mullinix / cuda_linear_model_mc_bs

linboot

CUDA-C (Monte-Carlo) non-parametric bootstrap confidence intervals on the slope of a linear model

Introduction/Basics

This software reads a comma-delimited input file of (x,y) pairs and computes percentile and BCa confidence intervals on the best-fit (BLUE) slope, intercept, and calculates the Pearson correlation coefficient rho.

Motivation

This code was written in support of academic research by the publisher, who is a doctoral student in the Computational Science Research Center (CSRC), as part of a multidisciplinary project with the Viral Information Instute (VII). The CSRC and VII are both located at San Diego State University.

Build Status

New: Use a Docker container:

• Docker repo: https://hub.docker.com/r/mullinix/linboot
• Github repo: https://github.com/mullinix/docker_linboot

Builds! Runs! Tested on GNU/Linux:

• Ubuntu 18.04 LTS: uname -a (kernel 4.15.0-39-generic)
• NVCC/toolkit: nvcc --version Cuda compilation tools, release 9.1, V9.1.85
• nVidia driver: nvidia-smi 396.54
• GNU Scientific Library: dpkg -l libgsl*
  • libgsl-dev, version: 2.4+dfsg-6
  • libgsl23, version: 2.4+dfsg-6
  • libgslcblas0, version: 2.4+dfsg-6

Compiling

A makefile is included.

make

Running

./linboot <filename> <alpha-value>

Output

The output includes a text file <filename>-CI.dat which lists the results in a comma-separated list of the following:

statistic,lower_percentile,upper_percentile,lower_BCa,upper_BCa,SE_lower,SE_upper,median,mean,#pts,#bs_iterations

as well as a text file <filename>-<statistic>-histogram.dat containing comma-separated histogram data, as

bin_RHS,bin_count

Post-processing

A sample application plot_hist.R to plot the histogram is included. It is written in the R language, and depends on ggplot2.

Test data

• Test data can be generated using generate_test_data.R.
• A test data file is included, called testdata.csv.

Legal

License -- MIT. Use/Share away!

Disclaimers:

1. Any/all libraries used have their own licenses/rights; be sure your use is compatible with these !
2. This content reflects the opinions of the content creator and is not meant to represent the CSRC, VII or SDSU.