model_specific.f90 File Reference

Go to the source code of this file.

Functions/Subroutines
subroutine	configure_model
	subroutine called initially to set up details and data for model specific functions More...
subroutine	reconfigure_model
	subroutine to reset variables that may change when the observation network changes More...
subroutine	solve_r (obsDim, nrhs, y, v, t)
	subroutine to take an observation vector y and return v in observation space. More...
subroutine	solve_rhalf (obsdim, nrhs, y, v, t)
	subroutine to take an observation vector y and return v in observation space. More...
subroutine	solve_hqht_plus_r (obsdim, y, v, t)
	subroutine to take an observation vector y and return v in observation space. More...
subroutine	q (nrhs, x, Qx)
	subroutine to take a full state vector x and return Qx in state space. More...
subroutine	qhalf (nrhs, x, Qx)
	subroutine to take a full state vector x and return \(Q^{1/2}x\) in state space. More...
subroutine	r (obsDim, nrhs, y, Ry, t)
	subroutine to take an observation vector x and return Rx in observation space. More...
subroutine	rhalf (obsDim, nrhs, y, Ry, t)
	subroutine to take an observation vector x and return Rx in observation space. More...
subroutine	h (obsDim, nrhs, x, hx, t)
	subroutine to take a full state vector x and return H(x) in observation space. More...
subroutine	ht (obsDim, nrhs, y, x, t)
	subroutine to take an observation vector y and return x \(= H^T(y)\) in full state space. More...
subroutine	dist_st_ob (xp, yp, dis, t)
	subroutine to compute the distance between the variable in the state vector and the variable in the observations More...
subroutine	bhalf (nrhs, x, bx)
	subroutine to take a full state vector x and return \(B^{1/2}x\) in state space. More...
subroutine	solve_b (nrhs, x, v)
	subroutine to take a state vector x and return v in state space. More...
subroutine	get_observation_data (y, t)
	Subroutine to read observation from a file . More...

Function/Subroutine Documentation

subroutine bhalf	(	integer, intent(in)	nrhs,
		real(kind=rk), dimension(state_dim,nrhs), intent(in)	x,
		real(kind=rk), dimension(state_dim,nrhs), intent(out)	bx
	)

subroutine to take a full state vector x and return \(B^{1/2}x\) in state space.

Given \(x\) compute \(B^{\frac{1}{2}}x\)

Parameters

[in]	nrhs	the number of right hand sides
[in]	x	the input vector
[out]	bx	the resulting vector where bx \(= B^{\frac{1}{2}}x\)

Definition at line 281 of file model_specific.f90.

Here is the caller graph for this function:

subroutine configure_model

(

)

subroutine called initially to set up details and data for model specific functions

By the end of this subroutine, the following must be set:

• state_dim in sizes
• obs_dim in sizes for the first observation
• total_timesteps in timestep_data

This is a very good place to load in data for the matrices B,Q,R,H etc

Definition at line 38 of file model_specific.f90.

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine dist_st_ob	(	integer, intent(in)	xp,
		integer, intent(in)	yp,
		real(kind=kind(1.0d0)), intent(out)	dis,
		integer, intent(in)	t
	)

subroutine to compute the distance between the variable in the state vector and the variable in the observations

Compute \(\mathrm{dist}(x(xp),y(yp))\)

Parameters

[in]	xp	the index in the state vector
[in]	yp	the index in the observation vector
[out]	dis	the distance between x(xp) and y(yp)
[in]	t	the current time index for observations

Definition at line 265 of file model_specific.f90.

Here is the caller graph for this function:

subroutine get_observation_data	(	real(kind=rk), dimension(obs_dim), intent(out)	y,
		integer, intent(in)	t
	)

Subroutine to read observation from a file
.

Parameters

[out]	y	The observation
[in]	t	the current timestep

Definition at line 319 of file model_specific.f90.

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine h	(	integer, intent(in)	obsDim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(state_dim,nrhs), intent(in)	x,
		real(kind=rk), dimension(obsdim,nrhs), intent(out)	hx,
		integer, intent(in)	t
	)

subroutine to take a full state vector x and return H(x) in observation space.

Given \(x\) compute \(Hx\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	x	the input vectors in state space
[out]	hx	the resulting vector in observation space where hx \(= Hx\)
[in]	t	the timestep

Definition at line 221 of file model_specific.f90.

Here is the caller graph for this function:

subroutine ht	(	integer, intent(in)	obsDim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(obsdim,nrhs), intent(in)	y,
		real(kind=rk), dimension(state_dim,nrhs), intent(out)	x,
		integer, intent(in)	t
	)

subroutine to take an observation vector y and return x \(= H^T(y)\) in full state space.

Given \(y\) compute \(x=H^T(y)\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	y	the input vectors in observation space
[out]	x	the resulting vector in state space where x \(= H^Ty\)
[in]	t	the timestep

Definition at line 243 of file model_specific.f90.

