If it's an optimization problem, it can run on the market.
Anywhere a hard problem can be written as an objective or a constraint set, Samādhān can encode it — and return both the solution and a return.
Allocation, calibration, and risk — as computation
Mean-variance allocation, model calibration, cointegration tests.
Encode the optimization / linear system across mean-reverting pairs.
The allocation is the equilibrium; the rebalancing flow is the return.
Molecular & materials optimization
Minimum-energy configurations (Lennard-Jones clusters), structure search.
Encode the energy landscape as a continuous-optimization force field.
Converged low-energy configuration — at a fraction of the usual compute cost.
Metabolic flux optimization
Optimize pathway fluxes (e.g. E. coli glycolysis) under constraints.
Constraints become spreads; objective becomes the relaxation force.
Optimal flux distribution recovered on the substrate.
Steering & structured search
LLM activation steering, hyperparameter and combinatorial search.
Encode the objective into the MCM gas engine.
A steered model / optimal configuration, plus PnL from the run.
Routing and scheduling
Shortest path, assignment, and satisfiability (SAT) problems.
Path-cost / clause constraints push flow onto the optimal solution.
The route or schedule reads out directly from the fills.
Bring your problem.
We'll scope whether it fits the substrate and what a run would look like.
Request access