.ipynb .pdf

repository open issue

Binder Colab

Contents

pyOptSparseDriver

pyOptSparseDriver wraps the optimizer package pyOptSparse, which provides a common interface for 11 optimizers, some of which are included in the package (e.g., SLSQP and NSGA2), and some of which are commercial products that must be obtained from their respective authors (e.g. SNOPT). The pyOptSparse package is based on pyOpt, but adds support for sparse specification of constraint Jacobians. Most of the sparsity features are only applicable when using the SNOPT optimizer.

Note

The pyOptSparse package does not come included with the OpenMDAO installation. It is a separate optional package that can be obtained from mdolab.

In this simple example, we use the SLSQP optimizer to minimize the objective of SellarDerivativesGrouped.

SellarDerivativesGrouped class definition

class SellarDerivativesGrouped(om.Group):
    """
    Group containing the Sellar MDA. This version uses the disciplines with derivatives.
    """

    def initialize(self):
        self.options.declare('nonlinear_solver', default=om.NonlinearBlockGS, recordable=False,
                             desc='Nonlinear solver (class or instance) for Sellar MDA')
        self.options.declare('nl_atol', default=None,
                             desc='User-specified atol for nonlinear solver.')
        self.options.declare('nl_maxiter', default=None,
                             desc='Iteration limit for nonlinear solver.')
        self.options.declare('linear_solver', default=om.ScipyKrylov, recordable=False,
                             desc='Linear solver (class or instance)')
        self.options.declare('ln_atol', default=None,
                             desc='User-specified atol for linear solver.')
        self.options.declare('ln_maxiter', default=None,
                             desc='Iteration limit for linear solver.')
        self.options.declare('mda_nonlinear_solver', default=om.NonlinearBlockGS, recordable=False,
                             desc='Nonlinear solver (class or instance)')
        self.options.declare('mda_linear_solver', default=om.ScipyKrylov, recordable=False,
                             desc='Linear solver (class or instance) for Sellar MDA')

    def setup(self):
        self.mda = mda = self.add_subsystem('mda', om.Group(), promotes=['x', 'z', 'y1', 'y2'])
        mda.add_subsystem('d1', SellarDis1withDerivatives(), promotes=['x', 'z', 'y1', 'y2'])
        mda.add_subsystem('d2', SellarDis2withDerivatives(), promotes=['z', 'y1', 'y2'])

        self.add_subsystem('obj_cmp', om.ExecComp('obj = x**2 + z[1] + y1 + exp(-y2)',
                                                  z=np.array([0.0, 0.0]), x=0.0, y1=0.0, y2=0.0),
                           promotes=['obj', 'x', 'z', 'y1', 'y2'])

        self.add_subsystem('con_cmp1', om.ExecComp('con1 = 3.16 - y1'), promotes=['con1', 'y1'])
        self.add_subsystem('con_cmp2', om.ExecComp('con2 = y2 - 24.0'), promotes=['con2', 'y2'])

        self.set_input_defaults('x', 1.0)
        self.set_input_defaults('z', np.array([5.0, 2.0]))

        nl = self.options['nonlinear_solver']
        self.nonlinear_solver = nl() if inspect.isclass(nl) else nl
        if self.options['nl_atol']:
            self.nonlinear_solver.options['atol'] = self.options['nl_atol']
        if self.options['nl_maxiter']:
            self.nonlinear_solver.options['maxiter'] = self.options['nl_maxiter']

        ln = self.options['linear_solver']
        self.linear_solver = ln() if inspect.isclass(ln) else ln
        if self.options['ln_atol']:
            self.linear_solver.options['atol'] = self.options['ln_atol']
        if self.options['ln_maxiter']:
            self.linear_solver.options['maxiter'] = self.options['ln_maxiter']

        nl = self.options['mda_nonlinear_solver']
        self.mda.nonlinear_solver = nl() if inspect.isclass(nl) else nl

        ln = self.options['mda_linear_solver']
        self.mda.linear_solver = ln() if inspect.isclass(ln) else ln

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SLSQP"

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')
prob.run_driver()

Optimization Problem -- Optimization using pyOpt_sparse
================================================================================
    Objective Function: _objfunc

