-                      ra7684e5
+                      rff7119b
 them with the desired polydispersity.
+The available kernel parameters are defined as a list, with each parameter
+defined as a sublist with the following elements:
+    *name* is the name that will be used in the call to the kernel
+    function and the name that will be displayed to the user.  Names
+    should be lower case, with words separated by underscore.  If
+    acronyms are used, the whole acronym should be upper case.
+    *units* should be one of *degrees* for angles, *Ang* for lengths,
+    *1e-6/Ang^2* for SLDs.
+    *default value* will be the initial value for  the model when it
+    is selected, or when an initial value is not otherwise specified.
+    [*lb*, *ub*] are the hard limits on the parameter value, used to limit
+    the polydispersity density function.  In the fit, the parameter limits
+    given to the fit are the limits  on the central value of the parameter.
+    If there is polydispersity, it will evaluate parameter values outside
+    the fit limits, but not outside the hard limits specified in the model.
+    If there are no limits, use +/-inf imported from numpy.
+    *type* indicates how the parameter will be used.  "volume" parameters
+    will be used in all functions.  "orientation" parameters will be used
+    in *Iqxy* and *Imagnetic*.  "magnetic* parameters will be used in
+    *Imagnetic* only.  If *type* is the empty string, the parameter will
+    be used in all of *Iq*, *Iqxy* and *Imagnetic*.
+    *description* is a short description of the parameter.  This will
+    be displayed in the parameter table and used as a tool tip for the
+    parameter value in the user interface.
 The kernel module must set variables defining the kernel meta data:
 …
     while the model parameters are being edited.
+    *parameters* is a list of parameters.  Each parameter is itself a
+    list containing *name*, *units*, *default value*,
+    [*lower bound*, *upper bound*], *type* and *description*.
+    *parameters* is the list of parameters.  Parameters in the kernel
+    functions must appear in the same order as they appear in the
+    parameters list.  Two additional parameters, *scale* and *background*
+    are added to the beginning of the parameter list.  They will show up
+    in the documentation as model parameters, but they are never sent to
+    the kernel functions.
     *source* is the list of C-99 source files that must be joined to
 …
     *VR* is a python function defining the volume ratio.  If it is not
     present, the volume ratio is 1.
+An *info* dictionary is constructed from the kernel meta data and
+returned to the caller.  It includes the additional fields
+The model evaluator, function call sequence consists of q inputs and the return vector,
+followed by the loop value/weight vector, followed by the values for
+the non-polydisperse parameters, followed by the lengths of the
+polydispersity loops.  To construct the call for 1D models, the
+categories *fixed-1d* and *pd-1d* list the names of the parameters
+of the non-polydisperse and the polydisperse parameters respectively.
+Similarly, *fixed-2d* and *pd-2d* provide parameter names for 2D models.
+The *pd-rel* category is a set of those parameters which give
+polydispersitiy as a portion of the value (so a 10% length dispersity
+would use a polydispersity value of 0.1) rather than absolute
+dispersity such as an angle plus or minus 15 degrees.
+The *volume* category lists the volume parameters in order for calls
+to volume within the kernel (used for volume normalization) and for
+calls to ER and VR for effective radius and volume ratio respectively.
+The *orientation* and *magnetic* categories list the orientation and
+magnetic parameters.  These are used by the sasview interface.  The
+blank category is for parameters such as scale which don't have any
+other marking.
 The doc string at the start of the kernel module will be used to
 …
 file names and extensions.
-Parameters are defined as follows:
-    *name* is the name that will be used in the call to the kernel
-    function and the name that will be displayed to the user.  Names
-    should be lower case, with words separated by underscore.  If
-    acronyms are used, the whole acronym should be upper case.
-    *units* should be one of *degrees* for angles, *Ang* for lengths,
-    *1e-6/Ang^2* for SLDs.
-    *default value* will be the initial value for  the model when it
-    is selected, or when an initial value is not otherwise specified.
-    *limits* are the valid bounds of the parameter, used to limit the
-    polydispersity density function.   In the fit, the parameter limits
-    given to the fit are the limits  on the central value of the parameter.
-    If there is polydispersity, it will evaluate parameter values outside
-    the fit limits, but not outside the hard limits specified in the model.
-    If there are no limits, use +/-inf imported from numpy.
-    *type* indicates how the parameter will be used.  "volume" parameters
-    will be used in all functions.  "orientation" parameters will be used
-    in *Iqxy* and *Imagnetic*.  "magnetic* parameters will be used in
-    *Imagnetic* only.  If *type* is none, the parameter will be used in
-    all of *Iq*, *Iqxy* and *Imagnetic*.  This will probably be a
-    is the empty string if the parameter is used in all model calculations,
-    it is "volu
-    *description* is a short description of the parameter.  This will
-    be displayed in the parameter table and used as a tool tip for the
-    parameter value in the user interface.
 The function :func:`make` loads the metadata from the module and returns
+the kernel source.  The function :func:`doc` extracts
+the kernel source.  The function :func:`doc` extracts the doc string
+and adds the parameter table to the top.  The function :func:`sources`
+returns a list of files required by the model.
 """
 # TODO: identify model files which have changed since loading and reload them.
 __all__ = ["make, doc"]
+__all__ = ["make, doc", "sources"]
 import os.path
 …
+    ]
+# Minimum width for a default value (this is shorter than the column header