Here is the caller graph for this function:

subroutine q	(	integer, intent(in)	nrhs,
		real(kind=rk), dimension(state_dim,nrhs), intent(in)	x,
		real(kind=rk), dimension(state_dim,nrhs), intent(out)	Qx
	)

subroutine to take a full state vector x and return Qx in state space.

Given \(x\) compute \(Qx\)

Parameters

[in]	nrhs	the number of right hand sides
[in]	x	the input vector
[out]	qx	the resulting vector where Qx \(= Qx\)

Definition at line 131 of file model_specific.f90.

Here is the call graph for this function:

Here is the caller graph for this function:

subroutine qhalf	(	integer, intent(in)	nrhs,
		real(kind=rk), dimension(state_dim,nrhs), intent(in)	x,
		real(kind=rk), dimension(state_dim,nrhs), intent(out)	Qx
	)

subroutine to take a full state vector x and return \(Q^{1/2}x\) in state space.

Given \(x\) compute \(Q^{\frac{1}{2}}x\)

Parameters

[in]	nrhs	the number of right hand sides
[in]	x	the input vector
[out]	qx	the resulting vector where Qx \(= Q^{\frac{1}{2}}x\)

Definition at line 156 of file model_specific.f90.

Here is the caller graph for this function:

subroutine r	(	integer, intent(in)	obsDim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(obsdim,nrhs), intent(in)	y,
		real(kind=rk), dimension(obsdim,nrhs), intent(out)	Ry,
		integer, intent(in)	t
	)

subroutine to take an observation vector x and return Rx in observation space.

Given \(y\) compute \(Ry\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	y	the input vector
[out]	ry	the resulting vectors where Ry \(= Ry\)
[in]	t	the timestep

Definition at line 176 of file model_specific.f90.

Here is the caller graph for this function:

subroutine reconfigure_model

(

)

subroutine to reset variables that may change when the observation network changes

Definition at line 58 of file model_specific.f90.

Here is the caller graph for this function:

subroutine rhalf	(	integer, intent(in)	obsDim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(obsdim,nrhs), intent(in)	y,
		real(kind=rk), dimension(obsdim,nrhs), intent(out)	Ry,
		integer, intent(in)	t
	)

subroutine to take an observation vector x and return Rx in observation space.

Given \(y\) compute \(R^{\frac{1}{2}}y\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	y	the input vector
[out]	ry	the resulting vector where Ry \(= R^{\frac{1}{2}}y\)
[in]	t	the timestep

Definition at line 198 of file model_specific.f90.

Here is the caller graph for this function:

subroutine solve_b	(	integer, intent(in)	nrhs,
		real(kind=rk), dimension(state_dim,nrhs), intent(in)	x,
		real(kind=rk), dimension(state_dim,nrhs), intent(out)	v
	)

subroutine to take a state vector x and return v in state space.

Given \(y\) find \(v\) such that \(Bv=x\)

Parameters

[in]	nrhs	the number of right hand sides
[in]	x	input vector
[out]	v	result vector where \(v=B^{-1}x\)

Definition at line 301 of file model_specific.f90.

Here is the caller graph for this function:

subroutine solve_hqht_plus_r	(	integer, intent(in)	obsdim,
		real(kind=rk), dimension(obsdim), intent(in)	y,
		real(kind=rk), dimension(obsdim), intent(out)	v,
		integer, intent(in)	t
	)

subroutine to take an observation vector y and return v in observation space.

Given \(y\) find \(v\) such that \((HQH^T+R)v=y\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	y	the input vector
[out]	v	the result where \(v = (HQH^T+R)^{-1}y\)
[in]	t	the timestep

Definition at line 111 of file model_specific.f90.

Here is the caller graph for this function:

subroutine solve_r	(	integer, intent(in)	obsDim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(obsdim,nrhs), intent(in)	y,
		real(kind=rk), dimension(obsdim,nrhs), intent(out)	v,
		integer, intent(in)	t
	)

subroutine to take an observation vector y and return v in observation space.

Given \(y\) find \(v\) such that \(Rv=y\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	y	input vector
[out]	v	result vector where \(v=R^{-1}y\)
[in]	t	the timestep

Definition at line 68 of file model_specific.f90.

Here is the caller graph for this function:

subroutine solve_rhalf	(	integer, intent(in)	obsdim,
		integer, intent(in)	nrhs,
		real(kind=rk), dimension(obsdim,nrhs), intent(in)	y,
		real(kind=rk), dimension(obsdim,nrhs), intent(out)	v,
		integer, intent(in)	t
	)

subroutine to take an observation vector y and return v in observation space.

Given \(y\) find \(v\) such that \(R^{\frac{1}{2}}v=y\)

Parameters

[in]	obsdim	the dimension of the observations
[in]	nrhs	the number of right hand sides
[in]	y	input vector
[out]	v	result vector where \(v=R^{-\frac{1}{2}}y\)
[in]	t	the timestep

Definition at line 89 of file model_specific.f90.

Here is the caller graph for this function: