BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Pablo Soberón (City University of New York) DTSTART:20240109T210000Z DTEND:20240109T220000Z DTSTAMP:20260423T022917Z UID:GEOTOP-A/75 DESCRIPTION:Title: New results on envy-free distributions\nby Pablo Soberón (City Univ ersity of New York) as part of GEOTOP-A seminar\n\n\nAbstract\nSimilarly t o the growth of Applied Topology\, the uses and applications of Geometry a re now expanding into scientific\, computational\, and engineering domains . First\, we'll review the recent history of this burgeoning Applied Geome try area. I'll mention a couple of collaborations\, developing and impleme nting algorithms inspired by the marked length spectrum that classify comp lex networks (with Eliassi-Rad and Torres) and analyzing digital images u sing a variant of curve-shortening flow (with Velazquez Richards). Then\, I'll present joint work with Evangelista and Ruiz Pantaleón on computatio nal Poisson geometry and its applications to learning symbolic expressions of Hamiltonian systems. We developed and released two Python packages tha t perform symbolic and numerical computation of objects in Poisson geometr y. We then used them to train neural networks (hybrids with CNN and LSTM c omponents) that learn symbolic expressions of Hamiltonian vector fields. F inally\, I'll briefly mention the theoretical limitations of computational ly analyzing Hamiltonian dynamics. I recently constructed an example of a Hamiltonian flow on the 4-sphere that is Turing complete. Therefore the mo st general cases of Hamiltonian learning problems are undecidable.\n LOCATION:https://researchseminars.org/talk/GEOTOP-A/75/ END:VEVENT END:VCALENDAR