    Solution: 
--------------------------------------------------------------------------------
    Total Time:                    0.0533
       User Objective Time :       0.0153
       User Sensitivity Time :     0.0317
       Interface Time :            0.0057
       Opt Solver Time:            0.0006
    Calls to Objective Function :       6
    Calls to Sens Function :            6


   Objectives
      Index  Name                   Value          Optimum
          0  obj_cmp.obj     3.183394E+00     0.000000E+00

   Variables (c - continuous, i - integer, d - discrete)
      Index  Name   Type      Lower Bound            Value      Upper Bound     Status
          0  z_0      c    -1.000000E+01     1.977639E+00     1.000000E+01           
          1  z_1      c     0.000000E+00     3.094744E-15     1.000000E+01          l
          2  x_0      c     0.000000E+00    -9.562171E-16     1.000000E+01          l

   Constraints (i - inequality, e - equality)
      Index  Name          Type          Lower           Value           Upper    Status  Lagrange Multiplier (N/A)
          0  con_cmp1.con1    i  -1.000000E+30   -8.491430E-11    0.000000E+00         u    9.00000E+100
          1  con_cmp2.con2    i  -1.000000E+30   -2.024472E+01    0.000000E+00              9.00000E+100

--------------------------------------------------------------------------------

False

print(prob.get_val('z', indices=0))

1.9776388834870342

pyOptSparseDriver Options

Option	Default	Acceptable Values	Acceptable Types	Description	Deprecation
debug_print	[]	N/A	['list']	List of what type of Driver variables to print at each iteration. Valid items in list are 'desvars', 'ln_cons', 'nl_cons', 'objs', 'totals'	N/A
gradient method	openmdao	['snopt_fd', 'openmdao', 'pyopt_fd']	N/A	Finite difference implementation to use	N/A
optimizer	SLSQP	['ALPSO', 'CONMIN', 'FSQP', 'IPOPT', 'NLPQLP', 'NSGA2', 'PSQP', 'SLSQP', 'SNOPT', 'NLPY_AUGLAG', 'NOMAD', 'ParOpt']	N/A	Name of optimizers to use	N/A
print_results	True	[True, False]	['bool']	Print pyOpt results if True	N/A
singular_jac_behavior	warn	['error', 'warn', 'ignore']	N/A	Defines behavior of a zero row/col check after first call tocompute_totals:error - raise an error.warn - raise a warning.ignore - don't perform check.	N/A
singular_jac_tol	1e-16	N/A	N/A	Tolerance for zero row/column check.	N/A
title	Optimization using pyOpt_sparse	N/A	N/A	Title of this optimization run	N/A
user_teriminate_signal	N/A	N/A	N/A	OS signal that triggers a clean user-termination. Only SNOPTsupports this option.	The option 'user_teriminate_signal' was misspelled and will be deprecated. Please use 'user_terminate_signal' instead.
user_terminate_signal	N/A	N/A	N/A	OS signal that triggers a clean user-termination. Only SNOPTsupports this option.	N/A

pyOptSparseDriver Constructor

The call signature for the pyOptSparseDriver constructor is:

pyOptSparseDriver.__init__(**kwargs)

Initialize pyopt.

Using pyOptSparseDriver

pyOptSparseDriver has a small number of unified options that can be specified as keyword arguments when it is instantiated or by using the “options” dictionary. We have already shown how to set the optimizer option. Next we see how the print_results option can be used to turn on or off the echoing of the results when the optimization finishes. The default is True, but here, we turn it off.

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver(optimizer='SLSQP')

prob.driver.options['print_results'] = False

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')
prob.run_driver()

print(prob.get_val('z', indices=0))

1.9776388834870342

Every optimizer also has its own specialized settings that allow you to fine-tune the algorithm that it uses. You can access these within the opt_setting dictionary. These options are different for each optimizer, so to find out what they are, you need to read your optimizer’s documentation. We present a few common ones here.

