def read_history( History_filename, nZones = 1):
""" reads a history file
returns an ordered bunch with the history file headers for keys
and a list of each header's floats for values.
if header is an optimization objective, its name is mapped to
the optimization name.
Iter and Time(min) headers are mapped to ITERATION and TIME
respectively.
"""
# read plot file
plot_data = read_plot( History_filename )
# initialize history data dictionary
history_data = ordered_bunch()
# map header names
for key in plot_data.keys():
var = key
for field in historyOutFields:
if key == historyOutFields[field]['HEADER'] and nZones == 1:
var = field
if key.split('[')[0] == historyOutFields[field]['HEADER'] and nZones > 1:
var = field + '[' + key.split('[')[1]
history_data[var] = plot_data[key]
return history_data
def read_history(History_filename):
""" reads a history file
returns an ordered bunch with the history file headers for keys
and a list of each header's floats for values.
if header is an optimization objective, its name is mapped to
the optimization name.
Iter and Time(min) headers are mapped to ITERATION and TIME
respectively.
"""
# read plot file
plot_data = read_plot(History_filename)
# initialize history data dictionary
history_data = ordered_bunch()
# header name to config file name map
map_dict = get_headerMap()
# map header names
for key in plot_data.keys():
if map_dict.has_key(key):
var = map_dict[key]
else:
var = key
history_data[var] = plot_data[key]
return history_data
def read_history( History_filename ):
""" reads a history file
returns an ordered bunch with the history file headers for keys
and a list of each header's floats for values.
if header is an optimization objective, its name is mapped to
the optimization name.
Iter and Time(min) headers are mapped to ITERATION and TIME
respectively.
"""
# read plot file
plot_data = read_plot( History_filename )
# initialize history data dictionary
history_data = ordered_bunch()
# header name to config file name map
map_dict = get_headerMap()
# map header names
for key in plot_data.keys():
if map_dict.has_key(key):
var = map_dict[key]
else:
var = key
history_data[var] = plot_data[key]
return history_data
def read_aerodynamics( History_filename , special_cases=[] ):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'UNSTEADY_SIMULATION', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename)
# list of functions to pull
func_names = optnames_aero
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in func_names:
if history_data.has_key(this_objfun):
Func_Values[this_objfun] = history_data[this_objfun]
# for unsteady cases, average time-accurate objective function values
if 'UNSTEADY_SIMULATION' in special_cases:
for key,value in Func_Values.iteritems():
Func_Values[key] = sum(value)/len(value)
# otherwise, keep only last value
else:
for key,value in Func_Values.iteritems():
Func_Values[key] = value[-1]
return Func_Values
def read_aerodynamics(History_filename,
nZones=1,
special_cases=[],
final_avg=0,
wnd_fct='SQUARE'):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'TIME_MARCHING', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename, nZones)
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in historyOutFields:
if nZones == 1:
if this_objfun in history_data:
if historyOutFields[this_objfun][
'TYPE'] == 'COEFFICIENT' or historyOutFields[
this_objfun]['TYPE'] == 'D_COEFFICIENT':
Func_Values[this_objfun] = history_data[this_objfun]
else:
for iZone in range(nZones):
if this_objfun + '[' + str(iZone) + ']' in history_data:
if historyOutFields[this_objfun][
'TYPE'] == 'COEFFICIENT' or historyOutFields[
this_objfun]['TYPE'] == 'D_COEFFICIENT':
Func_Values[this_objfun + '[' + str(iZone) +
']'] = history_data[this_objfun + '[' +
str(iZone) + ']']
if 'TIME_MARCHING' in special_cases:
# for unsteady cases, average time-accurate objective function values
for key, value in Func_Values.items():
if historyOutFields[key]['TYPE'] == 'COEFFICIENT':
if not history_data.get('TAVG_' + key):
raise KeyError('Key ' +
historyOutFields['TAVG_' + key]['HEADER'] +
' was not found in history output.')
Func_Values[key] = history_data['TAVG_' + key][-1]
elif historyOutFields[key]['TYPE'] == 'D_COEFFICIENT':
if not history_data.get('TAVG_' + key):
raise KeyError('Key ' +
historyOutFields['TAVG_' + key]['HEADER'] +
' was not found in history output.')
Func_Values[key] = history_data['TAVG_' + key][-1]
else:
# in steady cases take only last value.
for key, value in Func_Values.iteritems():
if not history_data.get(key):
raise KeyError('Key ' + historyOutFields[key]['HEADER'] +
' was not found in history output.')
