Name

sxresample - calculate resampled 3D reconstructions

TO BE WRITTEN

Usage

Usage in command lines:

sxbootstrap_bigdisk.py prj_stack wgtfile outdir bufprefix --nvol --nbufvol --seedbase --snr --genbuf --CTF --ngroup --npad --zero_wgts --MPI --core --verbose

Usage in python programming:

bootstrap_bigdisk(prj_stack, wgtfile, outdir, bufprefix, nvol, nbufvol, seedbase, snr, genbuf, sharebuf)

Input

prj_stack: projection stack file
wgtfile: the weight of each projection. This is a text file with the weights of projections int he first column. The weights is proportional to the probability of a projection to be selected in bootstrap calculation. This file is generated using sxbootstrap_calcwgts.py.
buf_prefix: the prefix for the buffer file. These files were generated using sxbootstrap_genbuf.py.
nvol: number of bootstrap volumes to be generated
nbufvol: number of fftvols in the memory
genbuf: whether generate the buffer
ngroup: how many groups to use (each group will share a buffer)
zero_wgts: Percentage of maximum weights to be set to zero (default 0.0)
seedbase: random seed base
core: keep projections with high weights without resampling
verbose: verbose level: 0 no, 1 yes

Output

outdir: bootstrap volumes will be in put in this directory with names "bsvolxxxx.hdf", where xxxx is the rank of CPU. Warning: If the output directory already exists, the program will crash. Please change the name of directory and restart the program.

Options

nvol: number of bootstrap volumes to be generated
nbufvol: number of volumes to be generated simultaneously. It depend on how much memory there is available for each CPU and the image size in pixels. A single 3D reconstruction will require 0.000512*nx^3 MB of memory. Thus, if image size is 75, one reconstruction will take 216 MB of memory. If image size is 64, it will take 134 MB of memory. The nbufvol is computed by dividing the total memory available for a CPU by the memory required for one 3D reconstruction. Note on a cluster memory per CPU is typically memory available on a node divided by the number of CPUs.
CTF: use CTF in reconstruction
snr: Signal to Noise ratio
seedbase: seed for the random number generator. If in order to compute sufficient number of bootstrap volumes the program is ran multiple times the seedbase has to be modified (increased by the number of volumes already generated).

Description

This is third and the most time consuming step of real-space variance calculations.

Method

Reference

Author / Maintainer

Wei Zhang

Keywords

category 1: APPLICATIONS

Files

applications.py, sxbootstrap_bigdisk.py

Maturity

stable: works for most people, has been tested; test cases/examples available.

Bugs

None. It is perfect.