+# width, so will be ignored).
 PARTABLE_VALUE_WIDTH = 10
 # Header included before every kernel.
+# This makes sure that the appropriate math constants are defined, and
+# This makes sure that the appropriate math constants are defined, and does
+# whatever is required to make the kernel compile as pure C rather than
+# as an OpenCL kernel.
 KERNEL_HEADER = """\
 // GENERATED CODE --- DO NOT EDIT ---
 …
+    }
 # Generic kernel template for opencl/openmp.
+# Generic kernel template for the polydispersity loop.
 # This defines the opencl kernel that is available to the host.  The same
 # structure is used for Iq and Iqxy kernels, so extra flexibility is needed
 …
 def kernel_name(info, is_2D):
+    """
+    Name of the exported kernel symbol.
+    """
     return info['name'] + "_" + ("Iqxy" if is_2D else "Iq")
 …
 def make_partable(info):
+    """
+    Generate the parameter table to include in the sphinx documentation.
+    """
     pars = info['parameters']
     column_widths = [
 …
     raise ValueError("%r not found in %s"%(filename, search_path))
+def make_model(search_path, info):
+def sources(info):
+    """
+    Return a list of the sources file paths for the module.
+    """
+    from os.path import abspath, dirname, join as joinpath
+    search_path = [ dirname(info['filename']),
+                    abspath(joinpath(dirname(__file__),'models')) ]
+    return [_search(search_path) for f in info['source']]
+def make_model(info):
+    """
+    Generate the code for the kernel defined by info, using source files
+    found in the given search path.
+    """
     kernel_Iq = make_kernel(info, is_2D=False)
     kernel_Iqxy = make_kernel(info, is_2D=True)
     source = [open(_search(search_path, f)).read() for f in info['source']]
+    source = [open(f).read() for f in sources(info)]
     kernel = "\n\n".join([KERNEL_HEADER]+source+[kernel_Iq, kernel_Iqxy])
     return kernel
 …
     Returns a dictionary of categories.
-    The function call sequence consists of q inputs and the return vector,
-    followed by the loop value/weight vector, followed by the values for
-    the non-polydisperse parameters, followed by the lengths of the
-    polydispersity loops.  To construct the call for 1D models, the
-    categories *fixed-1d* and *pd-1d* list the names of the parameters
-    of the non-polydisperse and the polydisperse parameters respectively.
-    Similarly, *fixed-2d* and *pd-2d* provide parameter names for 2D models.
-    The *pd-rel* category is a set of those parameters which give
-    polydispersitiy as a portion of the value (so a 10% length dispersity
-    would use a polydispersity value of 0.1) rather than absolute
-    dispersity such as an angle plus or minus 15 degrees.
-    The *volume* category lists the volume parameters in order for calls
-    to volume within the kernel (used for volume normalization) and for
-    calls to ER and VR for effective radius and volume ratio respectively.
-    The *orientation* and *magnetic* categories list the orientation and
-    magnetic parameters.  These are used by the sasview interface.  The
-    blank category is for parameters such as scale which don't have any
-    other marking.
     """
     partype = {
 …
     """
     # TODO: allow Iq and Iqxy to be defined in python
     from os.path import abspath, dirname, join as joinpath
+    from os.path import abspath
     #print kernelfile
     info = dict(
 …
     info['partype'] = categorize_parameters(info['parameters'])
+    search_path = [ dirname(info['filename']),
+                    abspath(joinpath(dirname(__file__),'models')) ]
+    source = make_model(search_path, info)
+    source = make_model(info)
     return source, info
 …
                  doc=kernel_module.__doc__)
     return DOC_HEADER%subst
+# Compiler platform details
+if sys.platform == 'darwin':
+    COMPILE = "gcc-mp-4.7 -shared -fPIC -std=c99 -fopenmp -O2 -Wall %s -o %s -lm -lgomp"
+elif os.name == 'nt':
+    COMPILE = "gcc -shared -fPIC -std=c99 -fopenmp -O2 -Wall %s -o %s -lm"
+else:
+    COMPILE = "cc -shared -fPIC -std=c99 -fopenmp -O2 -Wall %s -o %s -lm"
+DLL_PATH = "/tmp"
+def dll_path(info):
+    """
+    Path to the compiled model defined by *info*.
+    """
+    from os.path import join as joinpath, split as splitpath, splitext
+    basename = splitext(splitpath(info['filename'])[1])[0]
+    return joinpath(DLL_PATH, basename+'.so')
+def dll_model(kernel_module, dtype=None):
+    """
+    Load the compiled model defined by *kernel_module*.
+    Recompile if any files are newer than the model file.
+    *dtype* is ignored.  Compiled files are always double.
+    The DLL is not loaded until the kernel is called so models an
+    be defined without using too many resources.
+    """
+    import tempfile
+    from sasmodels import dll
+    source, info = make(kernel_module)
+    source_files = sources(info) + [info['filename']]
+    newest = max(os.path.getmtime(f) for f in source_files)
+    dllpath = dll_path(info)
+    if not os.path.exists(dllpath) or os.path.getmtime(dllpath)<newest:
+        # Replace with a proper temp file
+        srcfile = tempfile.mkstemp(suffix=".c",prefix="sas_"+info['name'])
+        open(srcfile, 'w').write(source)
+        os.system(COMPILE%(srcfile, dllpath))
+        ## comment the following to keep the generated c file
+        os.unlink(srcfile)
+    return dll.DllModel(dllpath, info)
+def opencl_model(kernel_module, dtype="single"):
+    """
+    Load the OpenCL model defined by *kernel_module*.
+    Access to the OpenCL device is delayed until the kernel is called
+    so models can be defined without using too many resources.
+    """
+    from sasmodels import gpu
+    source, info = make(kernel_module)
+    ## for debugging, save source to a .cl file, edit it, and reload as model
+    #open(modelname+'.cl','w').write(source)
+    #source = open(modelname+'.cl','r').read()
+    return gpu.GpuModel(source, info, dtype)

Note: See TracChangeset for help on using the changeset viewer.

SasView

Changeset ff7119b in sasmodels for sasmodels/gen.py

Legend:

sasmodels/gen.py

Download in other formats: