Quantum Machine Learning for Predicting Molecular Spectral Properties

Background

Molecular Dynamics (MD) modeling is a powerful computational method used to simulate the physical movements of atoms and molecules over time, allowing researchers to observe molecular systems in action (at the nanoscale). This technique, combined with in silico screening, plays a crucial role in predicting the behavior and interactions of new chemistries.

While MD modeling enables virtual testing of compound efficacy, safety, and interactions, a significant challenge remains: ensuring that these models accurately reflect experimental observations. It is essential to develop MD models capable of verifying experimental results and calculating relevant parameters comparable to those observed in lab experiments. In the context of cosmetic and laundry detergent applications, these models could support product development by predicting viscosity changes upon incorporation of various small molecules, simulating film formation on hair or textile fibers, and anticipating phase behavior in surfactant-polymer mixtures.

Developing hybrid models that can combine calculated parameters from MD with trained artificial intelligence (AI) models could lead to improved predictive accuracy and provide the opportunity for high throughput screening prior to laboratory testing.

What we're looking for

We are looking for modeling solutions with strong predictive accuracy to identify new ingredients for cosmetic applications and/or laundry detergents. We are particularly interested in models that can predict interactions of compounds with hair, skin, or textile substrates and estimate core functionalities, such as film formation on hair or color retention on textiles. We are also open to models capable of predicting interactions within complex formulations, such as viscosity changes and the complexation of polymers and surfactants.

Solutions of interest include:

• Molecular dynamics techniques
• Hybrid modeling of MD and AI
   

Our must-have requirements are:

• Proof of concept for any of the above-mentioned applications

Our nice-to-have's are:

• Uses open-source software like LAMMPS or GROMACS