index.json

[{"authors":["Shishir Adhikari"],"categories":[],"content":"The picture below summarizes very basic idea behind Thermodynamics, Equilibrium Statistical Mechanics, and Nonequilibrium Statistical Mechanics.  Basic Idea based on Elements of Nonequilibrium Statistical Mechanics book by V. Balakrishnan  \n","date":1554177600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1554177600,"objectID":"1e49c5fb01016caa75507f43350577d6","permalink":"https://arshishir.github.io/post/basic/","publishdate":"2019-04-02T00:00:00-04:00","relpermalink":"/post/basic/","section":"post","summary":"The picture below summarizes very basic idea behind Thermodynamics, Equilibrium Statistical Mechanics, and Nonequilibrium Statistical Mechanics.  Basic Idea based on Elements of Nonequilibrium Statistical Mechanics book by V. Balakrishnan","tags":["Thermodynamics","Equilibrium Stat mechanics","Non-equilibrium Stat mechanics"],"title":"Thermodynamics, ESM, and NESM","type":"post"},{"authors":["Shishir Adhikari"],"categories":[],"content":" For the past year or so, I have been dabbling in machine learning. Last summer, I decided to spend some serious amount of time exploring ML. As a beginner, I got overwhelmed by the plethora of resources out there. Meanwhile, I convinced one of my friends to go on this exploration with me. We decided to go through lectures on machine learning by Andrew Ng. The idea was to watch his old Coursera lectures and meet once a week and discuss those ideas.\nFortunately that summer, I also got an opportunity to mentor a high school student. He wanted to explore some computer science related topic. So, I decided to teach him machine learning. This decision forced me to pick a book and some sets of lecture notes. I choose Pattern Recognition and Machine Learning (PRML) by Christopher Bishop and Deep learning specialization by Andrew Ng. With the added responsibility of teaching, I strated going through first few chapters of PRML. Based on PRML and Deep Learning specialization, I made lecture notes so that I could convey those ideas clearly. It was fun!\nIn the mean that I came across some other very useful resources. Here’s a list of some of those resources:\nSupport Vector Machine  One of the most gentle and clear introductions to SVM by Patrick Winston\n Interesting and very helpful SVM recitation session by Jessica Noss\n  Machine Learning lectures by Nando Freitas I haven’t gone through the whole lecture series. However, I watched quite a few lectures by him and they are some of the clearest introductions those topics on ML that I found on youtube.\nMachine Learning lectures by Ali Ghodsi One of the hidden gems of youtube Machine learning lecture series . I have watched a couple of this lecture and they are brilliant. I would highly recommend his lecture on Decision Tree.\nThis is by no means a complete list. I went through many more resources. I will add those and the ones that I am currently going through soon.\n ","date":1549170000,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1549170000,"objectID":"074a4306d5932cf3558b174f03728e1a","permalink":"https://arshishir.github.io/post/expo_in_ml/","publishdate":"2019-02-03T00:00:00-05:00","relpermalink":"/post/expo_in_ml/","section":"post","summary":"For the past year or so, I have been dabbling in machine learning. Last summer, I decided to spend some serious amount of time exploring ML. As a beginner, I got overwhelmed by the plethora of resources out there. Meanwhile, I convinced one of my friends to go on this exploration with me. We decided to go through lectures on machine learning by Andrew Ng. The idea was to watch his old Coursera lectures and meet once a week and discuss those ideas.","tags":["Machine learning","Deep Learning","SVM","Book","PRML"],"title":"Exploration in ML","type":"post"},{"authors":null,"categories":["Resources"],"content":" Status Experimental \u0026amp; Ongoing\nResearch As an academician, we have been mostly exploring theoretical aspects of machine learning. This project grew out of our need to get our hands dirty. So, the idea is to explore deep learning by figuring out how to crack captcha. We will play with different DNN models. Here, the idea is not to come up with an original method but to use previously tested models. Hopefully, while doing so we learn how to code up the model in different platforms and develop a work flow for machine learning projects.\nTools For scientific computing, we have have been using $Mathematica$. So, firstly we will $Mathematica$’s inbuilt Neural Network functions and tools.\nWe have used $Python$ for data generation. We would also like to use $Python$’s ML packages like: $PyTorch, Keras, Fastai$.\nLastly, we would like to use $Julia$’s ML packages like: $Flux, KNet$.\nProcedure  Using the following python code to create 50,000 Captcha:  import random import numpy as np import string from captcha.image import ImageCaptcha image = ImageCaptcha() clist = list(string.ascii_uppercase+string.digits) random.shuffle(clist) labels = [] for j in range(50000): randstr = '' for i in range(4): randstr = randstr + random.choice(clist) labels.append(randstr) data = image.generate(randstr) image.write(randstr, '~/Desktop/Neural_nets/data_10/'+randstr+'.png') with open('~/Desktop/Neural_nets/data_10/'+'labels.txt','w') as f: for labs in labels: f.write('%s\\n' %labs)   We are in a process of writing and training a model like: VGG16  Note: This post is incomplete\u0026hellip;\n ","date":1548997200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548997200,"objectID":"548109ddd9f5067f214b25dcd077e7fa","permalink":"https://arshishir.github.io/projects/cracking_captach/","publishdate":"2019-02-01T00:00:00-05:00","relpermalink":"/projects/cracking_captach/","section":"projects","summary":"Status Experimental \u0026amp; Ongoing\nResearch As an academician, we have been mostly exploring theoretical aspects of machine learning. This project grew out of our need to get our hands dirty. So, the idea is to explore deep learning by figuring out how to crack captcha. We will play with different DNN models. Here, the idea is not to come up with an original method but to use previously tested models.","tags":["machine-learning","deep-learning","captcha","python","mathematica","julia"],"title":"Cracking Captcha using Deep Learning","type":"projects"},{"authors":["Shishir Adhikari"],"categories":[],"content":" This is my attempt to keep track of interesting science related websites and blogs\nBlogs  https://johncarlosbaez.wordpress.com/ https://mgbukov.github.io/  Websites  John Carlo Baez  http://math.ucr.edu/home/baez/ This webpage has many interesting topics. I have particularly gone through his exposition to information geometry (http://math.ucr.edu/home/baez/information/), which is engaging and interesting. http://math.ucr.edu/home/baez/stoch_stable.pdf Quantum Techniques for Stochastic Mechanics. A book, I hope to go through one day.   Gavin E. Crooks\n http://threeplusone.com/gec/ http://threeplusone.com/FieldGuide.pdf (Field Guide to Continuous Probability Distributions) http://threeplusone.com/on_information.pdf (On Measures of Entropy and Infromation) http://threeplusone.com/Crooks-FisherInfo.pdf (Fisher Information and Statistical Mechanics) http://threeplusone.com/Crooks-Whither.pdf (Whither Time’s Arrow?)  Machine Learning\n https://arxiv.org/pdf/1803.08823.pdf http://www.offconvex.org/    This is by no means a complete list. I will keeping adding.\n ","date":1548997200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548997200,"objectID":"dc1dccdc8cd62a33592189933c09d692","permalink":"https://arshishir.github.io/post/interesting-_website/","publishdate":"2019-02-01T00:00:00-05:00","relpermalink":"/post/interesting-_website/","section":"post","summary":"This is my attempt to keep track of interesting science related websites and blogs\nBlogs  https://johncarlosbaez.wordpress.com/ https://mgbukov.github.io/  Websites  John Carlo Baez  http://math.ucr.edu/home/baez/ This webpage has many interesting topics. I have particularly gone through his exposition to information geometry (http://math.ucr.edu/home/baez/information/), which is engaging and interesting. http://math.ucr.edu/home/baez/stoch_stable.pdf Quantum Techniques for Stochastic Mechanics. A book, I hope to go through one day.   Gavin E. Crooks\n http://threeplusone.","tags":[],"title":"Interesting Science Blogs and Websites","type":"post"},{"authors":["Shishir Adhikari"],"categories":[],"content":"Collections of papers that I would like to go over.\n Stochastic Thermodynamics\n Arcsine laws in stochastic thermodynamics (Barato et al.)\nhttps://arxiv.org/pdf/1712.00795.pdf Thermodynamic cost of external control(Barato et al.)\nhttps://arxiv.org/abs/1704.03480  Ecology\n Constrained optimization as ecological dynamics with applications to random quadratic programming in high dimensions (Mehta et al.)\nhttps://arxiv.org/pdf/1809.04221.pdf  Protein\n The strength of protein-protein interactions controls the information capacity and dynamical response of signaling networks (Wang et. al)\nhttps://arxiv.org/abs/1811.05371  Reinforcement Learning for Physics Problems\n Reinforcement Learning in Different Phases of Quantum Control (Bukov et al.)