Initial Release of Metcore Suite

We are pleased to announce the first official release of Metcore Suite.

Metcore is designed to answer a critical and expensive question in scientific computing: does the memory of your system actually matter, and is heavy non-Markovian computation worth the overhead? Instead of guessing or committing to weeks of heavy simulations like HEOM, pseudomodes, or TEDOPA, or blindly accepting the Markovian short-cut via Lindblad, Metcore allows you to measure the answer mathematically in seconds.

Based on the Markov Embedding Theorem (MET), this suite provides a unified mathematical engine usable across quantum, classical, rheological, and neural problems.

Key Features

This initial release delivers the core engine through three pathways:

1. Cross-Platform Desktop App (metcore-gui)

No Python environment is required. A production-ready desktop interface featuring five specialty tabs:

• Quantum: Spectral density to bath correlation and non-Markovian diagnosis.

• Classical and Rheology: Maxwell-Wiechert and Prony relaxation spectrum analysis.

• Neural: Retarded synaptic and dendritic kernel diagnostics.

• Diagnostics: Geometric non-Markovianity of qubit channels.

• Shortcuts: Closed-form, instantaneous answers for asymptotic Lindblad rates, Kuramoto synchronization thresholds, and reaction-time scaling.

The app also features academic-grade figure exporting in seven formats (including PDF, SVG, EPS, PNG, and TIFF) with custom lab branding and value tables.

2. Model Context Protocol Server (metcore-mcp)

This integrates the suite into any compatible LLM client like Claude Desktop, Cursor, VS Code, or Continue. It includes 17 custom tools allowing an LLM to run full memory diagnostics, choose Prony orders via Maximum Entropy (MaxEnt), and chart non-Markovianity directly from standard prompts.

3. Developer Python Library (metcore)

A robust Python API to integrate the MET pipeline into your own research scripts and automated workflows.

from memkern import zoo, full_diagnosis
import numpy as np

k = zoo.get_kernel("underdamped")
tau = np.linspace(0, 25, 800)
print(full_diagnosis(tau, k.C(tau), t_max=25)["summary"])

Installation and Getting Started

• For the Desktop App: Scroll down to the Assets section below and download the installer or binary for your operating system (.exe for Windows, .dmg for macOS, or .AppImage for Linux).

• For Python Developers and MCP users:

  Bash

   

  pip install metcore metcore-mcp
  metcore-mcp configure