New GENIE PINN and QMLHEP projects

Author: ereinha

_gsocproposals/2026/proposal_GENIE6.md

@@ -0,0 +1,72 @@
+---
+title: Physics-Informed Neural Network Diffusion Equation (PINNDE)
+layout: gsoc_proposal
+project: GENIE
+year: 2026
+organization:
+ - Alabama
+ - FSU
+ - Fermilab
+---
+
+## Description
+There is much interest in building ultra-fast samplers that map a density that is easy to sample from, typically, an n-dimensional normal to a desired n-dimensional density. One way to compute this mapping is to solve the reverse-time diffusion equation [1], which is an integro-differential equation. In Ref. [2], the integral in this equation is approximated using Monte Carlo integration where the integrand is averaged over N (~5K – 10K) points sampled from the desired distribution. Solving this equation is relatively slow, therefore, typically a neural network is trained to model the mapping from the normal to the desired density using training data generated by repeatedly solving the differential equation.  
+
+In last year's work, an alternative approach was investigated: modeling the solution to the differential equation using a physics-informed neural network (PINN) [3]. There is a large upfront cost in training the PINN, but this is subsequently amortized over the fast sampling using the PINN. This project extends last year's work by solving the reverse-time diffusion equation on the generated PINN to evaluate the physical realism of PINN models.
+
+## Duration
+Total project length: 175/350 hours.
+
+## Difficulty Level
+  * Intermediate/Advanced
+
+## Task ideas
+* Use CaloChallenge (Dataset 2 with ~6,400 voxels), build an accurate autoencoder of the showers.
+* Use the code developed in 2025 \[6\] for solving the reverse-time diffusion equation to map an d-dimensional standard normal to the d-dimensional latent space from step 1 and verify that we can accurately simulate showers.
+* Produce an accurate ML model of the q(t, x) d-dimensional vector field, tagged by the log of the energy of the incident particle that appears in the ODE and use the ML model for q(t, x) in lieu of the Monte Carlo approximation of q(t, x) used in the 2025 version of the reverse-time equation solver. I would expect a significant speed improvement in shower generation.
+* Finally, try to solve the reverse-time equation using a PINN, building on the promising results from last year. If this succeeds, we would have a superfast, accurate, particle shower generator.
+* Compare this approach with published CaloChallenge approaches and include our work and these comparisons in a ML paper.
+
+## Expected results
+* Trained graph-based jet classifier
+* Benchmarks on selected datasets
+
+## Test
+* Using PyTorch, solve the damped harmonic oscillator [5] using a PINN. Choose fixed initial conditions:  
+  x(0) = x₀, dx/dz(0) = v₀, with x₀ = 0.7 and v₀ = 1.2.  
+  Condition the PINN on damping ratios in the range ξ = 0.1 to 0.4.  
+  Solve on the domain z ∈ [0, 20]:  
+  d²x/dz² + 2ξ·dx/dz + x = 0
+
+<!-- ## Test 
+Please use [this link](https://docs.google.com/document/d/142YpKV7fJ49zaBZkSBekbBzw43KD71No2K_Jd-n5Neo/edit?usp=sharing) to access the test for this project. -->
+  
+## Requirements 
+  * Experience with numerical solution of ordinary differential equations. 
+  * Familiarity with PyTorch.  
+
+## Difficultly Level
+Advanced
+
+## Mentors
+  * [Harrison B. Prosper](mailto:[email protected]) (Florida State University)
+  * [Pushpalatha Bhat](mailto:[email protected]) (Fermilab)
+  * [Sergei Gleyzer](mailto:[email protected]) (University of Alabama)
+ 
+
+## Links
+1. Cheng Lu†, Yuhao Zhou†, Fan Bao†, Jianfei Chen†, Chongxuan Li‡, Jun Zhu, DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps, arXiv:2206.00927v3, 13 Oct 2022. 
+
+2. Yanfang Lui, Minglei Yang, Zezhong Zhang, Feng Bao, Yanzhao Cao, and Guannan Zhang, Diffusion-Model-Assisted Supervised Learning of Generative Models for Density Estimation, arXiv:2310.14458v1, 22 Oct 2023. 
+
+3. S. Cuomo et al., Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next, https://doi.org/10.48550/arXiv.2201.05624. 
+
+4. https://calochallenge.github.io/homepage/ 
+
+5. https://en.wikipedia.org/wiki/Harmonic_oscillator
+
+6. https://medium.com/@sijiljose.999/gsoc-2025-with-ml4sci-part-i-physics-informed-neural-network-for-diffusion-equation-pinnde-491d46a5b84d
+
+ 
+
+Please **DO NOT** contact mentors directly by email. Instead, please email [[email protected]](mailto:[email protected]) with Project Title and **include your CV** and **test results**. The mentors will then get in touch with you.