\nhttps://arxiv.org/pdf/1705.00565.pdf Reinforcement learning for autonomous preparation of Floquet-engineered states: Inverting the quantum Kapitza (Bukov) https://arxiv.org/pdf/1808.08910.pdf  Speed Limit\n Geometric Speed Limit of Accessible Many-Body State Preparation (Bukov et al.)\nhttps://arxiv.org/pdf/1804.05399.pdf   ","date":1548997200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548997200,"objectID":"e26625a22fd6b1383a1cb8d032710e80","permalink":"https://arshishir.github.io/post/journal-club/","publishdate":"2019-02-01T00:00:00-05:00","relpermalink":"/post/journal-club/","section":"post","summary":"Collections of papers that I would like to go over.\n Stochastic Thermodynamics\n Arcsine laws in stochastic thermodynamics (Barato et al.)\nhttps://arxiv.org/pdf/1712.00795.pdf Thermodynamic cost of external control(Barato et al.)\nhttps://arxiv.org/abs/1704.03480  Ecology\n Constrained optimization as ecological dynamics with applications to random quadratic programming in high dimensions (Mehta et al.)\nhttps://arxiv.org/pdf/1809.04221.pdf  Protein\n The strength of protein-protein interactions controls the information capacity and dynamical response of signaling networks (Wang et.","tags":[],"title":"Journal Club","type":"post"},{"authors":["Shishir Adhikari"],"categories":[],"content":"As a Ph.D. student, we are exposed to lots of new ideas and concepts. The posts on this website are my attempt to keep track of those ideas and concepts. I started a blog a couple of years ago but failed to maintain it. This is my second go at it. Better late than never.\nI set up this website using Hugo framework with the academic theme. I have high hopes from this set up as it natively supports $\\LaTeX$.\n","date":1548997200,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1548997200,"objectID":"cce922e87b95a04be05a6287b9a23955","permalink":"https://arshishir.github.io/post/why/","publishdate":"2019-02-01T00:00:00-05:00","relpermalink":"/post/why/","section":"post","summary":"As a Ph.D. student, we are exposed to lots of new ideas and concepts. The posts on this website are my attempt to keep track of those ideas and concepts. I started a blog a couple of years ago but failed to maintain it. This is my second go at it. Better late than never.\nI set up this website using Hugo framework with the academic theme. I have high hopes from this set up as it natively supports $\\LaTeX$.","tags":[],"title":"Why?","type":"post"},{"authors":["Shishir R. Adhikari","Jacob Moran","Christopher Weddle","Michael Hinczewski"],"categories":null,"content":"","date":1534478400,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1534478400,"objectID":"c56805872e9e4a9b26dbc3ededf6593e","permalink":"https://arshishir.github.io/publication/plos_2018/jimm/","publishdate":"2018-08-17T00:00:00-04:00","relpermalink":"/publication/plos_2018/jimm/","section":"publication","summary":"The adherens junctions between epithelial cells involve a protein complex formed by E-cadherin, β-catenin, α-catenin and F-actin. The stability of this complex was a puzzle for many years, since in vitro studies could reconstitute various stable subsets of the individual proteins, but never the entirety. The missing ingredient turned out to be mechanical tension: a recent experiment that applied physiological forces to the complex with an optical tweezer dramatically increased its lifetime, a phenomenon known as catch bonding. However, in the absence of a crystal structure for the full complex, the microscopic details of the catch bond mechanism remain mysterious. Building on structural clues that point to α-catenin as the force transducer, we present a quantitative theoretical model for how the catch bond arises, fully accounting for the experimental lifetime distributions. The underlying hypothesis is that force induces a rotational transition between two conformations of α-catenin, overcoming a significant energy barrier due to a network of salt bridges. This transition allosterically regulates the energies at the interface between α-catenin and F-actin. The model allows us to predict these energetic changes, as well as highlighting the importance of the salt bridge rotational barrier. By stabilizing one of the α-catenin states, this barrier could play a role in how the complex responds to additional in vivo binding partners like vinculin. Since significant conformational energy barriers are a common feature of other adhesion systems that exhibit catch bonds, our model can be adapted into a general theoretical framework for integrating structure and function