Andrea Zanetti, Senior Data Scientist, Intel Simulation of particle transport through matter is fundamental for understanding the physics of High Energy Physics (HEP) experiments, as the ones at the Large Hadron Collider (LHC) at CERN. Such experiments are very computing-intensive for geometry modelling and navigation through millions of objects and physics models. One of the recently developed approach is to introduce deep learning algorithms to train on different particle quantities and model the traditional simulation process. Using deep learning algorithms has shown to greatly improve the efficiency and reduce the computing cost. Several techniques are under testing at CERN with this purpose, such as generative adversarial networks (GANs), to replace the Monte Carlo approach. Indeed, at CERN, the GeantV project introduces fine grained parallelism, vectorisation, efficient memory management and NUMA in physics simulations. Within the GeantV framework we are developing a deep learning based tool for fast simulation to replace standard Monte Carlo simulation. This represents a completely generic approach in the sense that such a network could be designed and trained to simulate any kind of detector. Such development addresses the ever increasing need for simulated events expected for LHC experiments upgrades. In this presentation, we will show an example of how DL algorithms can contribute to scientific simulations in this area, and we will concisely describe how to implement a conditional Generative Adversarial Model with Auxiliary Classifiers to reproduce images of the energy deposited by particles traversing an electromagnetic calorimeter detector, using Intel Optimized Tensorflow. Besides, we will describe the main challenges and the many steps that are needed to obtain the sought result, and the future goals set for this work.