Skip to content

Commit

Permalink
Add Daniele's talk
Browse files Browse the repository at this point in the history
  • Loading branch information
IanBriggs authored Sep 26, 2024
1 parent 031f7a9 commit 33cf641
Showing 1 changed file with 48 additions and 1 deletion.
49 changes: 48 additions & 1 deletion index.html
Original file line number Diff line number Diff line change
@@ -52,7 +52,9 @@ <h1>FPBench</h1>
<details>
<summary>
<time>TIME</time>
TITLE, use sentence not title case
<div class="title">
TITLE, use sentence not title case
</div>
<div class="speaker">
TODO, University of TODO
</div>
@@ -102,6 +104,51 @@ <h1>FPBench</h1>
</details>
-->

<details>
<summary>
<time>Oct 3, 2024</time>
<div class="title">
Geometric predicates for unconditionally robust elastodynamics simulation
</div>
<div class="speaker">
Daniele Panozzo, Courant Institute of Mathematical Sciences in New York University
</div>
</summary>
<div class="abstract">
The numerical solution of partial differential equations (PDE) is
ubiquitously used for physical simulation in scientific computing and
engineering. Ideally, a PDE solver should be opaque: the user provides
as input the domain boundary, boundary conditions, and the governing
equations, and the code returns an evaluator that can compute the
value of the solution at any point of the input domain. This is
surprisingly far from being the case for all existing open-source or
commercial software, despite the research efforts in this direction
and the large academic and industrial interest. To a large extent,
this is due to lack of robustness in geometric algorithms used to
create the discretization, detect collisions, and evaluate element
validity.
I will present the incremental potential contact simulation paradigm,
which provides strong robustness guarantees in simulation codes,
ensuring, for the first time, validity of the trajectories accounting
for floating point rounding errors over an entire elastodynamic
simulation with contact. A core part of this approach is the use of a
conservative line-search to check for collisions between geometric
primitives and for ensuring validity of the deforming elements over
linear trajectories.
I will discuss both problems in depth, showing that SOTA approaches
favor numerical efficiency but are unfortunately not robust to
floating point rounding, leading to major failures in simulation. I
will then present an alternative approach based on judiciously using
rational and interval types to ensure provable correctness, while
keeping a running time comparable with non-conservative methods.
To conclude, I will discuss a set of applications enabled by this
approach in microscopy and biomechanics, including traction force
estimation on a live zebrafish and efficient modeling and simulation
of fibrous materials.
</div>
</details>

<details>
<summary>
<time>Sep 5, 2024</time>

0 comments on commit 33cf641

Please sign in to comment.