Physics-constrained materials analysis
Meridian hosts our 3D model canvas for virtual experiment design and efficient modules for materials structure analysis. Pedion treats magnetic, electronic, caloric, electrochemical and other physical properties. Mimesis (in design) integrates density-functional theory (DFT) and molecular dynamics (MD), our intent with Mimesis is that any user regardless of their expertise in theoretical physics can simulate, predict, and reverse-engineer materials.
MatSpace
Materials insight with AI
The benchmark problem
Ordinary tools will improve a plot if you ask them to. MatSpace is built to refuse numbers that cannot satisfy physics.
Unphysical fits
An XPS C 1s envelope can look like one peak with a slight deflection. A typical fitter deconvolutes two Gaussians and stops because the overlay matches. MatSpace deconvolutes the hidden third the envelope still requires.
Typical fitter · 2 components
Residual
Time to first result
A materials researcher still spends a whole afternoon in a spreadsheet, a notebook, and an Origin-class suite for structural data reduction and property analysis. MatSpace saves researchers time and eases the material data analysis.
Spreadsheet
55 min
Python notebook
40 min
Desktop suite
18 min
MatSpace
4 min
Typical time to a first named-model fit or property reduction (minutes)
The measured case on this graph is a model nonlinear fit in different suites vs the MatSpace CurveFit Lab.
AI chat bots
Generic AI chat bots and LLMs invent conclusions on your data from online scattered information that sounds satisfactory. Two studies on a single material can present contradicting information. But does that answer actually satisfy physics? Every MatSpace module is bound by physics and mathematical laws, so the generated tokens are physics constrained information about this dataset.
Two papers on this same C 1s envelope disagree — graphene oxide versus carbide. Which assignment should I write? - AI Chat bots
Literature consensus is graphene oxide with 32% C=O, confirming monolayer GO. Average the two papers and report that composition.
Scattered papers, averaged into a conclusion this spectrum did not contain.
Platform
Experimental design and materials structure
Experiment design and structural analysis with physically allowed parameters.
Meridian is the core MatSpace platform for planning synthesis workflows with physics bound 3D models and processing raw structural datasets. From raw powder XRD patterns and spectroscopic curves to microstructural images, Meridian treats every dataset as a physically constrained model rather than an unconstrained curve-fitting problem.
Platform
Materials property analysis
Functional property analysis from the governing relations, not from an unconstrained curve.
Pedion evaluates functional material performance through rigorous thermodynamic and constitutive models. Calculate magnetocaloric entropy changes (ΔSM), model thermoelectric transport parameters, extract piezoelectric coefficients, and evaluate electrochemical voltammetry datasets—all bounded by physical law.
Platform
Theory and simulation · in design
Electronic structure and atomistic dynamics under the same refusal to invent.
Mimesis is the MatSpace platform for theory, density-functional and molecular-dynamics engines that report what the Hamiltonian produced, and does not write a phase, a barrier, or a spectrum the calculation did not contain.
Access
Sign up via email, Google, or X to access the workbench environment.
Select a credit pack suited for individual runs or research group campaigns. Unused credits never expire.
Live laboratories on Meridian bill one credit per successful operation. Pedion and Mimesis join the same wallet as they open.
Session files are processed for that run; they are not stored as a training set. Unused credits remain in your wallet with no expiry.
Meridian is live for protocol design, powder diffraction, and equation-first fitting. Pedion is live for constitutive property analysis. Mimesis is in design.