in a variety of force-regulated protein complexes.","tags":["biophysics","cadherin","modeling","actin","catch-bond"],"title":"Unraveling the mechanism of the cadherin-catenin-actin catch bond","type":"publication"},{"authors":["Mark P. Taylor","Yuting Ye","Shishir R. Adhikari"],"categories":null,"content":"","date":1448254800,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1448254800,"objectID":"30ca1164fcf4a61d8cda6d5fb92afbda","permalink":"https://arshishir.github.io/publication/jcp_2015/jimm/","publishdate":"2015-11-23T00:00:00-05:00","relpermalink":"/publication/jcp_2015/jimm/","section":"publication","summary":"The conformation of a polymer chain in solution is coupled to the local structure of the surrounding solvent and can undergo large changes in response to variations in solvent density and temperature. The many-body effects of solvent on the structure of an n-mer polymer chain can be formally mapped to an exact n-body solvation potential. Here, we use a pair decomposition of this n-body potential to construct a set of two-body potentials for a Lennard-Jones (LJ) polymer chain in explicit LJ solvent. The solvation potentials are built from numerically exact results for 5-mer chains in solvent combined with an approximate asymptotic expression for the solvation potential between sites that are distant along the chain backbone. These potentials map the many-body chain-in-solvent problem to a few-body single-chain problem and can be used to study a chain of arbitrary length, thereby dramatically reducing the computational complexity of the polymer chain-in-solvent problem. We have constructed solvation potentials at a large number of state points across the LJ solvent phase diagram including the vapor, liquid, and super-critical regions. We use these solvation potentials in single-chain Monte Carlo (MC) simulations with n ≤ 800 to determine the size, intramolecular structure, and scaling behavior of chains in solvent. To assess our results, we have carried out full chain-in-solvent MC simulations (with n ≤ 100) and find that our solvation potential approach is quantitatively accurate for a wide range of solvent conditions for these chain lengths.","tags":["solvation","Lennard-Jones","Accurate"],"title":"Conformation of a flexible chain in explicit solvent: Accurate solvation potentials for Lennard-Jones chains","type":"publication"},{"authors":["Mark P. Taylor","Shishir R. Adhikari"],"categories":null,"content":"","date":1311825600,"expirydate":-62135596800,"kind":"page","lang":"en","lastmod":1311825600,"objectID":"1dc7fc42ee29c7b8968d25e229e551f8","permalink":"https://arshishir.github.io/publication/jcp_2011/jimm/","publishdate":"2011-07-28T00:00:00-04:00","relpermalink":"/publication/jcp_2011/jimm/","section":"publication","summary":"The average conformation of a flexible chain molecule in solution is coupled to the local solvent structure. In a dense solvent, local chain structure often mirrors the pure solvent structure, whereas, in a dilute solvent, the chain can strongly perturb the solvent structure which, in turn, can lead to either chain expansion or compression. Here we use Monte Carlo (MC) simulation to study such solvent effects for a short Lennard-Lones (LJ) chain in monomeric LJ solvent. For an n-site chain molecule in solution these many-body solvent effects can be formally mapped to an n-body solvation potential. We have previously shown that for hard-sphere and square-well chain-in-solvent systems this n-body potential can be decomposed into a set of two-body potentials. Here, we show that this decomposition is also valid for the LJ system. Starting from high precision MC results for the n = 5 LJ chain-in-solvent system, we use a Boltzmann inversion technique to compute numerically exact sets of two-body solvation potentials which map the many-body chain-in-solvent problem to a few-body single-chain problem. We have carried out this mapping across the full solvent phase diagram including the dilute vapor, dense liquid, and supercritical regions and find that these sets of solvation potentials are able to encode the complete range of solvent effects found in the LJ chain-in-solvent system. We also show that these two-site solvation potentials can be used to obtain accurate multi-site intramolecular distribution functions and we discuss the application of these exact short chain potentials to the study of longer chains in solvent.","tags":["solvation","Lennard-Jones","Exact"],"title":"Conformation of a flexible chain in explicit solvent: Exact solvation potentials for short Lennard-Jones chains","type":"publication"}]