public interface MechSystem
Modifier and Type | Interface and Description |
---|---|
static class |
MechSystem.ConstraintInfo
Contains information for a single constraint direction
|
static class |
MechSystem.FrictionInfo
Contains information for a single friction constraint set.
|
Modifier and Type | Field and Description |
---|---|
static int |
COMPUTE_CONTACTS
Flag passed to
updateConstraints()
indicating that contact information should be computed. |
static int |
UPDATE_CONTACTS
Flag passed to
updateConstraints() indicating
that contact information should be updated. |
Modifier and Type | Method and Description |
---|---|
void |
addActivePosImpulse(VectorNd q,
double h,
VectorNd u)
Computes an adjustment to the active positions of the system (stored
in the vector
q ) by applying
a velocity u for time
h . |
void |
addPosJacobian(SparseNumberedBlockMatrix S,
VectorNd f,
double h)
Adds the current force-position Jacobian, scaled by
h , to
the matrix S , which should have been previously created with
a call to buildSolveMatrix() . |
void |
addVelJacobian(SparseNumberedBlockMatrix S,
VectorNd f,
double h)
Adds the current force-velocity Jacobian, scaled by
h , to
the matrix S , which should have been previously created with
a call to buildSolveMatrix() . |
boolean |
buildMassMatrix(SparseBlockMatrix M)
Builds a mass matrix for this system.
|
void |
buildSolveMatrix(SparseNumberedBlockMatrix S)
Builds a solve matrix for this system.
|
void |
getActiveForces(VectorNd f)
Returns the generalized forces acting on all the active components in this
system.
|
void |
getActivePosDerivative(VectorNd dxdt,
double t)
Gets the current value of the position derivative for all active
components.
|
void |
getActivePosState(VectorNd q)
Gets the current position state for all active components.
|
int |
getActivePosStateSize()
Returns the size of the position state for all active components.
|
void |
getActiveVelState(VectorNd u)
Gets the current velocity state for all active components.
|
int |
getActiveVelStateSize()
Returns the size of the velocity state for all active components.
|
int |
getBilateralConstraints(SparseBlockMatrix GT,
VectorNd dg)
Obtains the transpose of the current bilateral constraint matrix G for this
system.
|
void |
getBilateralImpulses(VectorNd lam)
Returns from the system the most recently computed bilateral constraint
impulses.
|
void |
getBilateralInfo(MechSystem.ConstraintInfo[] ginfo)
Obtains information for all the constraint directions returned
by the most recent call to
getBilateralConstraints() . |
void |
getFrictionConstraints(SparseBlockMatrix DT,
MechSystem.FrictionInfo[] finfo)
Obtains the transpose of the current friction constraint matrix D for
this system.
|
void |
getInverseMassMatrix(SparseBlockMatrix Minv,
SparseBlockMatrix M)
Sets
Minv to the inverse of the mass matrix M . |
void |
getMassMatrix(SparseBlockMatrix M,
VectorNd f,
double t)
Sets
M to the current value of the mass matrix for this
system, evaluated at time t . |
void |
getParametricForces(VectorNd f)
Gets the forces associated with parametric components.
|
void |
getParametricPosState(VectorNd q)
Gets the current position state for all parametric components.
|
int |
getParametricPosStateSize()
Returns the size of the position state for all parametric components.
|
void |
getParametricPosTarget(VectorNd q,
double s,
double h)
Obtains the desired target position for all parametric components at a
particular time.
|
void |
getParametricVelState(VectorNd u)
Gets the current velocity state for all parametric components.
|
int |
getParametricVelStateSize()
Returns the size of the velocity state for all parametric components.
|
void |
getParametricVelTarget(VectorNd u,
double s,
double h)
Obtains the desired target velocity for all parametric components at a
particular time.
|
int |
getSolveMatrixType()
Returns information about the solve matrix for this system.
|
int |
getStructureVersion()
Returns the current structure version of this system.
|
void |
getUnilateralConstraints(SparseBlockMatrix NT,
VectorNd dn)
Obtains the transpose of the current unilateral constraint matrix N for this
system.
|
void |
getUnilateralImpulses(VectorNd the)
Returns from the system the most recently computed unilateral constraint
impulses.