_gsocproposals/2026/proposal_QMLHEP13.md

@@ -0,0 +1,52 @@
+---
+title: Quantum Reinforcement Learning for High Energy Physics
+layout: gsoc_proposal
+project: QMLHEP
+year: 2026
+organization:
+ - University of Alabama
+ - Wisconsin
+---
+
+#### Description
+The ambitious HL-LHC program will require enormous computing resources and datasets in the next two decades. New technologies are being sought after to replace the present computing infrastructure. A burning question is whether quantum computers can solve the ever growing demand of computing resources in High Energy Physics (HEP) in general and physics at LHC in particular. Our goal here is to explore and to demonstrate that Quantum Reinforcement Learning can be the new paradigm (Proof of Principle).
+Discovery of new physics requires the identification of rare signals against immense backgrounds. Development of reinforcement learning (RL) methods will greatly enhance our ability to achieve this objective. However, current RL algorithms require high computing resources on large datasets and excessive computing time to achieve good performance. Quantum Computing, where qubits are used instead of bits in classical computers, has the potential to improve the time complexity or data efficiency of classical algorithms.
+With this project we seek to implement Quantum Reinforcement Learning methods for LHC HEP analysis based on the Pennylane framework.
+
+#### Duration
+Total project length: 175 hours.
+
+#### Task Ideas
+  * Identify the bottlenecks of classical deep RL and propose quantum circuits to efficiently bypass computationally challenging parts of the pipeline.
+  * Implement a hybrid or fully quantum deep RL network that may utilize Quanvolutional Neural Networks and a Q-function approximator to minimize the temporal difference error defined by the Bellman equation.
+  * Apply the quantum RL method to HEP, benchmarking it against classical state-of-the-art approaches.
+
+#### Expected Results
+  * Quantum deep reinforcement learning framework.
+  * Applications within HEP data benchmarked against a classical reference model
+
+## Requirements
+  * Solid knowledge of machine learning and deep learning
+  * Knowledge of quantum mechanics
+  * Strong python skills
+  * Ability to work independently and proactive on a research project
+
+## Difficulty
+Intermediate/Advanced
+
+## Test
+Use [this link](https://docs.google.com/document/d/1t2cSxEx3vIa6uirfkMkF92rWZM4tTJ_V-lkpYSdukVQ/edit?usp=sharing) for instructions on completing the test.
+
+## Mentors
+* [Sergei Gleyzer](mailto:[email protected]) (University of Alabama)
+* [Rui Zhang](mailto:[email protected]) (University of Wisconsin-Madison)
+* [Mariia Baidachna](mailto:[email protected])
+
+## Links
+ * [HL-LHC](https://hilumilhc.web.cern.ch/)
+ * [LHC](https://home.cern/science/accelerators/large-hadron-collider)
+ * [Pennylane](https://pennylane.ai)
+ * [Classical RL for HEP](https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2022.929064/full)
+ * [Quantum RL](https://arxiv.org/abs/2510.14595)
+
+Please **DO NOT** contact mentors directly by email. Instead, please email [[email protected]](mailto:[email protected]) with Project Title and **include your CV**. The mentors will then get in touch with you.

_gsocproposals/2026/proposal_QMLHEP14.md

@@ -0,0 +1,49 @@
+---
+title: Automated Scientific Discovery of Quantum Machine Learning Architectures
+layout: gsoc_proposal
+project: QMLHEP
+year: 2026
+organization:
+ - University of Alabama
+---
+
+#### Description
+The ambitious HL-LHC program will require enormous computing resources and datasets in the next two decades. New technologies are being sought after to replace the present computing infrastructure. A burning question is whether quantum computers can solve the ever growing demand of computing resources in High Energy Physics (HEP) in general and physics at LHC in particular. Our goal here is to explore and to demonstrate that Automated Quantum Architecture Search can be used to discover novel solutions.
+Discovery of new physics requires the identification of rare signals against immense backgrounds, prompting the design of quantum machine learning methods. However, developing effective quantum circuits for these tasks is currently a difficult and manual process that relies on trial and error. To fully leverage the potential of Quantum Computing, we must move beyond manual design. Automated scientific discovery, specifically using machine learning to design better quantum machine learning architectures, can transform the way we design quantum circuits and optimize classifiers for HEP data.
+With this project we seek to implement automated methods for discovering optimal quantum machine learning architectures for LHC HEP analysis based on the Pennylane framework.
+
+#### Duration
+Total project length: 175 hours.
+
+#### Task Ideas
+  * Design an automated quantum architecture search pipeline that balances theory and RL driven methods to search a reasonable space of quantum circuits.
+  * Benchmark the discovered architectures against state-of-the-art quantum and classical approaches on HEP data.
+
+#### Expected Results
+  * A pipeline for the automated discovery of quantum circuit architectures tailored to HEP data.
+  * Comparative analysis against state-of-the-art approaches.
+
+## Requirements
+  * Solid knowledge of machine learning and deep learning
+  * Knowledge of quantum mechanics
+  * Strong python skills
+  * Ability to work independently and proactive on a research project
+
+## Difficulty
+Intermediate/Advanced
+
+## Test
+Use [this link](https://docs.google.com/document/d/1t2cSxEx3vIa6uirfkMkF92rWZM4tTJ_V-lkpYSdukVQ/edit?usp=sharing) for instructions on completing the test.
+
+## Mentors
+* [Sergei Gleyzer](mailto:[email protected]) (University of Alabama)
+* [Mariia Baidachna](mailto:[email protected])
+
+## Links
+ * [HL-LHC](https://hilumilhc.web.cern.ch/)
+ * [LHC](https://home.cern/science/accelerators/large-hadron-collider)
+ * [Pennylane](https://pennylane.ai)
+ * [SCP for automated scientific discovery](https://arxiv.org/abs/2512.24189)
+ * [AutoQML](https://arxiv.org/abs/2502.21025)
+
+Please **DO NOT** contact mentors directly by email. Instead, please email [[email protected]](mailto:[email protected]) with Project Title and **include your CV**. The mentors will then get in touch with you.