UB Home Page CCR Home Page HWI Home Page
DOWNLOAD
Announcements
DREAR
Bugs
FAQs
Funding
Manuals
Multiprocessing
Parameters
References
Personnel
Report Structures
SnB Server
SnB Successes
SnB Tutorial
Contact
SnB Home Page

Trials & Cycles

  • Starting Phases From: Select from Random Atoms, Random Phases, Variable Input Phases, Fixed Input Phases, and Model Structure Atoms.

    Typically, a user will use Random Atoms to generate trial structures with randomly positioned atoms, and a structure-factor calculation will be performed automatically to generate random phases. Alternatively, Random Phases can be generated directly. However, given partial knowledge of the structure, the user may want to prepare an input file consisting of either a set of phases or a set of atoms. Choose Variable Input Phases if you have a set of phases in a file and you want to allow these phases to vary during the Shake-and-Bake phasing procedure. Choose Fixed Input Phases if you have a file containing input phases that you want to remain fixed during the phasing procedure, varying only additional phases. In this case, starting values for the additional phases will be supplied by the random-number generator. Choose Model Structure Atoms if you want to input a set of atoms and process them as a single trial structure.
     
  • Number of Trials: Enter the number of trial structures you wish to process with the Shake-and-Bake phasing procedure.
     
  • Starting at Trial: Enter the serial number of the first trial structure to be processed. For example, this feature could be useful if you had previously been processing this set of trials but experienced a system crash after 4993 trials had been completed. By starting at trial 4994, you could resume where the aborted job stopped. This parameter can also be used in conjunction with Number of Trials to repeat a trial of particular interest. In this case, start at the desired trial, request that a single trial be processed, and be careful not to change any of the other parameters that would affect the outcome.
     
  • Random Seed: Enter a random seed. It is best if a large prime number is chosen. A feature is provided that will allow the user to click and choose from a stored list of prime numbers. Alternately, the user may simply enter a value into the text area.
     
  • Number of Shake-and-Bake Cycles: Enter the desired number of Shake-and-Bake refinement cycles that are to be applied to every trial structure.
     
  • Keep complete trace file: Selecting "yes" will cause SnB to produce an entry in the trace file for every cycle. Otherwise, entries are written only for the last cycle (i.e. once per trial).
     
  • Input Phase File: In the event that you have chosen to start with an input set of phases, enter the file name. This file should contain H, K, L, |F|, Sig(F), |E|, Sig(E), and the starting phase (in radians) in FORMAT(3I5, 2F9.2, 2F9.3, F9.4).
     
  • Input Atom File: In the event that you have chosen to start with an input set of atoms, enter the file name. This file should contain fractional coordinates (x, y, z) in FORMAT (3F10.6). SnB will read until an end of file is encountered, so all unwanted coordinates must be deleted. The output peak and atom files from previous SnB jobs are in the proper format but may contain spurious peaks.
     
  • R-Ratio test?: This feature can be used to cause early termination of trials that fail the R-Ratio test. Trials fail this test if the R-Ratio value in any cycle exceeds the specified cutoff value (the default cutoff value is 0.2). Trials with large R-Ratio values are likely to be false minima, so early termination in hopeless situations can reduce the time needed to find a solution. However, this test should be used cautiously, and it should never be used during the first job for any data set. It is difficult to be sure of the proper cutoff value, and an incorrect choice could result in rejection of a solution. This feature should be reserved for difficult cases where the occurrence of false minima is strongly indicated and a large number of trials will likely be required to obtain a solution.