invpinndata

Class which is used to implement data for inverse PINNs

invpinndata

Bases: data

Class for data on purely spatial inverse problems with a pinn.

__init__(domain, boundaries, dimdata, udata, n_clp=10000, n_bc=600)

Constructor for class

Parameters:
  • domain (domain) –

    Domain to generate data on.

  • boundaries (boundaries) –

    Boundary to generate data on.

  • dimdata (list) –

    List of data across spatial dimensions in which udata is recorded on for each u, each as (N,) shape tensors.

  • udata (list) –

    List containing data for each u solving for to solve inverse cosntants, each as (N,) shape tensors

  • n_clp (int, default: 10000 ) –

    Number of collocation points.

  • n_bc (int, default: 600 ) –

    Number of boundary condition points.

get_clp()

Returns:
  • tensor –

    Sampled collocation points.

get_bcp()

Returns:
  • tensor –

    Sampled boundary points.

get_domain()

Returns:
  • domain –

    Domain data is generated on.

get_boundaries()

Returns:
  • boundaries –

    Boundary data is generated on.

get_n_clp()

Returns:
  • int –

    Number of collocation points sampled.

get_n_bc()

Returns:
  • int –

    Number of boundary points sampled.

get_n_invp()

Returns:
  • int –

    Number of data points for inverse data given.

get_invp()

Returns:
  • tensor –

    Sampled data points for inverse data.

makeInverseData(dimdata, udata)

Combines data user provided into consistent trainable set

Parameters:
  • dimdata (list) –

    List of data across spatial dimensions in which udata is recorded on for each u, each as (N,) shape tensors.

  • udata (list) –

    List containing data for each u solving for to solve inverse cosntants, each as (N,) shape tensors