtual cycle agrees with
the Behrend- Fantechi virtual cycle.
Thm ( F- Kim, 2020) Thegeneral
GLSM invariants form a CohFT.
Construction
• Embedd Myrie in a tm°°ᵗh
Space
Mg,r,d -7 Ug,r,d
\
, .
. _
- -
"
I
, .FI
↓ →n÷
I%-% -
• Find a "virtual " matrix factorization
1kg,r,d on ( Ug ,r,d,Ѱw☒r)
supported
• [m ]; a := tdlbg.r.ae/chlKgisd)FeEetined
Construction of 1kg ,r,d
LG quasimap data :
• C genus g curve
• q-tgy-n.ir) marked points
• É principal f- bindle
• K : px
, Cly → WEE
• u :c → PIV
vector
87% bundle:
↓
L universal
curve
↓'
{ ( C , q , F , K) }
d
RiT*&×n%) = [ A → B)
RIT* (8%10)=[1--413]
her D= til 88%-7← extra
wantconstructtoa toÉp*Bᵈ*a
c.se#w ↓ ↑ '
using
≤Ugarit c-
"
'tot A
2- (B)
" 24%8, Huh ↓p
= M " LUGG)
{ ( C , q, 8,15 ) }
01
zlt) =Mig,;D
odd 2
A ≈[Ñf*B←→Ap*B)2
2 =) B -11 ✗ Supp= MgCK)-2-1%7
,rid
Problems : In general ✗
only exists locally on Ugigd
a-EIHYUg.ve/KlBH
Want: To find a matrix factorization which is
locally the Ko szul factorization
%¥T%Factorizations come
from sheaves of CDGAS
:
• * DM stalk
• (A, d) sheaf of CDGAS over
• we MCX , 0×3
• a C- M( ✗ A-,) s.t.dk)=w
Then even -2, odd
A ⇐ A
2 =D -1 • a
graded leibnitz rule 22__ w
- '
. this is a matrix factorization
- Idea :
To realize
✗ e-HI ' ( tot A)¥1B))
we need to replace 75lb)
by a d- acyclic complex
but retain the CDGA structure
Thom-Sullivan
✗ (B) É%%• # B) →Th%%B
e- acyclic P5h%¥
coslhplicial
sheaf of ¥817S
CDGAS
Then
• ✗ c- Tti ÉKIBL ,
• and alla) = w
1kg,r,d:=&h•G%TBÑ¥Hh%*Ñ
locally looks like
" p*Ñ→←iᵈᵈp* ]