Optimization
Compare objectives, stopping rules and the limits of a solution.
Questions to work through.
Explore the sources.
All 52 resources →[Re] Bandit Theory and Thompson Sampling-guided Directed Evolution for Sequence Optimization
ReScience C
A replication about Thompson sampling and bandit-guided sequence optimization.
Investigate How does Thompson sampling compare with uniform allocation on evaluator-generated finite reward landscapes at equal budgets?
[Re] Faster Teaching via POMDP Planning
ReScience C
A replication about automated teaching formulated as planning in a partially observable decision process.
Investigate Can a bounded planner match an exact small-state reference on a fixed teaching objective?
[Re] Least-cost modelling on irregular landscape graphs
ReScience
An R replication of least-cost modeling on irregular landscape graphs using graph and triangulation concepts.
Investigate Does the computed least-cost path agree with an independently enumerated optimum on bounded irregular graphs?
[Re] Predicting metabolic biomarkers of human inborn errors of metabolism
ReScience
A replication about predicting metabolic biomarkers with flux-variability analysis and metabolic reconstruction.
Investigate Can a bounded flux-variability computation recover analytically known feasible ranges in a small network?
[Re] Speedup Graph Processing by Graph Ordering
ReScience C
A C++ replication about graph ordering as a way to change graph-processing performance.
Investigate Does a chosen ordering preserve algorithm outputs while improving measured performance across graph families?
[Re] Thompson Sampling for Bandits with Clustered Arms
ReScience C
An R replication of Thompson sampling for bandits whose arms have a cluster structure.
Investigate When the assumed clusters are deliberately incorrect, how does cumulative regret change at a fixed sampling budget?
Accelerate
Upstream project repository
Training and inference orchestration for studying device placement, mixed precision and distributed execution without changing the scientific objective.
Investigate Does one controlled optimization step agree between a single-device reference and a fixed distributed configuration?
Advanced Features
CVXPY documentation
A guide to dual variables, transformations, problem arithmetic and canonical forms for convex optimization.
Investigate Do reported primal and dual variables satisfy independently computed feasibility and complementary-slackness checks on small convex problems?
Auction Verification
UCI Machine Learning Repository
Auction-verification workload descriptors for examining computational cost prediction.
Investigate How well can a bounded runtime model predict withheld auction configuration families?
BayesianOptimization
Upstream project repository
Bayesian optimization for bounded black-box objective functions.
Investigate Does a fixed acquisition strategy reduce regret faster than a prespecified random-search baseline on a bounded objective?
bitsandbytes
Upstream project repository
Low-precision tensor and optimizer components for investigating memory and numerical-error tradeoffs.
Investigate How does one supported quantization configuration change a small fixed matrix product relative to a high-precision reference?
Chex
Upstream project repository
Assertions and testing utilities for JAX programs, useful for expressing shape, type and transformation expectations.
Investigate Do declared shape and tree invariants detect deliberately malformed inputs to a tiny numerical function?