Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Very Low Activity

Commits : Listings

Analyzed about 7 hours ago. based on code collected 1 day ago.
Nov 05, 2025 — Nov 05, 2026
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
AsFer Design Document - Theory - Oracle version of Condorcet Jury Theorem circuit and conflicts
Srinivasan Kannan
More... over 8 years ago
AsFer Design Document - Theory - Oracle version of Condorcet Jury Theorem circuit and conflicts
Srinivasan Kannan
More... over 8 years ago
AsFer Design Document - Theory - Oracle version of Condorcet Jury Theorem circuit and conflicts
Srinivasan Kannan
More... over 8 years ago
AsFer Design Document - Theory - Oracle version of Condorcet Jury Theorem circuit and conflicts
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Scheduler Analytics and Perf
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - PRAM example (AMD Kaveri)
Srinivasan Kannan
More... over 8 years ago
-------------------------------------------------------------------------------------------------------------- Scheduler Analytics Update - Rewrite of scheduler analytics webserver functionality and Spark log mapreduce of "perf sched script" -------------------------------------------------------------------------------------------------------------- 1.SparkKernelLogMapReduceParser.py has been rewritten to remove hardcoded filenames and patterns - new function log_mapreducer() has been defined for passi$ and pattern as arguments. 2.software_analytics/DeepLearning_SchedulerAnalytics.py and software_analytics/SchedulerAnalytics_WebServer.py h$ been refactored and get_stream_data() (@Generic Socket Web Server) decorated function has been moved to software$ 3.software_analytics/DeepLearning_SchedulerAnalytics.py invokes log_mapreducer() defined in SparkKernelLogMapRed$ for new scheduler performance file perf.data.schedscript and pattern sched_stat_runtime. 4.New perf.data file has been created from perf utility by "perf sched record" commandline 5.New perf.data.schedscript has been created from perf.data by "perf sched script" commandline for printing scheduler context switches and runtime in nanoseconds 6.logs for both have been committed to software_analytics/testlogs/DeepLearning_SchedulerAnalytics.log.30May2018 and software_analytics/testlogs/SchedulerAnalytics_WebServer.log.30May2018 7.Perf utility support has been added for potential clockticks-to-processes dynamic hash table creation later.
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Correction to Ray Shooting Factorization Analysis
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Computational Geometric Factorization - Ray Shooting Queries and Cramer Prime Gap Updates - Correction to Commit Notes and Updated analysis of PRAM factorization by ray shooting queries
Srinivasan Kannan
More... over 8 years ago
-------------------------------------------------------------------------------------------------------------- Computational Geometric Factorization - Ray Shooting Queries and Cramer Prime Gap Updates -------------------------------------------------------------------------------------------------------------- 1.DiscreteHyperbolicFactorizationUpperbound_TileSearch_Optimized.py has been updated to include a new ray shooting query function for Cramer Prime Gaps (sqrt(p)*log(p)) 2.All Four Ray Shooting Query functions have been executed on two integers and factorization logs have been captured in testlogs/ (Single Core - Spark 2.3) 3.Both logs show don't show an exact shoot (but are sufficiently close) 4.Of the Four, Hardy-Ramanujan ray shooting is still the best and estimates are close to exact factors and sometimes exact. 5.Baker-Harman-Pintz, Hardy-Ramanujan-PrimeNumberTheorem Ray Shooting functions have been slightly changed for initial prime factor, constant for Prime Counting Function normal order etc., ---------- 6.Analysis: ---------- These ray shooting queries are Monte Carlo algorithms with an error bounded by the spacing between two prime factors. Ray shooting to find a prime factor by Hardy-Ramanujan succeeds with a probability loglogN/N which is loglogN/2^logN=loglogN/(2^(2^loglogN)) which is exponentially small. But real benefit of ray shooting queries is in the following factorization algorithm: - Sequentially ray shoot O(loglogN) queries in O(loglogN) time to create beacons - Distance between any two such beacon in the hyperbolic arc bow is O(N/loglogN) - This creates O(loglogN) hyperbolic pixelated segments of length O(N/loglogN) - Pixelation of O(N/loglogN) sized segment requires Parallel RAM sorting - Each pixelated segment can be binary searched in log(N/loglogN) time - Sequentially this requires O(loglogN)*O(log(N/loglogN)) time for all segments - Total factorization time = O(loglogN) + O(loglogN)*O(log(N/loglogN)) < O(loglogN) * O(logN) - This PRAM algorithm does factorization in at most O(loglogN*logN) time compared to O((logN)^2) or O((logN)^3) PRAM time algorithms using segment trees and sorting described earlier.
