Getting Started

Installation

Basic Installation

pip install rotalabs-ftms

With Embeddings Support

pip install rotalabs-ftms[embeddings]

Development Installation

pip install rotalabs-ftms[dev]

Core Concepts

Memory Fields

Memory in FTMS is represented as a continuous field that evolves according to the heat equation. Key parameters:

• grid_size: Resolution of the memory field (higher = more capacity)
• diffusion_rate (α): How quickly memories spread to nearby regions
• decay_rate (γ): How quickly memories fade without reinforcement
• noise_level: Amount of stochastic fluctuation

Importance Weighting

Not all memories are equal. FTMS uses importance scoring to determine how strongly memories resist decay:

• Entity importance: Named entities, facts
• Causal importance: Cause-effect relationships
• Temporal importance: Time-sensitive information
• Instructional importance: Commands, directives

Multi-Agent Systems

Multiple agents can share memory through:

• Field coupling: Coupled PDEs for memory synchronization
• Collective memory pools: Shared memory with consensus voting
• Topologies: Ring, star, mesh, hierarchical

Basic Usage

Creating an Agent

from rotalabs_ftms import FTCSAgent, AgentConfig, FieldConfig

config = AgentConfig(
    name="my_agent",
    field_config=FieldConfig(
        grid_size=64,
        diffusion_rate=0.1,
        decay_rate=0.01,
    ),
)

agent = FTCSAgent(config=config)

Storing Memories

# Store with automatic importance scoring
agent.store("Important fact to remember")

# Store with explicit importance
agent.store("Critical instruction", importance=0.95)

Querying Memories

# Semantic query
results = agent.query("relevant topic")

# Get most active memories
active = agent.get_active_memories(top_k=10)

Time Evolution

# Evolve the field forward in time
agent.step(dt=0.1)

# Run multiple steps
for _ in range(100):
    agent.step(dt=0.01)