finn fraser grathwol

hi, this is finn!

I am currently studying math & physics at uc berkeley. Prior to berkeley, I was living in vancouver, bc, focused on competitive & outdoor climbing, freestyle & backcountry skiing, wildlife photography, producing electronic music, and exploring the wilderness. It was a curiosity about patterns in nature, repeated across scales and contexts, that led me here.

I am deeply interested in the intersection of geometry, topology, & physics, both to make sense of questions in fundamental physics (quantum gravity, string theory, & nonperturbative formulations), and as a framework to approach more currently accessible, ‘real-world’ problems (topological quantum computation and interpretability of AI models).

I have worked on topological string theory & brane quantization as a 2025 berkeley SURF fellow, a TQFT-realization of perelman’s ricci flow with petr horava in the Berkeley Center for Theoretical Physics (BCTP), and on geometry of feature manifolds in LLMs with marmik chaudhari & rome thorstenson as part of the arcadia research team (ART).

I am passionate about understanding ideas to the deepest level, such that myself and others may be able to appreciate their importance and beauty, regardless of their background. In this spirit, I co-created and taught (w/ keshav deoskar & michelle dong) two berkeley ‘decal’ courses, on lie theory & differential geometry for physics students (fall 2024), and quantum field theory for math students (spring 2025). I plan to teach again in fall 2026 a revised combination of both.

Apart from intellectual pursuits, my previous-to-berkeley interests remain a definitive half of my life. Some life goals include a full coast mountain ski traverse from Vancouver to Alaska, landing a double cork 7 on skis, climbing v13/5.14 trad/sport, establishing multi pitch climbs ground up, and producing an entire album/dj set solely from samples of my environment.

Reach out to me if you are interested in...
intellectual & professional
fundamental questions
quantum gravity: string theory, m-theory, etc.
Do we live in 4 spacetime dimensions? If so, why/ why not?
Is there a fully nonperturbative formulation?
Is the mathematical structure of string theory rich enough to make & test quantum gravity / string theory predictions at energies experimentally accessible in the foreseeable future?
How does the standard model emerge?
What is the nature of geometry near / below the Planck scale?
What do dualities between string theories tell us physically? philosophically?
Is background independence possible?
QFT & mathematical aspects
Mathematical description of the path integral:
is supersymmetry or metric-independence (as in tqfts) needed for finite-dimensional localization?
is there a nonperturbative description in terms of geometry and moduli spaces?
what kind of configurations should be integrated over? smooth? continuous? fractal?
What is the physical meaning of analytic continuation / complexification? For instance, why is the A-model of topological string theory related to quantizing complexified classical phase spaces?
what does the Atiyah-Singer, or APS index theorem tell us about QFT?
Why are QFTs so good at computing topological/ geometric invariants?
Why are different QFTs, in different dimensions, related to each other
ie. WZW (1 + 1 D) - Chern-Simons (2 + 1 D) - N = 4 SYM (3 + 1 D), a la witten?
What is the physical significance of generalized symmetries?
Condensed matter physics & Quantum computation
Is intrinsic non-abelian topological order possible at finite temperature? Can we build and control a topological quantum computer?
The math & physics of the FQHE: noncommutative geometry, CS theory, etc.
For what tasks does a quantum computer (of any architecture) demonstrate a large advantage?
For a specific QC architecture, what problems is it uniquely poised for?
Can we simulate QFTs / QG?
applications of ideas in geometry, topology, & physics
interpretability of AI models
do models pack information onto interpretable "feature manifolds"? what does the geometry & topology of these spaces tell us about how the model works?
what controls the dynamic evolution of these spaces across layers?
Can we use interpretability to make progress in physics, say with understanding LHC data or designing quantum computers/algorithms?
Correspondences between ideas in QFT & AI models
geometry/ topology of discrete, finite data
how do we detect lower dimensional 'manifold(s)' in high dimensional data, unsupervised?
how do we compute cohomology, curvature, dimension, etc. of such spaces? what does this tell us about the data?
robot motion planning
what insights can we gain from the geometry/topology of the robot's configuration space?
suggest more!
type 2/2.5 'fun'
alpine ridge traverses, ideally in one 24-36 hr continuous push
  • Trinity horseshoe
  • howe sound crest trail
  • tantalus traverse
backcountry ski / mountaineering traverses, particularly in the coast mountains of bc
  • pemberton icefield traverse
  • clenndinning traverse
  • tahumming traverse
  • coast mountain traverse (a life goal)
developing new climbing areas - boulder, sport, trad, or multipitch

this is by far my favourite aspect of climbing, and a frequent subject of daydreams. I have developed areas across the sea-to-sky corridor, including a sport climbing area in whistler, as well as a (unreleased) bouldering area in squamish & 5.12 trad multi pitch at the top of the chief (squamish). When I was young, I used to draw in undone climbs into climbing guidebooks; many have since been completed (although mostly not by me, unfortunately).

type 1.5 fun
backcountry/freeride/freestyle skiing: I am looking to build a group of friends that can ski basically any terrain, and will push my limits doing tricks &/or jumping off cliffs
outdoor climbing trips: around california, british columbia, or abroad. Some places I'd love to visit/revisit: Squamish, the boulderfields, skaha, smith rocks, leavenworth, red rocks, bishop, ticino/val bavona, lapland/ flatanger, rocklands
routesetting: This has long been one of my favourite aspects of climbing, often making up climbs for my climbing team, and more recently setting at the hive (vancouver) & now mosaic (berkeley). I recently set for the 2026 mosaic telegraph turnup, and am interested in setting more competitions, etc.
music production: looking for other artists to collaborate with, particularly vocalists, and people who can play physical instruments; also would love to dj, produce music live like fred again, and hold events / parties
photo & video of any of the above, + wildlife, landscape, macro
making vegetarian food, particularly spicy & involving lots of herbs & flavours: I enjoy making up recipes and experimenting
geopolitical documentaries & discussion: I was brought up on documentaries; nyt & the economist are some of the most used apps on my phone. I love discussing geopolitics, history, philosophy, etc. from a wide range of perspectives
type 1 fun
listening to music: I enjoy a wide range of music, with some of my favourite artists being fred again (especially the live/ambient sets), men I trust, isoxo/knock2/isoknock, tyla, nikko, dave, flume, alice phoebe lou, odesza, and the syber djs (baby J, tye turner, baddiejuice), etc. You will often find me reading a paper on hep-th while the latest isoknock show blasts in my headphones at maximum volume.
EDM shows & festivals: in search of friends for EDM/trap (isoknock, fred again, nikko, skrillex, etc.) concerts & festivals. I won't ever touch drugs or alcohol, but I promise I'll turn up just as much or more. Currently trying to obtain niteharts 2026 tickets, iykyk.
media involving social commentary + comedy: south park, family guy, john oliver, etc.