source: inundation/pyvolution/domain.py @ 1862

"""Class Domain - 2D triangular domains for finite-volume computations of
   the shallow water wave equation


   Copyright 2004
   Ole Nielsen, Stephen Roberts, Duncan Gray, Christopher Zoppou
   Geoscience Australia
"""

from mesh import Mesh
from generic_boundary_conditions import *
import types

class Domain(Mesh):

    def __init__(self, coordinates, vertices, boundary = None,
                 conserved_quantities = None, other_quantities = None,
                 tagged_elements = None, geo_reference = None,
                 use_inscribed_circle=False):

        Mesh.__init__(self, coordinates, vertices, boundary,
                      tagged_elements, geo_reference, use_inscribed_circle)

        from Numeric import zeros, Float, Int
        from quantity import Quantity, Conserved_quantity

        #List of quantity names entering
        #the conservation equations
        #(Must be a subset of quantities)
        if conserved_quantities is None:
            self.conserved_quantities = []
        else:
            self.conserved_quantities = conserved_quantities

        if other_quantities is None:
            self.other_quantities = []
        else:
            self.other_quantities = other_quantities


        #Build dictionary of Quantity instances keyed by quantity names
        self.quantities = {}

        #FIXME: remove later - maybe OK, though....
        for name in self.conserved_quantities:
            self.quantities[name] = Conserved_quantity(self)
        for name in self.other_quantities:
            self.quantities[name] = Quantity(self)

        #Create an empty list for explicit forcing terms
        self.forcing_terms = []


        #Defaults
        from config import max_smallsteps, beta_w, beta_h, epsilon, CFL
        self.beta_w = beta_w
        self.beta_h = beta_h
        self.epsilon = epsilon

        #FIXME: Maybe have separate orders for h-limiter and w-limiter?
        #Or maybe get rid of order altogether and use beta_w and beta_h
        self.default_order = 1
        self.order = self.default_order
        self.smallsteps = 0
        self.max_smallsteps = max_smallsteps
        self.number_of_steps = 0
        self.number_of_first_order_steps = 0
        self.CFL = CFL

        #Model time
        self.time = 0.0
        self.finaltime = None
        self.min_timestep = self.max_timestep = 0.0
        self.starttime = 0 #Physical starttime if any (0 is 1 Jan 1970 00:00:00)

        ######OBSOLETE
        #Origin in UTM coordinates
        #FIXME: This should be set if read by a msh file
        #self.zone = zone
        #self.xllcorner = xllcorner
        #self.yllcorner = yllcorner


        #Checkpointing and storage
        from config import default_datadir
        self.datadir = default_datadir
        self.filename = 'domain'
        self.checkpoint = False

        #MH310505 To avoid calculating the flux across each edge twice, keep an integer (boolean) array,
        #to be used during the flux calculation
        N=self.number_of_elements
        self.already_computed_flux = zeros((N, 3), Int)

    #Public interface to Domain
    def get_conserved_quantities(self, vol_id, vertex=None, edge=None):
        """Get conserved quantities at volume vol_id

        If vertex is specified use it as index for vertex values
        If edge is specified use it as index for edge values
        If neither are specified use centroid values
        If both are specified an exeception is raised

        Return value: Vector of length == number_of_conserved quantities

        """

        from Numeric import zeros, Float

        if not (vertex is None or edge is None):
            msg = 'Values for both vertex and edge was specified.'
            msg += 'Only one (or none) is allowed.'
            raise msg

        q = zeros( len(self.conserved_quantities), Float)

        for i, name in enumerate(self.conserved_quantities):
            Q = self.quantities[name]
            if vertex is not None:
                q[i] = Q.vertex_values[vol_id, vertex]
            elif edge is not None:
                q[i] = Q.edge_values[vol_id, edge]
            else:
                q[i] = Q.centroid_values[vol_id]

        return q


    def set_quantity_vertices_dict(self, quantity_dict):
        """Set values for named quantities.
        The index is the quantity

        name: Name of quantity
        X: Compatible list, Numeric array, const or function (see below)

        The values will be stored in elements following their
        internal ordering.

        """
        for key in quantity_dict.keys():
            self.set_quantity(key, quantity_dict[key], location='vertices')


    def set_quantity(self, name, *args, **kwargs):
                     #X,
                     #location = 'vertices',
                     #indices = None):
        """Set values for named quantity

        name: Name of quantity
        X: Compatible list, Numeric array, const or function (see below)
        location: Where values are to be stored.
                  Permissible options are: vertices, edges, centroids

        In case of location == 'centroids' the dimension values must
        be a list of a Numerical array of length N, N being the number
        of elements. Otherwise it must be of dimension Nx3.

        Indices is the set of element ids that the operation applies to

        The values will be stored in elements following their
        internal ordering.

        #FIXME (Ole): I suggest the following interface
        set_quantity(name, X, location, region)
        where
            name: Name of quantity
            X:
              -Compatible list,
              -Numeric array,
              -const or function (see below)
              -another quantity Q or an expression of the form
              a*Q+b, where a is a scalar or a compatible array or a quantity
              Q is a quantity
              b is either a scalar, a quantity or a compatible array
              pts file
            location: Where values are to be stored.
                      Permissible options are: vertices, edges, centroid
            region: Identify subset of triangles. Permissible values are
                    - tag name (refers to tagged region)
                    - indices (refers to specific triangles)
                    - polygon (identifies region)

            This should work for all values of X


        """

        #from quantity import Quantity, Conserved_quantity

        #Create appropriate quantity object
        ##if name in self.conserved_quantities:
        ##    self.quantities[name] = Conserved_quantity(self)
        ##else:
        ##    self.quantities[name] = Quantity(self)

        #Set value
        #FIXME: use **kwargs
        #self.quantities[name].set_values(X,
        #                                 location = location,
        #                                 indices = indices)
        self.quantities[name].set_values(*args, **kwargs)


    def get_quantity(self, name, location='vertices', indices = None):
        """Get values for named quantity

        name: Name of quantity

        In case of location == 'centroids' the dimension values must
        be a list of a Numerical array of length N, N being the number
        of elements. Otherwise it must be of dimension Nx3.

        Indices is the set of element ids that the operation applies to.

        The values will be stored in elements following their
        internal ordering.
        """

        return self.quantities[name].get_values( location, indices = indices)


    def set_boundary(self, boundary_map):
        """Associate boundary objects with tagged boundary segments.

        Input boundary_map is a dictionary of boundary objects keyed
        by symbolic tags to matched against tags in the internal dictionary
        self.boundary.

        As result one pointer to a boundary object is stored for each vertex
        in the list self.boundary_objects.
        More entries may point to the same boundary object

        Schematically the mapping is from two dictionaries to one list
        where the index is used as pointer to the boundary_values arrays
        within each quantity.

        self.boundary:          (vol_id, edge_id): tag
        boundary_map (input):   tag: boundary_object
        ----------------------------------------------
        self.boundary_objects:  ((vol_id, edge_id), boundary_object)


        Pre-condition:
          self.boundary has been built.

        Post-condition:
          self.boundary_objects is built

        If a tag from the domain doesn't appear in the input dictionary an
        exception is raised.
        However, if a tag is not used to the domain, no error is thrown.
        FIXME: This would lead to implementation of a
        default boundary condition

        Note: If a segment is listed in the boundary dictionary and if it is
        not None, it *will* become a boundary -
        even if there is a neighbouring triangle.
        This would be the case for internal boundaries

        Boundary objects that are None will be skipped.

        FIXME: If set_boundary is called multiple times and if Boundary
        object is changed into None, the neighbour structure will not be
        restored!!!


        """

        self.boundary_objects = []
        self.boundary_map = boundary_map  #Store for use with eg. boundary_stats.

        #FIXME: Try to remove the sorting and fix test_mesh.py
        x = self.boundary.keys()
        x.sort()

        #Loop through edges that lie on the boundary and associate them with
        #callable boundary objects depending on their tags
        for k, (vol_id, edge_id) in enumerate(x):
            tag = self.boundary[ (vol_id, edge_id) ]

            if boundary_map.has_key(tag):
                B = boundary_map[tag]  #Get callable boundary object

                if B is not None:
                    self.boundary_objects.append( ((vol_id, edge_id), B) )
                    self.neighbours[vol_id, edge_id] = -len(self.boundary_objects)
                else:
                    pass
                    #FIXME: Check and perhaps fix neighbour structure


            else:
                msg = 'ERROR (domain.py): Tag "%s" has not been ' %tag
                msg += 'bound to a boundary object.\n'
                msg += 'All boundary tags defined in domain must appear '
                msg += 'in the supplied dictionary.\n'
                msg += 'The tags are: %s' %self.get_boundary_tags()
                raise msg


    def set_region(self, functions):
        # The order of functions in the list is used.
        if type(functions) not in [types.ListType,types.TupleType]:
            functions = [functions]
        for function in functions:
            for tag in self.tagged_elements.keys():
                function(tag, self.tagged_elements[tag], self)

                #Do we need to do this sort of thing?
                #self = function(tag, self.tagged_elements[tag], self)

    #MISC
    def check_integrity(self):
        Mesh.check_integrity(self)

        for quantity in self.conserved_quantities:
            msg = 'Conserved quantities must be a subset of all quantities'
            assert quantity in self.quantities, msg

        ##assert hasattr(self, 'boundary_objects')

    def write_time(self):
        print self.timestepping_statistics()

        #Old version
        #if self.min_timestep == self.max_timestep:
        #    print 'Time = %.4f, delta t = %.8f, steps=%d (%d)'\
        #          %(self.time, self.min_timestep, self.number_of_steps,
        #            self.number_of_first_order_steps)
        #elif self.min_timestep > self.max_timestep:
        #    print 'Time = %.4f, steps=%d (%d)'\
        #          %(self.time, self.number_of_steps,
        #            self.number_of_first_order_steps)
        #else:
        #    print 'Time = %.4f, delta t in [%.8f, %.8f], steps=%d (%d)'\
        #          %(self.time, self.min_timestep,
        #            self.max_timestep, self.number_of_steps,
        #            self.number_of_first_order_steps)        
        
    def timestepping_statistics(self):
        """Return string with time stepping statistics for printing or logging
        """

        msg = ''
        if self.min_timestep == self.max_timestep:
            msg += 'Time = %.4f, delta t = %.8f, steps=%d (%d)'\
                   %(self.time, self.min_timestep, self.number_of_steps,
                     self.number_of_first_order_steps)
        elif self.min_timestep > self.max_timestep:
            msg += 'Time = %.4f, steps=%d (%d)'\
                   %(self.time, self.number_of_steps,
                     self.number_of_first_order_steps)
        else:
            msg += 'Time = %.4f, delta t in [%.8f, %.8f], steps=%d (%d)'\
                   %(self.time, self.min_timestep,
                     self.max_timestep, self.number_of_steps,
                     self.number_of_first_order_steps)
            
        return msg    

            
    def write_boundary_statistics(self, quantities = None, tags = None):
        print self.boundary_statistics(quantities, tags)
        
    def boundary_statistics(self, quantities = None, tags = None):
        """Output statistics about boundary forcing at each timestep

        Example
        Tag 'wall':
            stage in [2, 5.5]
            xmomentum in []
            ymomentum in []
        Tag 'ocean'

        quantities and tags can be either None, a string or a list of strings

        If quantities are specified only report on those. Otherwise take all conserved quantities.
        If tags are specified only report on those, otherwise take all tags.

        """

        #Input checks
        import types, string
        
        if quantities is None:
            quantities = self.conserved_quantities
        elif type(quantities) == types.StringType:
            quantities = [quantities] #Turn it into a list

        msg = 'Keyword argument quantities must be either None, '
        msg += 'string or list. I got %s' %str(quantities)   
        assert type(quantities) == types.ListType, msg
            

        if tags is None:
            tags = self.get_boundary_tags()
        elif type(tags) == types.StringType:
            tags = [tags] #Turn it into a list

        msg = 'Keyword argument tags must be either None, '
        msg += 'string or list. I got %s' %str(tags)   
        assert type(tags) == types.ListType, msg            

        #Determine width of longest quantity name (for cosmetic purposes)
        maxwidth = 0
        for name in quantities:
            w = len(name) 
            if w > maxwidth:
                maxwidth = w
            
        #Output stats
        msg = 'Boundary values at time %.4f:\n' %self.time
        for tag in tags:
            msg += '    %s:\n' %tag
          
            for name in quantities:
                q = self.quantities[name]
        
