pypesto.profile.profile_next_guess

Functions

adaptive_step(x, par_index, par_direction, ...)	Group of more complex methods for point proposal.
do_line_search(next_x, step_size_guess, ...)	Perform the line search.
fixed_step(x, par_index, par_direction, ...)	Most simple method to create the next guess.
get_reg_polynomial(par_index, ...)	Compute the regression polynomial.
handle_profile_history(x, par_index, ...)	Compute the very first step direction update guesses.
next_guess(x, par_index, par_direction, ...)	Create the next initial guess for the optimizer.
next_x_interpolate(next_obj, last_obj, ...)	Interpolate between the last two steps.

pypesto.profile.profile_next_guess.adaptive_step(x, par_index, par_direction, options, current_profile, problem, global_opt, order=1, min_step_increase_factor=1.0, max_step_reduce_factor=1.0)

Group of more complex methods for point proposal.

Step size is automatically computed by a line search algorithm (hence: adaptive).

Parameters:

• x (ndarray) – The current position of the profiler, size dim_full.

• par_index (int) – The index of the parameter of the current profile.

• par_direction (Literal[1, -1]) – The direction, in which the profiling is done (1 or -1).

• options (ProfileOptions) – Various options applied to the profile optimization.

• current_profile (ProfilerResult) – The profile which should be computed.

• problem (Problem) – The problem to be solved.

• global_opt (float) – Log-posterior value of the global optimum.

• order (int) –

  Specifies the precise algorithm for extrapolation. Available options are:

  • 0: just one parameter is updated

  • 1: the last two points are used to extrapolate all parameters

  • np.nan: indicates that a more complex regression should be used as determined by pypesto.profile.ProfileOptions.reg_order.

• min_step_increase_factor (float) – Factor to increase the minimal step size bound.

• max_step_reduce_factor (float) – Factor to reduce the maximal step size bound.

Return type:

ndarray

Returns:

The updated parameter vector, of size dim_full.

pypesto.profile.profile_next_guess.fixed_step(x, par_index, par_direction, options, problem)

Most simple method to create the next guess.

Computes the next point based on the fixed step size given by pypesto.profile.ProfileOptions.default_step_size.

Parameters:

• x (ndarray) – The current position of the profiler, size dim_full.

• par_index (int) – The index of the parameter of the current profile.

• par_direction (Literal[1, -1]) – The direction, in which the profiling is done (1 or -1).

• options (ProfileOptions) – Various options applied to the profile optimization.

• problem (Problem) – The problem to be solved.

Return type:

ndarray

Returns:

The updated parameter vector, of size dim_full.

pypesto.profile.profile_next_guess.next_guess(x, par_index, par_direction, profile_options, update_type, current_profile, problem, global_opt, min_step_increase_factor=1.0, max_step_reduce_factor=1.0)

Create the next initial guess for the optimizer.

Used in order to compute the next profile point. Different proposal methods are available.

Parameters:

• x (ndarray) – The current position of the profiler.

• par_index (int) – The index of the parameter of the current profile.

• par_direction (Literal[1, -1]) – The direction, in which the profiling is done (1 or -1).

• profile_options (ProfileOptions) – Various options applied to the profile optimization.

• update_type (Literal['fixed_step', 'adaptive_step_order_0', 'adaptive_step_order_1', 'adaptive_step_regression']) –

  Type of update for next profile point. Available options are:

  • fixed_step (see fixed_step())

  • adaptive_step_order_0 (see adaptive_step()).

  • adaptive_step_order_1 (see adaptive_step()).

  • adaptive_step_regression (see adaptive_step()).

• current_profile (ProfilerResult) – The profile which should be computed.

• problem (Problem) – The problem to be solved.

• global_opt (float) – Log-posterior value of the global optimum.

• min_step_increase_factor (float) – Factor to increase the minimal step size bound. Used only in adaptive_step().

• max_step_reduce_factor (float) – Factor to reduce the maximal step size bound. Used only in adaptive_step().

Return type:

ndarray

Returns:

The next initial guess as base for the next profile point.