SLSQP-Specific Settings

Here, we set a convergence tolerance for SLSQP:

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SLSQP"

prob.driver.opt_settings['ACC'] = 1e-9

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')
prob.run_driver()

Optimization Problem -- Optimization using pyOpt_sparse
================================================================================
    Objective Function: _objfunc

    Solution: 
--------------------------------------------------------------------------------
    Total Time:                    0.0542
       User Objective Time :       0.0161
       User Sensitivity Time :     0.0319
       Interface Time :            0.0057
       Opt Solver Time:            0.0005
    Calls to Objective Function :       6
    Calls to Sens Function :            6


   Objectives
      Index  Name                   Value          Optimum
          0  obj_cmp.obj     3.183394E+00     0.000000E+00

   Variables (c - continuous, i - integer, d - discrete)
      Index  Name   Type      Lower Bound            Value      Upper Bound     Status
          0  z_0      c    -1.000000E+01     1.977639E+00     1.000000E+01           
          1  z_1      c     0.000000E+00     3.094744E-15     1.000000E+01          l
          2  x_0      c     0.000000E+00    -9.562171E-16     1.000000E+01          l

   Constraints (i - inequality, e - equality)
      Index  Name          Type          Lower           Value           Upper    Status  Lagrange Multiplier (N/A)
          0  con_cmp1.con1    i  -1.000000E+30   -8.491430E-11    0.000000E+00         u    9.00000E+100
          1  con_cmp2.con2    i  -1.000000E+30   -2.024472E+01    0.000000E+00              9.00000E+100

--------------------------------------------------------------------------------

False

print(prob.get_val('z', indices=0))

1.9776388834870342

Similarly, we can set an iteration limit. Here, we set it to just a few iterations, and don’t quite reach the optimum.

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SLSQP"

prob.driver.opt_settings['MAXIT'] = 3

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')
prob.run_driver()

Optimization Problem -- Optimization using pyOpt_sparse
================================================================================
    Objective Function: _objfunc

    Solution: 
--------------------------------------------------------------------------------
    Total Time:                    0.0364
       User Objective Time :       0.0110
       User Sensitivity Time :     0.0213
       Interface Time :            0.0037
       Opt Solver Time:            0.0003
    Calls to Objective Function :       4
    Calls to Sens Function :            4


   Objectives
      Index  Name                   Value          Optimum
          0  obj_cmp.obj     3.203561E+00     0.000000E+00

   Variables (c - continuous, i - integer, d - discrete)
      Index  Name   Type      Lower Bound            Value      Upper Bound     Status
          0  z_0      c    -1.000000E+01     1.983377E+00     1.000000E+01           
          1  z_1      c     0.000000E+00    -1.034460E-12     1.000000E+01          l
          2  x_0      c     0.000000E+00    -9.369164E-15     1.000000E+01          l

   Constraints (i - inequality, e - equality)
      Index  Name          Type          Lower           Value           Upper    Status  Lagrange Multiplier (N/A)
          0  con_cmp1.con1    i  -1.000000E+30   -2.043382E-02    0.000000E+00              9.00000E+100
          1  con_cmp2.con2    i  -1.000000E+30   -2.023325E+01    0.000000E+00              9.00000E+100

--------------------------------------------------------------------------------

True

print(prob.get_val('z', indices=0))

1.983377083308787

SNOPT-Specific Settings

SNOPT has many customizable settings. Here we show two common ones.

Setting the convergence tolerance:

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SNOPT"

prob.driver.opt_settings['Major feasibility tolerance'] = 1e-9

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')
prob.run_driver()

Optimization Problem -- Optimization using pyOpt_sparse
================================================================================
    Objective Function: _objfunc

    Solution: 
--------------------------------------------------------------------------------
    Total Time:                    0.0682
       User Objective Time :       0.0197
       User Sensitivity Time :     0.0403
       Interface Time :            0.0066
       Opt Solver Time:            0.0017
    Calls to Objective Function :       8
    Calls to Sens Function :            7


   Objectives
      Index  Name                   Value          Optimum
          0  obj_cmp.obj     3.183394E+00     0.000000E+00

   Variables (c - continuous, i - integer, d - discrete)
      Index  Name   Type      Lower Bound            Value      Upper Bound     Status
          0  z_0      c    -1.000000E+01     1.977639E+00     1.000000E+01           
          1  z_1      c     0.000000E+00     0.000000E+00     1.000000E+01          l
          2  x_0      c     0.000000E+00     0.000000E+00     1.000000E+01          l