                #Find range of boundary values for tag and q 
                maxval = minval = None
                for i, ((vol_id, edge_id), B) in\
                        enumerate(self.boundary_objects):
                    if self.boundary[(vol_id, edge_id)] == tag:
                        v = q.boundary_values[i]
                        if minval is None or v < minval: minval = v
                        if maxval is None or v > maxval: maxval = v
         
                if minval is None or maxval is None:
                    msg += '        Sorry no information available about' +\
                           ' tag %s and quantity %s\n' %(tag, name)
                else:
                    msg += '        %s in [%12.8f, %12.8f]\n'\
                           %(string.ljust(name, maxwidth), minval, maxval)


        return msg

    
    def get_name(self):
        return self.filename

    def set_name(self, name):
        self.filename = name

    def get_datadir(self):
        return self.datadir

    def set_datadir(self, name):
        self.datadir = name



    #def set_defaults(self):
    #    """Set default values for uninitialised quantities.
    #    Should be overridden or specialised by specific modules
    #    """#
    #
    #    for name in self.conserved_quantities + self.other_quantities:
    #        self.set_quantity(name, 0.0)


    ###########################
    #Main components of evolve

    def evolve(self, yieldstep = None, finaltime = None,
               skip_initial_step = False):
        """Evolve model from time=0.0 to finaltime yielding results
        every yieldstep.

        Internally, smaller timesteps may be taken.

        Evolve is implemented as a generator and is to be called as such, e.g.

        for t in domain.evolve(timestep, yieldstep, finaltime):
            <Do something with domain and t>

        """

        from config import min_timestep, max_timestep, epsilon

        #FIXME: Maybe lump into a larger check prior to evolving
        msg = 'Boundary tags must be bound to boundary objects before evolving system, '
        msg += 'e.g. using the method set_boundary.\n'
        msg += 'This system has the boundary tags %s ' %self.get_boundary_tags()
        assert hasattr(self, 'boundary_objects'), msg

        ##self.set_defaults()

        if yieldstep is None:
            yieldstep = max_timestep
        else:
            yieldstep = float(yieldstep)

        self.order = self.default_order


        self.yieldtime = 0.0 #Time between 'yields'

        #Initialise interval of timestep sizes (for reporting only)
        self.min_timestep = max_timestep
        self.max_timestep = min_timestep
        self.finaltime = finaltime
        self.number_of_steps = 0
        self.number_of_first_order_steps = 0

        #update ghosts
        self.update_ghosts()

        #Initial update of vertex and edge values
        self.distribute_to_vertices_and_edges()

        #Initial update boundary values
        self.update_boundary()

        #Or maybe restore from latest checkpoint
        if self.checkpoint is True:
            self.goto_latest_checkpoint()

        if skip_initial_step is False:    
            yield(self.time)  #Yield initial values

        while True:

            #Compute fluxes across each element edge
            self.compute_fluxes()

            #Update timestep to fit yieldstep and finaltime
            self.update_timestep(yieldstep, finaltime)

            #Update conserved quantities
            self.update_conserved_quantities()

            #update ghosts
            self.update_ghosts()

            #Update vertex and edge values
            self.distribute_to_vertices_and_edges()

            #Update boundary values
            self.update_boundary()

            #Update time
            self.time += self.timestep
            self.yieldtime += self.timestep
            self.number_of_steps += 1
            if self.order == 1:
                self.number_of_first_order_steps += 1

            #Yield results
            if finaltime is not None and abs(self.time - finaltime) < epsilon:

                #FIXME: There is a rare situation where the
                #final time step is stored twice. Can we make a test?

                # Yield final time and stop
                yield(self.time)
                break


            if abs(self.yieldtime - yieldstep) < epsilon:
                # Yield (intermediate) time and allow inspection of domain

                if self.checkpoint is True:
                    self.store_checkpoint()
                    self.delete_old_checkpoints()

                #Pass control on to outer loop for more specific actions
                yield(self.time)

                # Reinitialise
                self.yieldtime = 0.0
                self.min_timestep = max_timestep
                self.max_timestep = min_timestep
                self.number_of_steps = 0
                self.number_of_first_order_steps = 0


    def evolve_to_end(self, finaltime = 1.0):
        """Iterate evolve all the way to the end
        """

        for _ in self.evolve(yieldstep=None, finaltime=finaltime):
            pass



    def update_boundary(self):
        """Go through list of boundary objects and update boundary values
        for all conserved quantities on boundary.
        """

