When Semen samples are imaged under a microscope, they reveal how motile the sperm cells are in that sample (VISEM dataset). This 'motility' is directly related to the sperm's ability to travel the distance inside the female reproductive system to reach the egg and possibly fertilize it. Then a natural question arises: Can this motility be inferred from the microscale videos of semen samples? However, in a previous work from our group, we discovered that interaction between agents in a collective obfuscates the inference problem. This leads us to ask: Can the interactions between the sperm cells be inferred? We are also interested in finding out if these interactions lead to collective migration of these cells in large densities and numbers in models for the female reproductive system.

It is often a non-intuitive task to predict or understand the motion of droplets through a network of microchannels. These devices have been found to show aperiodic patterns, large scale oscillations, synchronisation, etc. The complexity observed arises because a droplet changes the resistance for fluid motion in the branch it enters, which alters the flow in  entire network. This in turn affects the motion of these droplets and their decisions at a bifurcating channel giving rise to droplet motion through these channels that is nontrivial. In this work, we aim to develop computer-aided solutions to the design and operation of microfluidic networks where droplets have to be parked at designated areas.

Order parameters characterising the polarisation of the collective, in finite sized flocks, exhibit large scale fluctuations. This motivates one to ask: What underlying stochastic dynamical processes lead to the observed features? We have developed an easy to use, Python based package (PyDaddy) which extracts an 'interpretable' Stochastic Differential Equation Model for a given time series data. We use PyDaddy on time series generated from different flocking models to understand the nature of the mesoscale dynamics and its connections to the underlying microscale interactions. We plan to extend the method to identify SPDEs which capture not only the temporal variation but also the spatial variation of the order parameter in larger flocks.

At high densities and flows, pedestrian systems may turn 'critical'; i.e. a small perturbation can spontaneously trigger a jammed state or lead to stampede. Hence, from a planning/crowd-control viewpoint, it is important to ask how likely is a stampede or any undesirable event in a crowd. In other words, we ask what is the 'crowd risk' associated with a pedestrian system. Existing methods to quantify crowd risk, in India, involves traffic experts looking at video clippings of pedestrian movement to assign a value for crowd risk. In this context, it is important to ask: Can we quantify crowd risk from the movement information? What features are the experts looking at, when they quantify risk? 

Mean field models are often employed to study the collective behavior exhibited by organismal systems which assumes 'fully connectedness' -- an agent can perceive all agents in the system. However, we know that organisms have finite field of vision and only interact with nearby neighbors. If that is the case, when is any mean-field analysis valid? We observe that the agents exhibit, what we call, configurational space mixing, due to the probabilistic nature of the interactions. We study how this mechanism may allow the agents to access information available in the entire system from just its neighbors---resulting in 'meanfield-ness'.