tfp.substrates.numpy.math.ode.BDF

Backward differentiation formula (BDF) solver for stiff ODEs.

Inherits From: Solver

Implements the solver described in [Shampine and Reichelt (1997)][1], a variable step size, variable order (VSVO) BDF integrator with order varying between 1 and 5.

Algorithm details

Each step involves solving the following nonlinear equation by Newton's method:

0 = 1/1 * BDF(1, y)[n+1] + ... + 1/k * BDF(k, y)[n+1]
  - h ode_fn(t[n+1], y[n+1])
  - bdf_coefficients[k-1] * (1/1 + ... + 1/k) * (y[n+1] - y[n] - BDF(1, y)[n]
                                                        -  ... - BDF(k, y)[n])

where k >= 1 is the current order of the integrator, h is the current step size, bdf_coefficients is a list of numbers that parameterizes the method, and BDF(m, y) is the m-th order backward difference of the vector y. In particular, BDF(0, y)[n] = y[n] and BDF(m + 1, y)[n] = BDF(m, y)[n] - BDF(m, y)[n - 1] for m >= 0.

Newton's method can fail because

  • the method has exceeded the maximum number of iterations,
  • the method is converging too slowly, or
  • the method is not expected to converge (the last two conditions are determined by approximating the Lipschitz constant associated with the iteration).

When evaluate_jacobian_lazily is True, the solver avoids evaluating the Jacobian of the dynamics function as much as possible. In particular, Newton's method will try to use the Jacobian from a previous integration step. If Newton's method fails with an out-of-date Jacobian, the Jacobian is re-evaluated and Newton's method is restarted. If Newton's method fails and the Jacobian is already up-to-date, then the step size is decreased and Newton's method is restarted.

Even if Newton's method converges, the solution it generates can still be rejected if it exceeds the specified error tolerance due to truncation error. In this case, the step size is decreased and Newton's method is restarted.

References

[1]: Lawrence F. Shampine and Mark W. Reichelt. The MATLAB ODE Suite. SIAM Journal on Scientific Computing 18(1):1-22, 1997.

rtol Optional float Tensor specifying an upper bound on relative error, per element of the dependent variable. The error tolerance for the next step is tol = atol + rtol * abs(state) where state is the computed state at the current step (see also atol). The next step is rejected if it incurs a local truncation error larger than tol. Default value: 1e-3.
atol Optional float Tensor specifying an upper bound on absolute error, per element of the dependent variable (see also rtol). Default value: 1e-6.
first_step_size Optional scalar float Tensor specifying the size of the first step. If unspecified, the size is chosen automatically. Default value: None.
safety_factor Scalar positive float Tensor. When Newton's method converges, the solver may choose to update the step size by applying a multiplicative factor to the current step size. This factor is factor = clamp(factor_unclamped, min_step_size_factor, max_step_size_factor) where factor_unclamped = error_ratio**(-1. / (order + 1)) * safety_factor (see also min_step_size_factor and max_step_size_factor). A small (respectively, large) value for the safety factor causes the solver to take smaller (respectively, larger) step sizes. A value larger than one, though not explicitly prohibited, is discouraged. Default value: 0.9.
min_step_size_factor Scalar float Tensor (see safety_factor). Default value: 0.1.
max_step_size_factor Scalar float Tensor (see safety_factor). Default value: 10..
max_num_steps Optional scalar integer Tensor specifying the maximum number of steps allowed (including rejected steps). If unspecified, there is no upper bound on the number of steps. Default value: None.
max_order Scalar integer Tensor taking values between 1 and 5 (inclusive) specifying the maximum BDF order. Default value: 5.
max_num_newton_iters Optional scalar integer Tensor specifying the maximum number of iterations per invocation of Newton's method. If unspecified, there is no upper bound on the number iterations. Default value: 4.
newton_tol_factor Scalar float Tensor used to determine the stopping condition for Newton's method. In particular, Newton's method terminates when the distance to the root is estimated to be less than newton_tol_factor * norm(atol + rtol * abs(state)) where state is the computed state at the current step. Default value: 0.1.
newton_step_size_factor Scalar float Tensor specifying a multiplicative factor applied to the size of the integration step when Newton's method fails. Default value: 0.5.
bdf_coefficients 1-D float Tensor with 5 entries that parameterize the solver. The default values are those proposed in [1]. Default value: (-0.1850, -1. / 9., -0.0823, -0.0415, 0.).
evaluate_jacobian_lazily Optional boolean specifying whether the Jacobian should be evaluated at each point in time or as needed (i.e., lazily). Default value: True.
make_adjoint_solver_fn Callable that takes no arguments that constructs a Solver instance. The created solver is used in the adjoint senstivity analysis to compute gradients (if they are requested). Default value: A callable that returns this solver.
validate_args Whether to validate input with asserts. If validate_args is False and the inputs are invalid, correct behavior is not guaranteed. Default value: False.
name Python str name prefixed to Ops created by this function. Default value: None (i.e., 'bdf').

name

Methods

solve

View source

Solves an initial value problem.