Func_Values[key] = value[-1]
return Func_Values
def read_aerodynamics( History_filename , special_cases=[] ):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'UNSTEADY_SIMULATION', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename)
# list of functions to pull
func_names = optnames_aero
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in func_names:
if history_data.has_key(this_objfun):
Func_Values[this_objfun] = history_data[this_objfun]
# for unsteady cases, average time-accurate objective function values
if 'UNSTEADY_SIMULATION' in special_cases:
for key,value in Func_Values.iteritems():
Func_Values[key] = sum(value)/len(value)
# otherwise, keep only last value
else:
for key,value in Func_Values.iteritems():
Func_Values[key] = value[-1]
return Func_Values
def read_aerodynamics(History_filename,
nZones=1,
special_cases=[],
final_avg=0):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'UNSTEADY_SIMULATION', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename, nZones)
# list of functions to pull
func_names = optnames_aero + grad_names_directdiff + optnames_turbo
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in func_names:
if this_objfun in history_data:
Func_Values[this_objfun] = history_data[this_objfun]
# for unsteady cases, average time-accurate objective function values
if 'UNSTEADY_SIMULATION' in special_cases and not final_avg:
for key, value in Func_Values.items():
Func_Values[key] = sum(value) / len(value)
# average the final iterations
elif final_avg:
for key, value in Func_Values.iteritems():
# only the last few iterations
i_fin = min([final_avg, len(value)])
value = value[-i_fin:]
Func_Values[key] = sum(value) / len(value)
# otherwise, keep only last value
else:
for key, value in Func_Values.iteritems():
Func_Values[key] = value[-1]
return Func_Values
def find_cfl_number(config, state):
# setup result data
results = ordered_bunch()
results.CONFIG = copy.deepcopy(config)
results.ANALYSES = ordered_dict()
# run a bracketing algorithm
sp.optimize.brent(
func=run_su2,
args=(config, state, results),
brack=tuple(SEARCH_BRACKET),
tol=0.001,
maxiter=MAX_ANALYSES -
3, #analyses are used in the first bracket pass and aren't counted by this algorithm
)
## run a search method
#sp.optimize.fminbound(
#func = run_su2,
#x1 = SEARCH_BRACKET[0],
#x2 = SEARCH_BRACKET[2],
#args = ( config, state, results ),
#xtol = 0.001,
#maxfun = MAX_ANALYSES,
#disp = 0,
#)
# get final result
cfl = np.array([r.CFL_NUMBER for r in results.ANALYSES.values()])
its = np.array([r.ITERATIONS for r in results.ANALYSES.values()])
ind = np.lexsort((-cfl, its))
# done!
print 'Final Result'
print ' CFL = %.4f' % cfl[ind[0]]
print ' Iterations = %i' % its[ind[0]]
# save the data
SU2.io.save_data('experiment_results.pkl', results)
return
def find_cfl_number(config,state):
# setup result data
results = ordered_bunch()
results.CONFIG = copy.deepcopy(config)
results.ANALYSES = ordered_dict()
# run a bracketing algorithm
sp.optimize.brent(
func = run_su2,
args = ( config, state, results ),
brack = tuple(SEARCH_BRACKET),
tol = 0.001,
maxiter = MAX_ANALYSES-3, #analyses are used in the first bracket pass and aren't counted by this algorithm
)
## run a search method
#sp.optimize.fminbound(
#func = run_su2,
#x1 = SEARCH_BRACKET[0],
#x2 = SEARCH_BRACKET[2],
#args = ( config, state, results ),
#xtol = 0.001,
#maxfun = MAX_ANALYSES,
#disp = 0,
#)
# get final result
cfl = np.array([ r.CFL_NUMBER for r in results.ANALYSES.values() ])
its = np.array([ r.ITERATIONS for r in results.ANALYSES.values() ])
ind = np.lexsort((-cfl,its))
# done!
print 'Final Result'
print ' CFL = %.4f' % cfl[ind[0]]
print ' Iterations = %i' % its[ind[0]]
# save the data
SU2.io.save_data('experiment_results.pkl',results)
return
def read_aerodynamics( History_filename , nZones = 1, special_cases=[], final_avg=0 ):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'UNSTEADY_SIMULATION', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename, nZones)
# list of functions to pull
func_names = optnames_aero + grad_names_directdiff + optnames_turbo
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in func_names:
if this_objfun in history_data:
Func_Values[this_objfun] = history_data[this_objfun]
