Skip to content



For Simmate's workflow module, we are currently reliant on VASP, which is an expensive DFT software that can be difficult to install for beginners. We are working to move away from propriatary softwares and toward free/open-source codes like ABINIT, Quantum Espresso, or DFTK.jl. That way you can install Simmate and we will take care of the rest. This will take time though... so we recommend that fully experimental labs wait until Simmate hits this milestone. If you'd like to be notified when this occurs, send us an email at

Before you begin

If you are new to Simmate, jump over to our main website and take a look at what we have to offer. This page is for when you're ready to use Simmate in your own research and access some advanced functionality. Our software is open-source and free to use, so come back to try it when you're ready!

What is Simmate?

The Simulated Materials Ecosystem (Simmate) is a toolbox and framework for computational materials research. It lets you explore various crystal databases, predict new materials, and quickly calculate properties (electronic, elastic, thermodynamic, and more).

Computational research can be intimidating because there are so many programs to choose from, and it's hard to mix-and-match them for your specific project. Simmate aims to be the glue between all these different programs, databases, and utilities. We do the heavy lifting and explain these other programs to you along the way.

Even if you consider yourself an experimentalist and have little-to-no coding experience, Simmate's barrier to entry is built to be as low as possible. Our web interface can generate property predictions with a single mouse click. And for learning how to code, we wrote our tutorials and documentation for those that have never used python before.

At the other end of the spectrum, we provide an extremely powerful toolbox and API for experts. Those familiar with the field can view Simmate as an alternative to the Materials Project stack (Atomate, PyMatGen, MatMiner, and more), where we operate under a very different coding philosphy. Here, usability and readability are our top priorities. We therefore distribute Simmate as an "all-in-one" package rather than many separate programs. This includes a core material science toolkit, workflow management, database orm, and a website interface. Simmate also focuses heavily on cloud-based storage, which enables large scale collaborations and avoids researchers repeating calculations. To learn more about the different design choices made in Simmate compared to competing codes, read through our comparisons and benchmarks page.


Don't panic if you're new to coding and Python. When you're ready, head to our tutorials where we teach you everything from the beginning.

If you're comfortable with Python, you can install Simmate with...

conda create -n my_env -c conda-forge python=3.10 simmate  # (1)
  1. While it is the recommended way to install Simmate, conda can be extremely slow. It can take >20 minutes to solve the environment on some systems because of Simmate's many large dependencies.

    We therefore suggest experienced users try the following:

    • set conda config --set channel_priority strict

    • use the libmamba solver

mamba create -n my_env -c conda-forge python=3.10 simmate
pip install simmate


Pip should only be used as an absolute last resort for installing Simmate. Do this only if you are an advanced python user and understand how to manage your python env's effectively.


Simmate itself is <2MB, but when installed to a clean conda environment, the entire download for Simmate and all it's dependencies comes to ~1.2GB. Additional disk space is also needed for optional downloads -- such as third-party data.

Running a Server

Once installed, running a local test server is as simple as...

# On first-time setup, you must intialize an empty database.
simmate database reset

# then start the server!
simmate run-server

After a few seconds, you can open in your browser to view your local server!


Read our website tutorials and documentation in order to switch to a production-ready server that's accessible through the internet and can be shared among a team.

A Sneak-Peak of Features

Again, take a look at our main website if you'd like to see the end-result of what Simmate has to offer. There are many more functions and utilities once you download Simmate, so this section showcases a few of those features.

Prebuilt Workflows

All of the most common material properties have workflows ready to go. These range from simple XRD pattern predictions to intensive dynamic simulations. Simmate also builds off of Prefect for orchestrating and managing workflows. This means that it's up to you whether to run jobs via (i) an advanced user-interface, (ii) the command-line, or (iii) in custom python scripts:

# The command line let's you quickly run a workflow
# from a structure file (CIF or POSCAR).
simmate workflows run relaxation.vasp.matproj --structure NaCl.cif
# Workflows can also be ran from YAML-based configuration
# files, such as the one shown here (named `example.yaml`).
# This would be submitted with the command:
#   `simmate workflows run example.yaml`
workflow_name: relaxation.vasp.matproj
structure: NaCl.cif
command: mpirun -n 8 vasp_std > vasp.out
# Workflows can also be ran from TOML-based configuration
# files, such as the one shown here (named `example.toml`).
# This would be submitted with the command:
#   `simmate workflows run example.toml`
workflow_name = "relaxation.vasp.matproj"
structure = "NaCl.cif"
command = "mpirun -n 8 vasp_std > vasp.out"
# Python let's you run workflows within scripts
# which enables advanced setting configurations.

from simmate.workflows.relaxation import Relaxation__Vasp__Matproj as workflow

state ="NaCl.cif")
result = state.result()

Full-Feature Database

Using all the data on our official site along with your own private data, you can take advantage of Simmate's extremely powerful database that is built off of Django ORM. Simmate also brings together third-party databases and their data -- including those like the COD, Materials Project, JARVIS, and others. With so much data, being able to easily download and navigate it is critical:

# Be sure to follow the database tutorial where we build our 
# initial database with the command `simmate database reset`

from simmate.database import connect
from simmate.database.third_parties import MatprojStructure

# EXAMPLE 1: all structures that have less than 6 sites in their unitcell
structures = MatprojStructure.objects.filter(nsites__lt=6).all()

# EXAMPLE 2: complex filtering
structures = MatprojStructure.objects.filter(
    nsites__gte=3,  # greater or equal to 3 sites
    energy__isnull=False,  # the structure DOES have an energy
    density__range=(1,5),  # density is between 1 and 5
    elements__icontains='"C"',  # the structure includes the element Carbon
    spacegroup__number=167,  # the spacegroup number is 167

# Quickly convert to excel, a pandas dataframe, or toolkit structures.
df = structures.to_dataframe()
structures = structures.to_toolkit()

Utilities & Toolbox

A lot of times in research, a new method is needed to analyze a structure, so a prebuilt workflow won't exist for you yet. Here, you'll need common functions ready to go (such as grabbing the volume of a crystal or running symmetry analysis). Our toolkit functions and classes largely inherit from PyMatGen, which gives you a wide variety of functions to use:

# Load the structure file you'd like to use
from simmate.toolkit import Structure
structure = Structure.from_file("NaCl.cif")

# Access a wide variety of properties. Here are some simple ones.

# Also access methods that run deformations or analysis on your structure.

Scalable to Clusters

At the beginning of a project, you may want to write and run code on a single computer and single core. But as you run into some intense calculations, you may want to use all of your CPU and GPU to run calculations. At the extreme, some projects require thousands of computers across numerous locations, including university clusters (using SLURM or PBS) and cloud computing (using Kubernetes and Docker). Simmate can meet all of these needs thanks to integration with a custom SimmateExecutor (the default), Dask, and/or Prefect:

# On your local computer, schedule your workflow run.
# This is as easy as replacing "run" with "run_cloud".
# This returns a "future-like" object.
state = workflow.run_cloud(...)

# Calling result will wait until the job completes 
# and grab the result! Note, the job won't run 
# until you start a worker that is connected to the
# same database (see command below)
result = state.result()
# In a separate terminal or even on a remote HPC cluster, you
# can start a worker that will start running any scheduled jobs
simmate engine start-worker

Need help?

Post your question here in our discussion section.

Even if it's something like "How do I download all structures with x, y, and z properties?", let us help out and point you in the right direction!

Extra resources