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Computational Geometric Factorization - Ray Shooting Queries - Hardy-Ramanujan-Prime Number Theorem and Baker-Harman-Pintz Gap between Prime Numbers - Implementation
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Hardy-Ramanujan-Prime Number Theorem and Baker-Harman-Pintz Ray Shooting Queries - Factorization - Theory
Srinivasan Kannan
More... over 8 years ago
------------------------------------------------------------------------------------------------------------- Computational Geometric Factorization - Ray Shooting Queries - Hardy-Ramanujan-Prime Number Theorem and Baker-Harman-Pintz Gap between Prime Numbers ------------------------------------------------------------------------------------------------------------- 1.DiscreteHyperbolicFactorizationUpperbound_TileSearch_Optimized.py has been updated to include two new functions for ray shooting queries - 1) based on Hardy-Ramanujan-Prime Number Theorem estimate for number of primes between two prime factors 2) based on Baker-Harman-Pintz Gap between prime numbers. 2.These two functions are quite nascent and might be refined in next commits. 3.Two numbers have been factored and all three ray shooting query algorithms print the approximate location of factors. 4.Of the three, Hardy-Ramanujan query is very simple and finds the factor exactly in most executions (logs show an exact match for one factor in both integers). Hardy-Ramanujan-Prime Number Theorem and Baker-Harman-Pintz estimates are too skewed. But theoretically latter two estimates are tighter approximations compared to Hardy-Ramanujan ray shooting. 5.Logs for these two integer factorizations have been committed to testlogs/
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Hardy-Ramanujan Approximate Ray Shooting
Srinivasan Kannan
More... over 8 years ago
--------------------------------------------------------------------------------------------------------------- Computational Geometric Factorization - Single Core - Hardy-Ramanujan Ray Shooting Queries --------------------------------------------------------------------------------------------------------------- 1. Hard-Ramanujan Approximate Ray Shooting Queries function has been refactored to compute the angle and tangent of the ray simply as functions of multiples of N/k*loglogN. 2. Precisely angle between the rays must be a function of spacing/gap between prime factors of N which is at present assumed to be equal. 3. Cramer's bound for Spacing between primes is O(sqrt(p)*log(p)) assuming Riemann Hypothesis is true while the best known modern bound proved in 1995 and 2001 by [Baker-Harman-Pintz] is p^(0.525) which is unconditional. 4. Prime number theorem implies prime counting function ~ N/logN. Probability that a prime below N is a prime factor of N is roughly: (loglogN*logN)/N - ratio of Hardy-Ramanujan estimate and Prime Counting Function. 5. Even with this equal spacing between prime factors, two experimental logs committed herewith show previous approximate ray shooting is exact, passing through an exact factor found by tile binary search. Both logs have an exact match for one factor - approximate and exact ray shooting coincide. 6. By replacing equal spacing by a function of Cramer and Baker-Harman-Pintz Prime Gap estimate might accurately fine tune the location of a factor further.
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - fMRI to Connectome Deep Learning
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - ThoughtNet Evocation based on Graph Tensor Neuron Network Intrinsic Merit
Srinivasan Kannan
More... over 8 years ago
-------------------------------------------------------------------------------------------------------------- ThoughtNet Reinforcement Learning - Contextual Multi Armed Bandit Evocation Update -------------------------------------------------------------------------------------------------------------- 1. RecursiveGlossOverlap_Classifier.py has been updated to return both the definition graph NetworkX object and Recursive Gloss Overlap Intrinsic Merit 2. RecursiveLambdaFunctionGrowth.py has been updated to return all intrinsic merits from grow_lambda_function3(): 2.1 Korner Entropy of Definition Graph 2.2 Bose Einstein Intrinsic Fitness 2.3 Graph Tensor Neuron Network Instrinsic Merit 2.4 Definition Graph Density 2.5 Definition Graph Recursive Gloss Overlap Intrinsic Merit 2.6 Maximum Per Random Walk Graph Tensor Neuron Network Intrinsic Merit 3.ThoughtNet/Create_ThoughtNet_Hypergraph.py has been updated to make a choice for merit criterion by choosing between sentiment scoring and graph tensor neuron network intrinsic merit for ThoughtNet HyperGraph edges 4.ThoughtNet/ThoughtNet_Hypergraph_Generated.txt has been recreated based on Graph Tensor Neuron Network Intrinsic Merit and the sorted order of hyperedges for most vertices is different compared to sentiment scoring earlier 5.Logs for previous ThoughtNet creation has been committed to ThoughtNet/testlogs/Create_ThoughtNet_Hypergraph.GraphTensorNeuronNetwork.log.22May2018 6.Logs for ThoughtNet Contextual Multi Armed Bandit Evocation has been committed to testlogs/DeepLearning_ReinforcementLearningThoughtNet.log.22May2018
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - ThoughtNet
Srinivasan Kannan
More... over 8 years ago
------------------------------------------------------------------------------------------------------------- ThoughtNet - Contextual Multi Armed Bandit Evocation - Update ------------------------------------------------------------------------------------------------------------- 1.Python implementation DeepLearning_ReinforcementLearningThoughtNet.py has been updated to Classify the inpu text by Recursive Gloss Overlap Definition Graph Core Number Classifier and lookup the classes in the Thought Hypergraph 2.ThoughtNet Hypergraph has been recreated having classes mostly belonging to terrorism, economics, social pr extracted from chronologically older thought edges. 3.Input to ThoughtNet Reinforcement Learning has been updated to have recent events/thoughts/stimuli 4.Logs for the evocation of the new thoughts on the ThoughtNet hypergraph is in ThoughtNet/ThoughtNet_Hypergr 5.Logs show a recent terrorist event evoking an older event of similar class among others.