# for unsteady cases, average time-accurate objective function values
if 'UNSTEADY_SIMULATION' in special_cases and not final_avg:
for key,value in Func_Values.items():
Func_Values[key] = sum(value)/len(value)
# average the final iterations
elif final_avg:
for key,value in Func_Values.iteritems():
# only the last few iterations
i_fin = min([final_avg,len(value)])
value = value[-i_fin:]
Func_Values[key] = sum(value)/len(value)
# otherwise, keep only last value
else:
for key,value in Func_Values.iteritems():
Func_Values[key] = value[-1]
return Func_Values
def read_aerodynamics( History_filename , nZones = 1, special_cases=[], final_avg=0 ):
""" values = read_aerodynamics(historyname, special_cases=[])
read aerodynamic function values from history file
Outputs:
dictionary with function keys and thier values
if special cases has 'TIME_MARCHING', returns time averaged data
otherwise returns final value from history file
"""
# read the history data
history_data = read_history(History_filename, nZones)
# pull only these functions
Func_Values = ordered_bunch()
for this_objfun in historyOutFields:
if this_objfun in history_data:
if historyOutFields[this_objfun]['TYPE'] == 'COEFFICIENT' or historyOutFields[this_objfun]['TYPE'] == 'D_COEFFICIENT':
Func_Values[this_objfun] = history_data[this_objfun]
# for unsteady cases, average time-accurate objective function values
if 'TIME_MARCHING' in special_cases and not final_avg:
for key,value in Func_Values.items():
Func_Values[key] = sum(value)/len(value)
# average the final iterations
elif final_avg:
for key,value in Func_Values.iteritems():
# only the last few iterations
i_fin = min([final_avg,len(value)])
value = value[-i_fin:]
Func_Values[key] = sum(value)/len(value)
# otherwise, keep only last value
else:
for key,value in Func_Values.iteritems():
Func_Values[key] = value[-1]
return Func_Values
#: optnames_geo
grad_names_directdiff = [
"D_LIFT", "D_DRAG", "D_SIDEFORCE", "D_MOMENT_X", "D_MOMENT_Y",
"D_MOMENT_Z", "D_FORCE_X", "D_FORCE_Y", "D_FORCE_Z", "D_EFFICIENCY",
"D_CUSTOM_OBJFUNC", "D_HEAT", "D_MAX_HEAT", "D_TOTAL_PRESSURE_LOSS",
"D_TOTAL_EFFICIENCY", "D_TOTAL_PRESSURE_LOSS", "D_KINETIC_ENERGY_LOSS",
"D_TOTAL_STATIC_EFFICIENCY", "D_FLOW_ANGLE_OUT", "D_FLOW_ANGLE_IN",
"D_MASS_FLOW_IN", "D_MASS_FLOW_OUT", "D_PRESSURE_RATIO", "D_ENTHALPY_OUT",
"D_TOTAL_ENTHALPY_OUT", "D_SURFACE_UNIFORMITY", "D_SURFACE_SECONDARY",
"D_SURFACE_MOM_DISTORTION", "D_SURFACE_SECOND_OVER_UNIFORM",
"D_SURFACE_PRESSURE_DROP"
]
grad_names_map = ordered_bunch()
grad_names_map.MASS_FLOW_IN = "D_MASS_FLOW_IN"
grad_names_map.MOMENT_Z = "D_MOMENT_Z"
grad_names_map.FLOW_ANGLE_OUT = "D_FLOW_ANGLE_OUT"
grad_names_map.MASS_FLOW_OUT = "D_MASS_FLOW_OUT"
grad_names_map.FLOW_ANGLE_IN = "D_FLOW_ANGLE_IN"
grad_names_map.FORCE_Z = "D_FORCE_Z"
grad_names_map.FORCE_Y = "D_FORCE_Y"
grad_names_map.FORCE_X = "D_FORCE_X"
grad_names_map.TOTAL_EFFICIENCY = "D_TOTAL_EFFICIENCY"
grad_names_map.TOTAL_STATIC_EFFICIENCY = "D_TOTAL_STATIC_EFFICIENCY"
grad_names_map.PRESSURE_RATIO = "D_PRESSURE_RATIO"
grad_names_map.EFFICIENCY = "D_EFFICIENCY"
grad_names_map.DRAG = "D_DRAG"
grad_names_map.LIFT = "D_LIFT"
grad_names_map.TOTAL_ENTHALPY_OUT = "D_TOTAL_ENTHALPY_OUT"
"D_FORCE_Z",
"D_EFFICIENCY",
"D_TOTAL_PRESSURE_LOSS",
"D_TOTAL_EFFICIENCY",
"D_TOTAL_PRESSURE_LOSS",
"D_KINETIC_ENERGY_LOSS",
"D_TOTAL_STATIC_EFFICIENCY",
"D_FLOW_ANGLE_OUT",
"D_FLOW_ANGLE_IN",
"D_MASS_FLOW_IN",
"D_MASS_FLOW_OUT",
"D_PRESSURE_RATIO",
"D_ENTHALPY_OUT",
"D_TOTAL_ENTHALPY_OUT"]
grad_names_map = ordered_bunch()
grad_names_map.MASS_FLOW_IN = "D_MASS_FLOW_IN"
grad_names_map.MOMENT_Z = "D_MOMENT_Z"
grad_names_map.FLOW_ANGLE_OUT = "D_FLOW_ANGLE_OUT"
grad_names_map.MASS_FLOW_OUT = "D_MASS_FLOW_OUT"
grad_names_map.FLOW_ANGLE_IN = "D_FLOW_ANGLE_IN"
grad_names_map.FORCE_Z = "D_FORCE_Z"
grad_names_map.FORCE_Y = "D_FORCE_Y"
grad_names_map.FORCE_X = "D_FORCE_X"
grad_names_map.TOTAL_EFFICIENCY = "D_TOTAL_EFFICIENCY"
grad_names_map.TOTAL_STATIC_EFFICIENCY = "D_TOTAL_STATIC_EFFICIENCY"
grad_names_map.PRESSURE_RATIO = "D_PRESSURE_RATIO"
grad_names_map.EFFICIENCY = "D_EFFICIENCY"
grad_names_map.DRAG = "D_DRAG"
grad_names_map.LIFT = "D_LIFT"
grad_names_map.TOTAL_ENTHALPY_OUT = "D_TOTAL_ENTHALPY_OUT"