   Constraints (i - inequality, e - equality)
      Index  Name          Type          Lower           Value           Upper    Status  Lagrange Multiplier
          0  con_cmp1.con1    i  -1.000000E+30   -5.966339E-12    0.000000E+00         u    -9.86840E-01
          1  con_cmp2.con2    i  -1.000000E+30   -2.024472E+01    0.000000E+00               0.00000E+00

--------------------------------------------------------------------------------

False

print(prob.get_val('z', indices=0))

1.9776388834648047

Setting a limit on the number of major iterations. Here, we set it to just a few iterations, and don’t quite reach the optimum.

import numpy as np

import openmdao.api as om
from openmdao.test_suite.components.sellar import SellarDerivativesGrouped

prob = om.Problem()
model = prob.model = SellarDerivativesGrouped()

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SNOPT"

# after upgrading to SNOPT 7.5-1.1, this test failed unless iter limit raised from 4 to 5
prob.driver.opt_settings['Major iterations limit'] = 5

model.add_design_var('z', lower=np.array([-10.0, 0.0]), upper=np.array([10.0, 10.0]))
model.add_design_var('x', lower=0.0, upper=10.0)
model.add_objective('obj')
model.add_constraint('con1', upper=0.0)
model.add_constraint('con2', upper=0.0)

prob.set_solver_print(level=0)

prob.setup(check=False, mode='rev')

prob.run_driver()

Optimization Problem -- Optimization using pyOpt_sparse
================================================================================
    Objective Function: _objfunc

    Solution: 
--------------------------------------------------------------------------------
    Total Time:                    0.0574
       User Objective Time :       0.0179
       User Sensitivity Time :     0.0323
       Interface Time :            0.0057
       Opt Solver Time:            0.0016
    Calls to Objective Function :       7
    Calls to Sens Function :            6


   Objectives
      Index  Name                   Value          Optimum
          0  obj_cmp.obj     3.183402E+00     0.000000E+00

   Variables (c - continuous, i - integer, d - discrete)
      Index  Name   Type      Lower Bound            Value      Upper Bound     Status
          0  z_0      c    -1.000000E+01     1.977641E+00     1.000000E+01           
          1  z_1      c     0.000000E+00     0.000000E+00     1.000000E+01          l
          2  x_0      c     0.000000E+00     0.000000E+00     1.000000E+01          l

   Constraints (i - inequality, e - equality)
      Index  Name          Type          Lower           Value           Upper    Status  Lagrange Multiplier
          0  con_cmp1.con1    i  -1.000000E+30   -8.621022E-06    0.000000E+00              -9.86840E-01
          1  con_cmp2.con2    i  -1.000000E+30   -2.024472E+01    0.000000E+00               0.00000E+00

--------------------------------------------------------------------------------

True

print(prob.get_val('z', indices=0))

1.9776413083133966

If you have pyoptsparse 1.1 or greater, then you can send a signal such as SIGUSR1 to a running SNOPT optimization to tell it to terminate cleanly. This is useful if an optimization has gotten close enough to an optimum. How to do this is dependent on your operating system in all cases, on your mpi implementation if you are running mpi, and on your queuing software if you are on a supercomputing cluster. Here is a simple example for unix and mpi.

    ktmoore1$ ps -ef |grep sig
      502 17955   951   0  4:05PM ttys000    0:00.02 mpirun -n 2 python sig_demo.py
      502 17956 17955   0  4:05PM ttys000    0:00.03 python sig_demo.py
      502 17957 17955   0  4:05PM ttys000    0:00.03 python sig_demo.py
      502 17959 17312   0  4:05PM ttys001    0:00.00 grep sig

    ktmoore1$ kill -SIGUSR1 17955

You can enable this feature by setting the “user_terminate_signal” option and giving it a signal (imported from the signal library in Python). By default, user_terminate_signal is None, which disables the feature. Here, we set the signal to SIGUSR1:

import openmdao.api as om
import signal

prob = om.Problem()
model = prob.model

prob.driver = om.pyOptSparseDriver()
prob.driver.options['optimizer'] = "SNOPT"
prob.driver.options['user_terminate_signal'] = signal.SIGUSR1

You can learn more about the available options in the SNOPT_Manual.