An initial value problem consists of a system of ODEs and an initial condition:

dy/dt(t) = ode_fn(t, y(t), **constants)
y(initial_time) = initial_state

Here, t (also called time) is a scalar float Tensor and y(t) (also called the state at time t) is an N-D float or complex Tensor. constants is are values that are constant with respect to time. Passing the constants here rather than just closing over them in ode_fn is only necessary if you want gradients with respect to these values.

Example

The ODE dy/dt(t) = dot(A, y(t)) is solved below.

t_init, t0, t1 = 0., 0.5, 1.
y_init = tf.constant([1., 1.], dtype=tf.float64)
A = tf.constant([[-1., -2.], [-3., -4.]], dtype=tf.float64)

def ode_fn(t, y):
  return tf.linalg.matvec(A, y)

results = tfp.math.ode.BDF().solve(ode_fn, t_init, y_init,
                                   solution_times=[t0, t1])
y0 = results.states[0]  # == dot(matrix_exp(A * t0), y_init)
y1 = results.states[1]  # == dot(matrix_exp(A * t1), y_init)

If the exact solution times are not important, it can be much more efficient to let the solver choose them using solution_times=tfp.math.ode.ChosenBySolver(final_time=1.). This yields the state at various times between t_init and final_time, in which case results.states[i] is the state at time results.times[i].

Gradients

The gradients are computed using the adjoint sensitivity method described in [Chen et al. (2018)][1].

grad = tf.gradients(y1, y0) # == dot(e, J)
# J is the Jacobian of y1 with respect to y0. In this case, J = exp(A * t1).
# e = [1, ..., 1] is the row vector of ones.

This is not capable of computing gradients with respect to values closed over by ode_fn, e.g., in the example above:

def ode_fn(t, y):
  return tf.linalg.matvec(A, y)

with tf.GradientTape() as tape:
  tape.watch(A)
  results = tfp.math.ode.BDF().solve(ode_fn, t_init, y_init,
                                     solution_times=[t0, t1])
tape.gradient(results.states, A)  # Undefined!

There are two options to get the gradients flowing through these values:

  1. Use tf.Variable for these values.
  2. Pass the values in explicitly using the constants argument:
def ode_fn(t, y, A):
  return tf.linalg.matvec(A, y)

with tf.GradientTape() as tape:
  tape.watch(A)
  results = tfp.math.ode.BDF().solve(ode_fn, t_init, y_init,
                                     solution_times=[t0, t1],
                                     constants={'A': A})
tape.gradient(results.states, A)  # Fine.

By default, this uses the same solver for the augmented ODE. This can be controlled via make_adjoint_solver_fn.

References

[1]: Chen, Tian Qi, et al. "Neural ordinary differential equations." Advances in Neural Information Processing Systems. 2018.

Args
ode_fn Function of the form ode_fn(t, y, **constants). The input t is a scalar float Tensor. The input y and output are both Tensors with the same shape and dtype as initial_state. constants is are values that are constant with respect to time. Passing the constants here rather than just closing over them in ode_fn is only necessary if you want gradients with respect to these values.
initial_time Scalar float Tensor specifying the initial time.
initial_state N-D float or complex Tensor specifying the initial state. The dtype of initial_state must be complex for problems with complex-valued states (even if the initial state is real).
solution_times 1-D float Tensor specifying a list of times. The solver stores the computed state at each of these times in the returned Results object. Must satisfy initial_time <= solution_times[0] and solution_times[i] < solution_times[i+1]. Alternatively, the user can pass tfp.math.ode.ChosenBySolver(final_time) where final_time is a scalar float Tensor satisfying initial_time < final_time. Doing so requests that the solver automatically choose suitable times up to and including final_time at which to store the computed state.
jacobian_fn Optional function of the form jacobian_fn(t, y). The input t is a scalar float Tensor. The input y has the same shape and dtype as initial_state. The output is a 2N-D Tensor whose shape is initial_state.shape + initial_state.shape and whose dtype is the same as initial_state. In particular, the (i1, ..., iN, j1, ..., jN)-th entry of jacobian_fn(t, y) is the derivative of the (i1, ..., iN)-th entry of ode_fn(t, y) with respect to the (j1, ..., jN)-th entry of y. If this argument is left unspecified, the solver automatically computes the Jacobian if and when it is needed. Default value: None.
jacobian_sparsity Optional 2N-D boolean Tensor whose shape is initial_state.shape + initial_state.shape specifying the sparsity pattern of the Jacobian. This argument is ignored if jacobian_fn is specified. Default value: None.
batch_ndims Optional nonnegative integer. When specified, the first batch_ndims dimensions of initial_state are batch dimensions. Default value: None.
previous_solver_internal_state Optional solver-specific argument used to warm-start this invocation of solve. Default value: None.
constants Optional dictionary with string keys and values being (possibly nested) float Tensors. These represent values that are constant with respect to time. Specifying these here allows the adjoint sentitivity method to compute gradients of the results with respect to these values.

Returns
Object of type Results.