GPU-Accelerated Combinatorial Optimization

DSC-3 Isomorphic Engine

100 Million Spins. 21 Seconds. One GPU.

100,000,000 spins solved on a single RTX 6000 Ada — 3.63 billion ops/sec peak
100M Spins Solved in 21s
3.63 B/s Peak Throughput
16 Solvers Cooperative Ensemble
1331 GFlops HPL LINPACK (91%)
31 Papers 500+ Checks, 0 Falsified
1M in 276ms 10M in 3.0s 25M in 5.7s 50M in 17s 75M in 16s 100M in 21s
Jobs solved: | Avg solve: | Active:
SmartLedger Solutions NVIDIA Inception Origin Neural AI

Interactive Solver

Configure a problem and watch the engine solve it in real time.

Preset:

Showcase

Live demonstrations proving engine capability across 6 real-world domains.

Graph Theory

Petersen Graph MaxCut

Classic 3-regular graph with known optimal cut of 12/15 edges. Verify the engine finds the exact solution.

Expected: Optimal: 12 edges cut
Biology

Quantum Photosynthesis

The FMO light-harvesting complex transfers energy at 95%+ efficiency via quantum coherence. Find the optimal energy pathway.

Expected: 7 chromophore sites
Healthcare

Cancer Mutation Clusters

Partition co-occurring cancer mutations to separate driver mutations from passenger noise using graph partitioning.

Expected: 12-gene network
Mathematics

Ramsey R(4,3) on K9

Color edges of the complete graph on 9 vertices avoiding monochromatic triangles and K4 cliques. A frontier combinatorics problem.

Expected: 36 edges to color
Logistics

Traveling Salesman

Find the shortest tour visiting 10 cities exactly once. The quintessential NP-hard routing problem solved via Ising encoding.

Expected: 10 random cities
Computer Science

Phase Transition 3-SAT

At the critical clause ratio α=4.267, satisfiability undergoes a sharp phase transition. Navigate this computational boundary.

Expected: 20 vars, 85 clauses
Physics

SK Spin Glass

Sherrington-Kirkpatrick model with dense random couplings. The prototypical NP-hard problem from statistical mechanics.

Expected: N=50, dense random
Finance

Portfolio Optimization

Markowitz mean-variance portfolio selection encoded as QUBO with cardinality constraints. Find the minimum-risk asset allocation.

Expected: 20 assets
Healthcare

Drug Interaction Network

Partition a drug-drug interaction graph to identify potentially dangerous combination clusters.

Expected: 15 drugs, 25 interactions
Logistics

Supply Chain Routing

Optimize warehouse-to-store delivery routes across a clustered distribution network.

Expected: 12 delivery points
Cybersecurity

Network Vulnerability Clustering

Partition a network graph to isolate vulnerable nodes from critical assets.

20 nodes, 30 edges
Operations

Warehouse Layout

Optimize item placement to minimize picker travel distance using co-order frequency.

25 items, 5 categories
Biotechnology

Protein Contact Map

Optimize a simplified protein folding energy landscape on a lattice model.

18 amino acids (HP model)
Operations

Exam Scheduling

Schedule 16 exams across time slots to minimize student conflicts.

16 exams, 24 conflicts

Benchmarks

Performance data from reproducible benchmark suites.

Run a live GPU benchmark at 500K, 1M, 2M, 5M, and 10M spins on the RTX 6000 Ada (48GB VRAM).

Problem Size DSC-3 D-Wave Gurobi SimCIM Winner
MaxCut G14 800 3064 3048 3064 3055 Tie
MaxCut G22 2000 13359 13180 13346 13310 DSC-3
SK Spin Glass 5000 -3.812 -3.790 -3.805 DSC-3
3-SAT (hard) 200v SAT SAT SAT UNSAT Tie
TSP Berlin-52 52 7544 7890 7542 7620 Gurobi
Portfolio QUBO 50 -142.7 -139.2 -142.7 -141.0 Tie

API Access

RESTful JSON API for programmatic access. All endpoints are proxied with server-side authentication.

POST /v1/solve

Submit a problem for solving. Returns a job ID for async polling.

Example Request / Response
// Request
{
  "problem_type": "maxcut",
  "n": 10,
  "edges": [[0,1,1.0], [1,2,-0.5], [2,3,1.0]],
  "preset": "production"
}

// Response
{
  "job_id": "j_8f3a2b1c",
  "status": "queued"
}
GET /v1/jobs/{id}

Poll the status and result of a solve job.

Example Response
{
  "job_id": "j_8f3a2b1c",
  "status": "completed",
  "result": {
    "best_energy": -3064.0,
    "best_solver": "SA",
    "best_spins": [1, -1, 1, "..."],
    "elapsed_secs": 0.142,
    "i_value": 0.847,
    "t_class": "GueRandom",
    "solver_order": ["SA", "SBM", "..."],
    "all_results": [
      {"solver_name": "SA", "energy": -3064.0, "elapsed_secs": 0.142, "stagnated": false},
      {"solver_name": "SBM", "energy": -3060.0, "elapsed_secs": 0.098, "stagnated": false}
    ]
  }
}
GET /v1/info

Returns engine capabilities, supported problem types, and limits.

Example Response
{
  "name": "DSC-3 Isomorphic Engine",
  "version": "3.2.0",
  "solvers": 16,
  "gpu_available": true,
  "features": ["auto_sparse", "z2_symmetry", "spectral_analysis"],
  "problem_types": ["maxcut", "ising", "sat", "tsp", "ramsey", "qubo"],
  "presets": ["fast", "production", "quality"],
  "max_n": 50000000
}
GET /v1/health

Returns engine health status and uptime.

Example Response
{
  "status": "healthy",
  "uptime_secs": 84321
}
POST /v1/generate

Generate a random problem instance for testing or benchmarking.

Example Request / Response
// Request
{
  "n": 500,
  "density": 0.1,
  "seed": 42
}

// Response
{
  "problem_type": "ising",
  "n": 500,
  "matrix": [[0.0, 0.73, -1.2, "..."], "..."],
  "field": [0.1, -0.3, "..."]
}