source: anuga_work/development/alpha_validation/find_alpha.py @ 6732

from anuga.abstract_2d_finite_volumes import domain
from anuga.shallow_water import Domain
from anuga.geospatial_data.geospatial_data import Geospatial_data
from anuga.coordinate_transforms import Geo_reference
from anuga.pmesh.mesh_interface import create_mesh_from_regions
from anuga.shallow_water import Reflective_boundary
from anuga.utilities.numerical_tools import cov
from Numeric import array, resize,shape,Float,zeros,take,argsort,argmin
from pylab import plot, ion, hold,savefig,semilogx,plotting,loglog


poly_topo = [[664000,7752000],[664000,7754000],
             [666000,7754000],[666000,7752000]]
mesh_dir_name = 'temp.msh'
                  
create_mesh_from_regions(poly_topo,
                         boundary_tags={'back': [2], 'side': [1,3],
                                        'ocean': [0]},
                         maximum_triangle_area=1000,
                         filename=mesh_dir_name,
                         use_cache=True,
                         verbose=True)


#topo_dir_name = 'pt_hedland_export.txt'
topo_dir_name = 'pt_hedland_small.txt'
#sample = 'pt_hedland_small.txt'
remainder ='remainder.txt'
#split topo data
G = Geospatial_data(file_name = topo_dir_name)
print 'start split'





G_sample,G_loss= G.split(0.01, True)

#G_sample.export_points_file(sample)

G_small, G_other = G_sample.split(0.1,True)

#import sys
#sys.exit()

alphas = [0.000001, 0.00001, 0.0001, 0.001, 0.01, 0.1, 1.0, 10.0, 100.0,1000.0,10000.0]
#alphas = [0.001,1,100]
domains = {}
#domains = []

print 'Setup computational domain'
for alpha in alphas:
    
    #think about making domain absolute when it is created
    domain = Domain(mesh_dir_name, use_cache=False, verbose=True)
    print domain.statistics()
    domain.set_quantity('elevation', 
                    geospatial_data = G_other,
                    use_cache = True,
                    verbose = True,
                    alpha = alpha)
    domains[alpha]=domain

points=G_small.get_data_points()
data=array([],typecode=Float)
data=resize(data,(len(points),3+len(alphas)))
#gets relative point from sample

data[:,0]=points[:,0]
data[:,1]=points[:,1]
print'geo',domain.geo_reference,points[0:10],points[0:10,0]
elevation_sample=G_small.get_attributes(attribute_name='elevation')

data[:,2]=elevation_sample

print'data',data[0:10],len(alphas)
normal_cov=array(zeros([len(alphas),2]),typecode=Float)
i=0

for domain in domains:
#    print'domain',domain
    points_geo=domains[domain].geo_reference.change_points_geo_ref(points)
    

    print'poin',points_geo[0:10],elevation_sample[0:10]
    print'quantities',domains[domain].get_quantity_names()
    elevation_predicted=domains[domain].quantities['elevation'].get_values(interpolation_points=points_geo)
    print'elevation_predicted',elevation_predicted[0:10],len(points),len(elevation_predicted)
    
    data[:,i+3]=elevation_predicted
    sample_cov= cov(elevation_sample)

    ele_cov= cov(elevation_sample-elevation_predicted)
    normal_cov[i,:]= [domain,ele_cov/sample_cov]

    print'cov',alpha,'= ',normal_cov, ele_cov, sample_cov
    i+=1
    
    
print'data',data[100:200]
print'normal cov',normal_cov
#to sort array by column
normal_cov0=normal_cov[:,0]
normal_cov_new=take(normal_cov,argsort(normal_cov0))
semilogx(normal_cov_new[:,0],normal_cov_new[:,1])
#loglog(normal_cov_new[:,0],normal_cov_new[:,1])
print 'normal_cov_new',normal_cov_new
savefig("alphas",dpi=300)

print argmin(normal_cov_new,axis=1)
print min(normal_cov_new[:,1]) , normal_cov_new[(argmin(normal_cov_new,axis=0))[1],0]