Main Reading Room, Library of Congress (Credit: Carol M. Highsmith, Public Domain)
SUBTOPIC

Economics & Markets

Running a million tomorrows to price the risk in today's decision.

Image: Carol M. Highsmith, Library of Congress

Simulating the Economy Before It Happens

Computational economics runs on Monte Carlo methods — simulating tens of thousands of possible market paths to price derivatives, stress-test portfolios, or model how a policy shock ripples through an agent-based economy. Real-time behavioral analytics adds another layer, processing streaming transaction and sentiment data to detect regime shifts as they happen rather than after the fact. Malgukke provides the CPU-dense, low-latency infrastructure that lets these simulations scale from thousands to billions of trajectories.

HPC Solution Architecture: Market Data to Risk Model

A Monte Carlo and agent-based simulation pipeline, comparable to workflows run on NSF ACCESS systems like Expanse:

Market Data Prices, transactions, sentiment streams Simulation Core CPU Nodes Monte Carlo path generation GPU Nodes Agent-based model, real-time analytics Scenario Set Millions of simulated paths Risk Metrics Value-at-risk, stress scenarios Decision Support & policy design Realized market outcomes recalibrate the scenario generator for the next cycle

Running on the TOP500: Expanse

TOP500-CLASS — NSF / SDSC

Expanse, operated by the San Diego Supercomputer Center (SDSC) at UC San Diego, is a 5.16-peak-petaflop heterogeneous system funded by the National Science Foundation and made available through the ACCESS program. It is explicitly built to serve the "long tail" of science — including economics and the social sciences alongside astrophysics and genomics — with composable CPU/GPU nodes and direct cloud-bursting integration for the kind of large-scale Monte Carlo and agent-based economic simulations described on this page.

Yes — this is a real, currently operating NSF-funded system, directly serving the computational-economics research community described here.

5.16
Peak Petaflops
12+7
PB Lustre + Ceph Storage

Voices from the Field

Michael Norman

Director, San Diego Supercomputer Center; Principal Investigator, Expanse

On why Expanse was built around cloud integration and composable systems to answer questions previously not possible for the long tail of science.

Read article
Shawn Strande

Deputy Director, San Diego Supercomputer Center

On how Expanse's predecessor system supported tens of thousands of users across every research domain, including the social sciences.

Read article