        #FIXME: Update only those that change (if that can be worked out)
        for i, ((vol_id, edge_id), B) in enumerate(self.boundary_objects):
            q = B.evaluate(vol_id, edge_id)

            for j, name in enumerate(self.conserved_quantities):
                Q = self.quantities[name]
                Q.boundary_values[i] = q[j]


    def compute_fluxes(self):
        msg = 'Method compute_fluxes must be overridden by Domain subclass'
        raise msg


    def update_timestep(self, yieldstep, finaltime):

        from config import min_timestep

        # self.timestep is calculated from speed of characteristics
        # Apply CFL condition here
        timestep = self.CFL*self.timestep

        #Record maximal and minimal values of timestep for reporting
        self.max_timestep = max(timestep, self.max_timestep)
        self.min_timestep = min(timestep, self.min_timestep)

        #Protect against degenerate time steps
        if timestep < min_timestep:

            #Number of consecutive small steps taken b4 taking action
            self.smallsteps += 1

            if self.smallsteps > self.max_smallsteps:
                self.smallsteps = 0 #Reset

                if self.order == 1:
                    msg = 'WARNING: Too small timestep %.16f reached '\
                          %timestep
                    msg += 'even after %d steps of 1 order scheme'\
                           %self.max_smallsteps
                    print msg
                    timestep = min_timestep  #Try enforcing min_step

                    #raise msg
                else:
                    #Try to overcome situation by switching to 1 order
                    self.order = 1

        else:
            self.smallsteps = 0
            if self.order == 1 and self.default_order == 2:
                self.order = 2


        #Ensure that final time is not exceeded
        if finaltime is not None and self.time + timestep > finaltime:
            timestep = finaltime-self.time

        #Ensure that model time is aligned with yieldsteps
        if self.yieldtime + timestep > yieldstep:
            timestep = yieldstep-self.yieldtime

        self.timestep = timestep



    def compute_forcing_terms(self):
        """If there are any forcing functions driving the system
        they should be defined in Domain subclass and appended to
        the list self.forcing_terms
        """

        for f in self.forcing_terms:
            f(self)



    def update_conserved_quantities(self):
        """Update vectors of conserved quantities using previously
        computed fluxes specified forcing functions.
        """

        from Numeric import ones, sum, equal, Float

        N = self.number_of_elements
        d = len(self.conserved_quantities)

        timestep = self.timestep

        #Compute forcing terms
        self.compute_forcing_terms()

        #Update conserved_quantities
        for name in self.conserved_quantities:
            Q = self.quantities[name]
            Q.update(timestep)

            #Clean up
            #Note that Q.explicit_update is reset by compute_fluxes

            #MH090605 commented out the following since semi_implicit_update is now re-initialized
            #at the end of the _update function in quantity_ext.c (This is called by the
            #preceeding Q.update(timestep) statement above).
            #For run_profile.py with N=128, the time of update_conserved_quantities is cut from 14.00 secs
            #to 8.35 secs

            #Q.semi_implicit_update[:] = 0.0

    def update_ghosts(self):
        pass

    def distribute_to_vertices_and_edges(self):
        """Extrapolate conserved quantities from centroid to
        vertices and edge-midpoints for each volume

        Default implementation is straight first order,
        i.e. constant values throughout each element and
        no reference to non-conserved quantities.
        """

        for name in self.conserved_quantities:
            Q = self.quantities[name]
            if self.order == 1:
                Q.extrapolate_first_order()
            elif self.order == 2:
                Q.extrapolate_second_order()
                Q.limit()
            else:
                raise 'Unknown order'
            Q.interpolate_from_vertices_to_edges()



##############################################
#Initialise module

#Optimisation with psyco
from config import use_psyco
if use_psyco:
    try:
        import psyco
    except:
        import os
        if os.name == 'posix' and os.uname()[4] == 'x86_64':
            pass
            #Psyco isn't supported on 64 bit systems, but it doesn't matter
        else:
            msg = 'WARNING: psyco (speedup) could not import'+\
                  ', you may want to consider installing it'
            print msg
    else:
        psyco.bind(Domain.update_boundary)
        #psyco.bind(Domain.update_timestep)     #Not worth it
        psyco.bind(Domain.update_conserved_quantities)
        psyco.bind(Domain.distribute_to_vertices_and_edges)


if __name__ == "__main__":
    pass