Postdoctoral Research Fellowship: Network Thermodynamics of Distributed Computation

SANTA FE INSTITUTE LIBRARY; Santa Fe (NM); On-site; USD 65,000 - 90,000; Full time

Benefits offered by this job

J1 visa sponsorship

Job summary

The Santa Fe Institute in Santa Fe, NM, invites applications for a two-year, full-time postdoctoral fellowship in physics or computer science to explore the thermodynamic cost of distributed computation across digital circuits, neural networks, and human brains.

The successful candidate will work with PI David Wolpert on projects using mismatch cost (MMC) to relate network topology to cost, speed, robustness, and computational power, scaling MMC to large state spaces.

Qualifications

• Ph.D. in Physics or Computer Science by start date.
• Background in stochastic thermodynamics or theoretical computer science.
• Familiarity with Monte Carlo methods and uncertainty quantification.
• Experience with advanced mathematical modeling and computational analysis.

Responsibilities

• Collaborate with PI and team to advance the project through research and publications.
• Develop and apply mismatch-cost theory and large-state-space techniques.
• Design, train, and analyze distributed computational systems to relate topology to cost, robustness, and speed.
• Investigate how hardness of problems relates to thermodynamic cost.

Skills

distributed computation

programming

mathematical modeling

team collaboration

Education

Ph.D. in Physics

Ph.D. in Computer Science

Tools

Monte Carlo methods

Tensor networks

Uncertainty quantification

Backward differential equations

Job description

The Santa Fe Institute - a private, not-for-profit research and education organization - has an opening for a two-year full-time postdoctoral fellowship. We are seeking a highly motivated scholar with expertise in physics (or in special cases in computer science), who has a desire to apply their expertise to understand the thermodynamic cost of distributed computation, from digital circuits and neural networks to human brains.

The candidate will work with PI David Wolpert on a project investigating how the network coupling the components of the distributed computer controls the tradeoff among the thermodynamic cost of running the computer, the computer's speed, its robustness against component error, and the precise computation it performs. A particular focus will be to see how the hierarchical and / or modular structure of the network controls the tradeoff among these aspects of distributed computers.

The primary tool in this investigation will be "mismatch cost" (MMC), a recently derived strengthening of the second law that applies to all physical systems - in particular those implementing computation - independent of the physical details of those systems. The central challenge, and the focus of this project, is scaling MMC calculations up to systems with very large state spaces and complicated dynamics, using techniques such as coarse-graining (in both time and space), uncertainty quantification, backward differential equations, tensor networks, and Monte Carlo approximation. The project will apply these tools to trained feedforward neural networks, restricted Boltzmann machines, and lightweight LLMs, relating their network topology-modularity, hierarchy, communication structure to their thermodynamic cost and the difficulty of the computations they perform (including canonical problems relating computer science and thermodynamics, such as KSAT).

This position is based in-person in Santa Fe, NM. The desired start date is no later than June 1, 2027. The term of this position may be extended if appropriate and if funding allows.

Responsibilities:

Collaborate with PI and other team members to advance this project through research, publication, workshop organization, etc.

Develop and apply mismatch-cost theory and large-state-space approximation techniques (e.g., coarse-graining, uncertainty quantification, backward differential equations tensor networks, Monte Carlo) to distributed computational systems

Design, train, and analyze distributed computational systems - such as feedforward neural networks, restricted Boltzmann machines, and lightweight LLMs - to relate their network topology to thermodynamic cost, robustness, and speed, and computational power.

Investigate how the "hardness" of computational problems (e.g., KSAT) relates to the thermodynamic cost of solving them.

Preferred Qualifications:

Ph.D. in Physics, Computer Science, or a related field (by start date).

Background in stochastic thermodynamics, statistical physics, or theoretical computer science (computational complexity).

Familiarity with Monte Carlo methods, uncertainty quantification, backward differential equations, and/or tensor network approximations.

Strong programming skills and interest in interdisciplinary collaboration spanning physics and computer science theory.

SFI is an Equal Opportunity Employer and is committed to fostering a diverse and inclusive academic global community. Women and members of underrepresented groups are especially encouraged to apply. U.S. citizenship is not a requirement, however, you must be legally able to work in the US. SFI will sponsor a J1 Visa for successful candidates. SFI is not able to sponsor a H1B Visa for candidates.