def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
# if os.path.isfile('codecache.pkl'):
# with open('codecache.pkl','rb') as f:
# deriv_func_bck = pickle.load(f)
#
# else:
# deriv_func_bck = None
out_ws = QCPICodeGen({'problem_data': problem_data})
print(out_ws['bc_func_code'])
print(out_ws['deriv_func_code'])
self.bvp = BVP(out_ws['deriv_func_fn'],
out_ws['bc_func_fn'], None) #out_ws['compute_control_fn'])#out_ws['compute_control_fn'])
self.stm_ode_func = self.bvp.deriv_func
self.module = out_ws['code_module']
self.left_bc_mask = out_ws['code_module'].bc_free_mask
self.num_dae_vars = out_ws['code_module'].num_dae_vars
# self.left_bc_mask = problem_data['bc_free_mask']
# print(self.left_bc_mask)
self.bc_left_fn = out_ws['code_module'].bc_func_left
self.bc_right_fn = out_ws['code_module'].bc_func_right
# self.deriv_func = njit(parallel=True)(out_ws['code_module'].deriv_func_mcpi)
# if deriv_func_bck is not None:
# print('Loaded deriv_func from cache')
# self.deriv_func = deriv_func_bck
# self.saved_deriv_func = True
# else:
# self.deriv_func = out_ws['code_module'].deriv_func_mcpi
self.deriv_func = out_ws['code_module'].deriv_func
sys.modules['_beluga_'+problem_data['problem_name']] = out_ws['code_module']
return out_ws['code_module']
def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
cache_exists = os.path.isfile('codecache.pkl')
out_ws = PythonCodeGen({'problem_data': problem_data})
if not self.cached or not cache_exists:
print(out_ws['bc_func_code'])
print(out_ws['deriv_func_code'])
self.stm_ode_func = self.make_stmode(out_ws['deriv_func_fn'],
problem_data['nOdes'])
self.saved_code = False
else:
bvp_data = self.load_code()
out_ws['deriv_func_fn'] = bvp_data['deriv_fn']
out_ws['code_module'].deriv_func = bvp_data['deriv_fn']
# self.stm_ode_func = bvp_data['stmode_fn']
self.stm_ode_func = self.make_stmode(bvp_data['deriv_fn'],
problem_data['nOdes'])
out_ws['code_module'].deriv_func_ode45 = self.deriv_func_ode45
self.out_ws = out_ws
self.bvp = BVP(
out_ws['deriv_func_fn'], out_ws['bc_func_fn'],
out_ws['compute_control_fn']) #out_ws['compute_control_fn'])
# self.stm_ode_func = ft.partial(self._stmode_fd, odefn=self.bvp.deriv_func)
self.bc_jac_multi = ft.partial(self.__bc_jac_multi,
bc_func=self.bvp.bc_func)
sys.modules['_beluga_' +
problem_data['problem_name']] = out_ws['code_module']
return out_ws['code_module']
def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
out_ws = PythonCodeGen({'problem_data': problem_data})
logging.debug(out_ws['bc_func_code'])
logging.debug(out_ws['deriv_func_code'])
if self.use_numba:
deriv_func = numba.njit(parallel=False)(
out_ws['code_module'].deriv_func_nojit)
else:
deriv_func = out_ws['code_module'].deriv_func_nojit
out_ws['deriv_func_fn'] = deriv_func
out_ws['code_module'].deriv_func = deriv_func
self.saved_code = False
self.out_ws = out_ws
# self.stm_ode_func = self.make_stmode(out_ws['deriv_func_fn'], problem_data['nOdes'])
self.bvp = BVP(
out_ws['deriv_func_fn'], out_ws['bc_func_fn'],
out_ws['compute_control_fn']) #out_ws['compute_control_fn'])
# self.stm_ode_func = ft.partial(self._stmode_fd, odefn=self.bvp.deriv_func)
self.stm_ode_func = self.make_stmode(self.bvp.deriv_func,
problem_data['nOdes'])
self.bc_jac_multi = ft.partial(self.__bc_jac_multi,
bc_func=self.bvp.bc_func)
sys.modules['_beluga_' +
problem_data['problem_name']] = out_ws['code_module']
return out_ws['code_module']
def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
out_ws = PythonCodeGen({'problem_data': problem_data})
logging.debug(out_ws['bc_func_code'])
logging.debug(out_ws['deriv_func_code'])
cache_exists = os.path.isfile('codecache.pkl')
# if not self.cached or not cache_exists:
deriv_func = numba.njit(parallel=True)(
out_ws['code_module'].deriv_func_nojit)
out_ws['deriv_func_fn'] = deriv_func
out_ws['code_module'].deriv_func = deriv_func
self.saved_code = False
# else:
# bvp_data = self.load_code()
# print('deriv func',id(bvp_data['deriv_fn']))
# out_ws['deriv_func_fn'] = bvp_data['deriv_fn']
# out_ws['code_module'].deriv_func = bvp_data['deriv_fn']
self.out_ws = out_ws
# self.stm_ode_func = self.make_stmode(out_ws['deriv_func_fn'], problem_data['nOdes'])
self.bvp = BVP(
out_ws['deriv_func_fn'], out_ws['bc_func_fn'],
out_ws['compute_control_fn']) #out_ws['compute_control_fn'])
# self.stm_ode_func = ft.partial(self._stmode_fd, odefn=self.bvp.deriv_func)
# self.stm_ode_func = self.make_stmode(self.bvp.deriv_func, problem_data['nOdes'])
# self.bc_jac_multi = ft.partial(self.__bc_jac_multi, bc_func=self.bvp.bc_func)
sys.modules['_beluga_' +
problem_data['problem_name']] = out_ws['code_module']
return out_ws['code_module']
def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
out_ws = PythonCodeGen({'problem_data': problem_data})
print(out_ws['bc_func_code'])
print(out_ws['deriv_func_code'])
self.bvp = BVP(
out_ws['deriv_func_fn'], out_ws['bc_func_fn'],
out_ws['compute_control_fn']) #out_ws['compute_control_fn'])
# self.stm_ode_func = ft.partial(self.__stmode_fd, odefn=self.bvp.deriv_func)
# self.bc_jac_func = ft.partial(self.__bcjac_fd, bc_func=self.bvp.bc_func)
return out_ws['code_module']
def preprocess(self, problem_data):
"""Code generation and compilation before running solver."""
out_ws = PythonCodeGen({'problem_data': problem_data})
self.bvp = BVP(out_ws['deriv_func_fn'], out_ws['bc_func_fn'],
out_ws['compute_control_fn'])
if self.derivative_method == 'csd':
self.stm_ode_func = ft.partial(self.__stmode_csd,
odefn=self.bvp.deriv_func)
self.bc_jac_func = ft.partial(self.__bcjac_csd,
bc_func=self.bvp.bc_func)
elif self.derivative_method == 'fd':
self.stm_ode_func = ft.partial(self.__stmode_fd,
odefn=self.bvp.deriv_func)
self.bc_jac_func = ft.partial(self.__bcjac_fd,
bc_func=self.bvp.bc_func)
return self.bvp