Optimization API

Solve optimization problems using Sematryx with configurable AI intelligence through the AEAO Tetrad.

Basic Optimization

The simplest way to optimize a function. Define your objective function and bounds, then let Sematryx find the optimal solution.

Python SDK - Basic Optimization
from aeao import aeao
import numpy as np

# Define your objective function
def sphere(x):
    return sum(xi**2 for xi in x)

# Run optimization
result = aeao(
    objective_function=sphere,
    bounds=[[-5, 5], [-5, 5]],
    max_evaluations=1000
)

print(f"Best solution: {result['best_solution']}")
print(f"Best fitness: {result['best_fitness']}")

Parameters

  • objective_function (required): Function to optimize f(x) → float
  • bounds (required): Search bounds [[min1, max1], [min2, max2], ...]
  • max_evaluations (optional): Maximum function evaluations (default: 1000)
  • mode (optional): "fast", "balanced", or "accurate" (default: "balanced")

AEAO Tetrad Configuration

Configure the four pillars of Sematryx's AEAO Engine intelligence: Agentic, Expository, Autodidactic, and Domain Extension.

Tetrad Configuration Examples
from sematryx import sematryx, AEAOTetradCompleteConfig

# Use preset configuration
result = sematryx(
    objective_function=sphere,
    bounds=[[-5, 5], [-5, 5]],
    preset="production"  # development, production, research, enterprise, minimal
)

# Or enable specific tetrad pillars
result = aeao(
    objective_function=sphere,
    bounds=[[-5, 5], [-5, 5]],
    use_agentic_intelligence=True,
    use_autodidactic_intelligence=True,
    explanation_level=3
)

# Or complete custom configuration
config = AEAOTetradCompleteConfig.enterprise()
config.expository.explanation_level = 4
result = aeao(objective_function=sphere, bounds=[[-5, 5], [-5, 5]], config=config)

Tetrad Pillars

  • 🤖 Agentic Intelligence: Multi-agent coordination for strategy selection
  • 📖 Expository Intelligence: Explainability with configurable levels (0-5)
  • 🧠 Autodidactic Intelligence: Self-improvement and learning from experience
  • 🏗️ Domain Extension: Business domain libraries for rapid adoption

Preset Configurations

  • development: Fast iteration, basic explanations
  • production: Balanced performance, standard explanations
  • research: Maximum capabilities, comprehensive explanations
  • enterprise: Full features, advanced monitoring
  • minimal: Core optimization only

Domain-Specific Optimization

Use specialized optimization libraries for specific business domains with pre-configured constraints and objectives.

Domain-Specific Optimization
from aeao import financial_optimize, healthcare_optimize, supply_chain_optimize

# Financial portfolio optimization
result = financial_optimize(
    problem_type="portfolio",
    config={
        "assets": ["AAPL", "GOOGL", "MSFT"],
        "risk_tolerance": 0.3
    },
    max_evaluations=2000
)

# Healthcare drug discovery
result = healthcare_optimize(
    problem_type="drug_discovery",
    config={
        "target_protein": "protein_id_123",
        "constraints": {"toxicity": "< 0.1"}
    }
)

# Supply chain routing
result = supply_chain_optimize(
    problem_type="vehicle_routing",
    config={
        "locations": [...],
        "vehicle_capacity": 1000
    }
)

Available Domains

  • Financial: Portfolio optimization, trading strategies, risk management
  • Healthcare: Drug discovery, clinical trial design, treatment protocols
  • Supply Chain: Vehicle routing, inventory management, warehouse optimization
  • AI/ML: Hyperparameter tuning, neural architecture search
  • Marketing: Campaign optimization, budget allocation

REST API

Use the REST API for server-side optimization requests. Upload your objective function first, then call the optimize endpoint.

POST /v1/optimize
curl -X POST https://api.sematryx.com/v1/optimize \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "objective_function_id": "func_1234567890",
    "bounds": [[-10, 10], [-10, 10], [-10, 10]],
    "max_evaluations": 2000,
    "preset": "production",
    "tetrad_config": {
      "use_agentic_intelligence": true,
      "use_expository_intelligence": true,
      "explanation_level": 3
    }
  }'
Response
{
  "success": true,
  "best_solution": [0.001, -0.002, 0.003],
  "best_fitness": 0.000014,
  "evaluations": 847,
  "duration_seconds": 2.34,
  "strategy_used": "shgo",
  "tetrad_config": {
    "use_agentic_intelligence": true,
    "use_expository_intelligence": true,
    "use_autodidactic_intelligence": false,
    "use_domain_extension": true
  },
  "features_active": {
    "agentic_intelligence": true,
    "expository_intelligence": true,
    "explanation_level": 2,
    "cross_problem_learning": false
  }
}

Response Format

Optimization Result

  • success: Whether optimization succeeded
  • best_solution: Optimal parameter values found
  • best_fitness: Best objective value achieved
  • evaluations: Number of function evaluations used
  • duration_seconds: Time taken for optimization
  • strategy_used: Optimization algorithm selected
  • tetrad_config: Tetrad configuration that was active
  • features_active: Which tetrad features were enabled

Advanced Features

Explainability

Get detailed explanations of optimization decisions with configurable explanation levels (0-5).

result = aeao(objective, bounds, explanation_level=4)

Self-Improvement

Enable learning from optimization experience to improve performance on repeated problems.

result = aeao(objective, bounds, use_autodidactic_intelligence=True)

Multi-Agent Coordination

Use multiple AI agents to collaboratively select the best optimization strategy.

result = aeao(objective, bounds, use_agentic_intelligence=True)

GPU Acceleration

Accelerate optimization with GPU/CUDA support for large-scale problems.

result = aeao(objective, bounds, use_gpu_acceleration=True)