MolKet offers management consulting and AI services for:
- quantum molecular dynamics,
- cryptography,
- neuromorphic-based computing
The company also offers consulting services for training AI on datasets from fields such as chemistry, biology, materials science, and cosmology. These services are supported by physics-based kernels (e.g., quantum-mechanical kernels), enabling the development of advanced machine learning algorithms with significant predictive power.
MolKet harnesses the power of quantum intelligence to revolutionize simulations in chemistry and physics. Our AI agents, trained on intricate Hamiltonian derivations, predict the behavior of complex systems, significantly accelerating research and discovery. Dr. Taha Selim proposes that algorithm design tailored to the physics of a problem cultivates what he terms 'quantum intelligence,' driving complex simulations. He explores how models can learn and derive the Hamiltonian from data, and crucially, predict Hamiltonians for larger or specified systems by leveraging knowledge gained from smaller, known systems. Dr. Selim also addresses the limitations of the universal function approximator concept and highlights the promise of physics-inspired kernels in advancing AI applications.
Alain Chancé and Dr. Taha Selim propose leveraging AI agents to accelerate chemical design and quantum dynamical simulations. These AI agents can be trained on the inputs and outputs of quantum chemistry and dynamics algorithms, effectively replacing computationally expensive components. Moreover, they can learn the scaling behavior of these algorithms, enabling extrapolation to larger systems. Their recent proof-of-concept demonstrates the potential of this approach. They successfully developed an AI agent to generate vibrational wavefunctions of small molecules such as CO₂, using only frequency and potential parameters as input, thus bypassing the need for computationally intensive variational algorithms. This work highlights the promising synergy between quantum computing and AI in advancing the field of quantum dynamical simulation.
Our Complementary Services
Alongside the cloud services, we can help you in the followingDesigning algorithms for quantum molecular dynamics
Adapt your application using HPC/QC
Training, consultancy, and knowledge transfer
We provide fast and accurate molecular design simulations assisted by AI, performed and automated by hybrid quantum computing + HPC
Chemistry simulations are time consuming and complicated to prepare. Our goal is to make the design process easier and fast. We use AI to train general Hamiltonians for chemistry simulations that can be executed efficiently on HPC, quantum, computers, and on simulations of quantum computers.
We provide highly engineered, standard process for quantum chemistry simulations over MolKet Studio
How Does It Work?
-
Choose your AI-assisted Environment
You can choose a jupyterhub like environment Or MolKet studio
You can use MolKet’s engine on your computer in visual studio or program it online on the web.
import molket_engine as me import molket_visual as mv import molket_data as md # define the molecular structure # insert H2 molecule molecule = me.Molecule('H_2') # define the geometry of the molecule: example H2 molecule.geometry = [('H', (0, 0, 0)), ('H', (0, 0, 0.74))] # you can also define a grid of points to sample the wavefunction r = [0.5,0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2,1.3,1.4,1.5,1.6, 1.7,1.8,1.9,2] # in Angstrom molecule.geometry = [('H', (0, 0, -r/2)), ('H', (0, 0, r/2))] -
Type your code or drag and design your simulation
Type your code in natural language, or choose from our library. Our AI-assistant will help you filling the code writing and suggest modules that you can use.
Alternatively, you can design the simulation in a flow chart format and our AI-assistant will convert it into code. Then, it will choose the proper modules and print the code sequence for to confirm.
# create the Hamiltonian operator with the molecule and data structure H_op = me.Operator('H_2') H_op.molecule = molecule # define the electronic basis set H_op.elec_basis = 'sto-3g' # define the nuclear motion basis set, the vibrational basis set H_op.nuc_basis = ('harmonic','Gaussian')
-
MolKet's AI engine will assist in selecting the algorithms and computing platforms.
Depending on the computational complexity and the type of each step in the simulation, MolKet’s AI engine will choose the proper quantum/HPC architecture for you.
It will also optimize the execution on each type of hardware by training the general Hamiltonians and map them onto quantum
# define the rotational basis set (wavefunctions), ... # Default: spherical harmonics for linear molecules, and Wigner D-matrices for non-linear molecules. H_op.rot_basis = ('Ylm') # you can also define phase or choose real harmonics depending on the symmetry of the molecule # Define the chips for the computing part: the choice of the hardware and the backend H_op.Qchip = 'Qqx2' ## variations of Qqx4: Qqx2, Qqx3, Qqx4, Qqx5, Qqx20, Qqx_qasm_simulator ## accelerated computing with GPU H_op.HPCchip = 'GPU'
-
Get the results and post-processing using builtin ML and AI libraries.
Depending on the computational complexity and the type of each step in the simulation, MolKet’s AI engine will choose the proper quantum/HPC architecture for you.
# choose the simulators for the quantum and HPC backends, to be used in analysis as well H_op.Qbackend = 'statevector_simulator' H_op.HPCbackend = 'local_qasm_simulator' vib_groundstate_energy= H_op.vib_eigE('0','groundstate') vib_WF0 = H_op.vib_eigW('0','groundstate') #H = H_vib + H_rot + H_elec + H_nuc
Our Development Model
We can help you by providing what you need from built-in libraries that help you in your execution.
Molket Founders
Dr. Taha Selim
Directeur Général - General Manager
Lecturer/Researcher (QC Education Officer), Amsterdam University of Applied Sciences
MVP Founder Member QSECDEF
PhD Quantum Theoretical & Computational Chemistry
M.Sc. of Quantum physics, Lasers, & Materials.
LinkedIn: https://www.linkedin.com/in/tiselim/
Talks
Co-speaker lightning talk Simulating quantum molecular dynamics with Julia at JuliaCon 2024
Speaker, Carbon dioxide (CO2) as a quantum molecular sensor in protoplanetary disks, at APS March Meeting 2024
Podcasts and videos
Quantum AI Podcast, Quantum AI for Chemistry Simulations, June 2024
Publications
Carbon dioxide (CO2) as a quantum molecular sensor in protoplanetary disks, at APS March Meeting 2024
State-to-state rovibrational transition rates for CO2 in the bend mode in collisions with He atoms, Taha Selim, Ad van der Avoird, and Gerrit C. Groenenboom, J. Chem. Phys., 159, 164310 (2023)
Efficient computational methods for rovibrational transition rates in molecular collisions, Taha Selim, Ad van der Avoird, and Gerrit C. Groenenboom, J. Chem. Phys., 157, 064105 (2022)
Multi-channel distorted-wave Born approximation for rovibrational transition rates in molecular collisions , Taha Selim, Arthur Christianen, Ad van der Avoird, and Gerrit C. Groenenboom J. Chem. Phys., 155, 034105 (2021)
Certificates and Badges
IBM Quantum Challenge: Spring 2023 Achievement, Badge
