PAW Physicist Analysis Workstation What is PAW A

PAW: Physicist Analysis Workstation • What is PAW? – A tool to display and manipulate data. • Learning PAW – See ref. in your induction week notes. • Running PAW: 2 Versions: – PAW: 2 windows: • A terminal window for command input • A graphics window for output – PAW++: GUI version • A terminal window for command input • A terminal window for log output • Multiple graphics windows for in/output – Same functionality, but PAW++ allows interaction with data points. • Using PAW – – Involves operations on 3 data types: Vectors: 1, 2 or 3 dimensional arrays Histograms: 1 or 2 dimensional N-tuples: Tables of events 1

PAW: Data Flows Vector ASCII File PAW Histogram N-tuple External data flows but can also create internally HBOOK File • Multiple open at once Main internal data flows • Organised into directories • Each directory can have (but all are possible) multiple versions of same object (n-tuple or histogram) • Can duplicate and merge 2

PAW: Vectors • Vectors – – Have character identifier e. g. vec 1, 2 or 3 dim. arrays e. g. vec(10, 3) Arbitrary size and number (almost!) Create in memory, Read from disk (can filter) and Write to disk – Combine e. g. a~ = b~ * c~ – Select subrange e. g. vec(2: 5, 2) – Draw (as histogram bins), Plot (histogram points) and Fit to function 3

PAW: Functions • Functions – Plot user or built-in functions – 1, 2 or 3 dimensional – Wide range of representations (this is just a few of them!) 4

PAW: Histograms • Histograms – – Have a numeric identifier e. g. 123 1 or 2 dimensional Can associate errors with bins Read from / Write to disk / Create in memory. – Combine e. g. A = B * C – Select subrange e. g. 123(1: 20) – Wide range of plotting and fitting facilities 5

PAW: N-tuples • N-tuples – Have numeric identifiers e. g. 123 – Record a set of n numbered (1. . n) events each with m named attributes: **************************** * NTUPLE ID= 10 ENTRIES= 3354 CERN Population * **************************** * Var numb * Name * Lower * Upper * **************************** * 1 * CATEGORY * 0. 102000 E+03 * 0. 567000 E+03 * * 2 * DIVISION * 0. 100000 E+01 * 0. 130000 E+02 * * 3 * FLAG * 0. 000000 E+00 * 0. 310000 E+02 * * 4 * AGE * 0. 210000 E+02 * 0. 640000 E+02 * * 5 * SERVICE * 0. 000000 E+00 * 0. 350000 E+02 * **************************** – Two types • Row-Wise – Stored as a set of rows – Each attribute is a 4 -byte floating point • Column-Wise – Stored as a set of attributes – Attributes can be a integer, real or char – Fixed or variable length – Create in memory, I/O to disk. – Merge two or more. 6

PAW: N-tuples (cont) – Can plot functions of attributes, e. g. if have attributes x, y plot: sqrt(x**2+y**2) – Can apply cuts on points to plot e. g. : sin(x)+log(y) z>1. 0. and. z<10. 0 This is very powerful! – Can also Print rows on basis of cuts – Masks: • Used to store commonly used cuts • Saves time: only computed once • Can have 32 masks per n-tuple – Can plot in 1, 2 or 3 dimensions, e. g. z%y%x to plot Z v. Y v. X – Plotting, PAW generates histograms: • Automatically choosing suitable scales • For fine control can define histogram and the plot into that. – Scanning: Each entry in n-tuple is processed by user code 7

PAW: Programming • SIGMA – A system for vector operation e. g: sigma x=array(200, 0#2*pi) sigma s=sin(x) • Create a 200 point vector x running 0. . 2 • Create a 200 point vector s of sin(x) • COMIS – A FORTRAN interpreter – Supports a subset of FORTRAN 77 – Has access to PAW’s internal data structures – Can make calls to a wide range of CERNlib routines – Allows users to add their own code, without linking, for: • N-tuple plotting and cutting functions • Fitting and plotting functions • Non-standard operations on internal data structures. Another very powerful feature! 8

PAW: Command Interface • Commands – Have a tree structure e. g. vector/operations/vscale – Can abbreviate if unambiguous e. g. ve/op/vsc or even vsc – Case insensitive – Built-in help: summary (usage) or detailed (help) – Command line recall with arrow keys – Can be stored in files as macros. Files have the extension. kumac – Complete programming language with: • • Local variables Global variables Flow control Argument passing - so that macros can be used like subroutines • Embedded data files – Typically users develop kumac files as part of their analysis tool set 9

PAW: Graphics • Options to Control Appearance – Position on screen. Direct plots into tiled zones on the screen – Log and linear scales – Titles – Attributes of graphical elements: • • Colour Thickness Character font Line type (solid or broken) • HIGZ – Is the underlying plotting package – Has a wide range of primitives e. g. titles, axes, polylines etc. that can supplement raw plots – All graphical output can be stored in files and output as Postscript. 10