discussion hugh thomas probabilistic
TRANSCRIPT
Discussion of Hugh Thomas' talk
A probabilistic perspective on ol ' theory
im Tom KLOSE Amphituhedron DayNikolas TAPIA 8 April 2022
I
• STOCHASTIC QUANT ISAT 1 ON goes back to [ PARISI & WU,1981] i
consider Euclidean QFT measure as stationary measure of a stochastic process
* static picture : old f) i = exp f- 54 ) ) d¥ ( formal !)
• functional derivative* Langevin dynamics :
'¥4 = - 81¥ t 3.
10=0/4 , xD ,t E fictitious time space
- time white noise
x space - timeEL3A ,
a) 54yd] = 8ft - s) f'd 'Cx -y)
• scold := f I ↳ fowl ' t £ local ' d '" x1124
* off = DOI - ol'
t 3 ④ equation on IR+ × Id, loco, -1=0)
2
• Fixed - point argument for : to =P * C- f't 3)
* of'"=O → of
"'= P * 3 = :P
- 1 - K
- K-3+2 - k
NB : 3e C-'¥
as. ⇒ P e C a. s .(Sikander )
* ol '" = P * C - P'
t 3)
ill - posed : C'
xcpacf.gl to f.ge Gap is well- def IFF atp > 0.
• Cure : consider abstract FP problem [ Hairy] :
OI = pf - I ' t )
-2k 1-3k y -k
Iterate ! -1- K O n
- Eft ) = - t 9411 - 2 - 29th t CD9H ,X7
3• ITE e S
'
( IR"' ) w/ * ITE i= 3
,I smoothed STWN
* TIE := P * IT't
* IT'II :-. IT4 1TE
Ex.: IT
'=P * IT
'= P * IT' I
'
=P * I E regularised at scale e
• Problem in regularity structures : counter terms to ensure ex.
③
^ aof it :-&:P. .
ft - D) be = - ok + 3. t§g, God k Id* [ Brunel
,Chandra
, Cheryrev , Hairer]with C
,led := ELTIEA
-
I 101]A
Negative twisted antipodes. " Extraction - contraction procedure"
F [ Bruised, Haireritambohti) & (Chandra , Hairer ]
4• Explicitly for the ol ' eq .
( Berglund & Bruised) :
dtif, = DOT, - TO} t 3, t ↳ let t GCE ) Toe
* Cited = ELTIEA-ti lol) for some Ti w I degt;) a 0
Examples:
• A-
= - t 4 I . - 4 I • I •
• Each of the shaded trees gives a Gum At vacuum Feynman diagram !( IT acts multiplicatively on forests )
• Eft' lol)
-9Let's look at this one in some
more detail !
5
Eft' lol) = ELITE ,lol )' ] = Eff L . . . ? 3×06 Idea .
. . - ded)+ use Nick 's thin !
=
2t 4 t 4 t [ 0 's ]
q
Contest
•
•I • 5-1
> -
2 cohesion rule 8'
•a • *"
•
•t
6are. These computations become unwieldy very quickly !
• We need to exploit cancellations for of :-. fig toe to exist !
In the spirit of the original motivation :
2 Can we use AMPLITUHEDRON - like structures to organise
our Feynman diagram computations ?
( RS allow to treat locally subcortical = superrenomalisabk stochastic PDES,
i. e. # It : deg G) a 03 so .
Ever more important when one wants to study critical SPDES ! )
THANK YOU !