Srinivasan Kannan
More... over 8 years ago
--------------------------------------------------------------------------------------------------------------- ThoughtNet - Contextual Multi Armed Bandit Evocation - Update --------------------------------------------------------------------------------------------------------------- 1.Python implementation DeepLearning_ReinforcementLearningThoughtNet.py has been updated to Classify the input text by Recursive Gloss Overlap Definition Graph Core Number Classifier and lookup the classes in the ThoughtNet Hypergraph 2.ThoughtNet Hypergraph has been recreated having classes mostly belonging to terrorism, economics, social progress etc., extracted from chronologically older thought edges. 3.Input to ThoughtNet Reinforcement Learning has been updated to have recent events/thoughts/stimuli 4.Logs for the evocation of the new thoughts on the ThoughtNet hypergraph is in ThoughtNet/ThoughtNet_Hypergraph_Generated.txt 5.Logs show a recent terrorist event evoking an older event of similar class among others.
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Two Level Locality Sensitive Hashing for Electronic Voting Machines
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Graph Mining of EventNet DOT files
Srinivasan Kannan
More... over 8 years ago
-------------------------------------------------------------------------------------------------------------- Graph Mining - GSpan Algorithm Implementation Update - for mining patterns in EventNet DOT files -------------------------------------------------------------------------------------------------------------- 1.GSpan GraphMining implementation has been updated to read EventNet DOT file as input 2.This enables mining EventNet Event-Partakers Graph for patterns 3.EventNet has been implemented in NeuronRain AsFer in C++ and Python for persisting general purpose Cloud Event-Actors conversations as Graph. 4.EventNet is also an Economic Trading Network/Money Flow Market for KingCobra Fictitious Neuro Electronic Currency and Buyer-Seller networks in Neuro Money Flow Markets e.g AsFer-KingCobra Cloud Move of Neuro Currency writes out an EventNet edge in a DOT file from Buyers to Sellers 5.Mining this Economic Network EventNet requires Frequent SubGraph Mining(GSpan) 6.log for this has been committed to testlogs/GraphMining_GSpan.log.21May2018
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Brain Graph (Connectomes) and Graph Expanders
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Contradition in Depth-2 Majority Circuit Composition and ZFAOC
Srinivasan Kannan
More... over 8 years ago
Updated Requirements.txt
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - SAT Solver LSMR initial guess
Srinivasan Kannan
More... over 8 years ago
------------------------------------------------------------------------------------------------------------- SAT Solver Update - Initial Guess for LSMR - 1000 variables and Alpha=4.267 and 1.0 ------------------------------------------------------------------------------------------------------------- 1. SAT Solver has been updated to invoke lsmr() function with an initial guess assignment vector - hardcoded to a decimal value < 1 and close to 0 for all variables. 2. Binary and Real parities (Riemann Discrete Integral Sum) are printed along side the assignment vector for each random SAT 3. SAT Solver has been executed for 1000 variables - Alpha=1.0 and 1000 variables - Alpha=4.267 for this initial guess and logs have been captured in testlogs/CNFSATSolver.1000variablesAlpha1.0.log.initial_guess.18May2018 and testlogs/CNFSATSolver.1000variablesAlpha4.267.log.initial_guess.18May2018 4. As usual Alpha=1.0 converges to ~95% and Alpha=4.267 to ~90% for 150 and 20 random SAT iterations respectively. 5. Non-uniform choice probability for Alpha=4.267 is alomst close to theoretical random matrix probability 1/sqrt(m*n)
Srinivasan Kannan
More... over 8 years ago
Updated AsFer Design Document - Computation Geometric Factorization on Single Core, Spark 2.3 + jdk 1.8
Srinivasan Kannan
More... over 8 years ago