|
void |
getUnilateralInfo(MechSystem.ConstraintInfo[] ninfo)
Obtains information for all the constraint directions returned
by the most recent call to
getUnilateralConstraints() . |
int |
maxFrictionConstraintSets()
Returns that maximum number of friction constraint set that may be added by
the method
getFrictionConstraints() . |
void |
mulInverseMass(SparseBlockMatrix M,
VectorNd a,
VectorNd f) |
int |
numActiveComponents()
Returns the current number of active components in this system.
|
int |
numParametricComponents()
Returns the current number of parametric components in this system.
|
void |
setActiveForces(VectorNd f)
Sets the generalized forces acting on all the active components in this
system.
|
void |
setActivePosState(VectorNd q)
Sets the current position state for all active components.
|
void |
setActiveVelState(VectorNd u)
Sets the current velocity state for all active components.
|
void |
setBilateralImpulses(VectorNd lam,
double h)
Supplies to the system the most recently computed bilateral constraint
impulses.
|
void |
setParametricForces(VectorNd f)
Sets the forces associated with parametric components.
|
void |
setParametricPosState(VectorNd q)
Sets the current position state for all parametric components.
|
void |
setParametricVelState(VectorNd u)
Sets the current velocity state for all parametric components.
|
void |
setUnilateralImpulses(VectorNd the,
double h)
Supplies to the system the most recently computed unilateral constraint
impulses.
|
void |
updateConstraints(double t,
StepAdjustment stepAdjust,
int flags)
Updates the constraints associated with this system to be consistent with
the current position and indicated time.
|
void |
updateForces(double t)
Updates all internal forces associated with this system to be
consistent with the current position and velocity and the indicated time
t . |
static final int COMPUTE_CONTACTS
updateConstraints()
indicating that contact information should be computed.static final int UPDATE_CONTACTS
updateConstraints()
indicating
that contact information should be updated. This means that there has
only been an adjustment in position and existing contacts between bodies
should be maintained, though possibly with a modified set of constraints.int getStructureVersion()
int getActivePosStateSize()
getStructureVersion()
remains unchanged.void getActivePosState(VectorNd q)
q
(which will be sized appropriately by the
system).q
- vector in which the state is storedvoid setActivePosState(VectorNd q)
q
, whose size should be greater than
or equal to the value returned by getActivePosStateSize()
.q
- vector supplying the state informationint getActiveVelStateSize()
getStructureVersion()
remains unchanged.void getActiveVelState(VectorNd u)
u
(which will be sized appropriately by the
system).u
- vector in which the state is storedvoid setActiveVelState(VectorNd u)
u
, whose size should be greater than
or equal to the value returned by getActiveVelStateSize()
.u
- vector supplying the state informationint getParametricPosStateSize()
getStructureVersion()
remains unchanged.void getParametricPosTarget(VectorNd q, double s, double h)
q
(which will
be sized appropriately by the system).
The system interpolates between the current parametric position values
and those desired for the end of the time step, using a parameter
s
defined on the interval [0,1]. In particular, specifying
s = 0 yields the current parametric positions, and s = 1 yields the
positions desired for the end of the step. The actual time step size
h
is also provided, as this may be needed by some
interpolation methods.
The interpolation uses the current parametric positions, and possibly
the current parametric velocities as well. Hence this method should be
called before either of these are changed using setParametricPosState()
or setParametricVelState()
.
q
- vector returning the state informations
- specifies time relative to the current time steph
- time step sizevoid getParametricPosState(VectorNd q)
q
(which will be sized appropriately by
the system).q
- vector in which the state is storedvoid setParametricPosState(VectorNd q)
q
, whose size should be greater than
or equal to the value returned by getParametricPosStateSize()
.q
- vector supplying the state informationint getParametricVelStateSize()
getStructureVersion()
remains unchanged.void getParametricVelTarget(VectorNd u, double s, double h)
u
(which will
be sized appropriately by the system).
The system interpolates between the current parametric velocity values
and those desired for the end of the time step, using a parameter
s
defined on the interval [0,1]. In particular, specifying
s = 0 yields the current parametric velocities, and s = 1 yields the
velocities desired for the end of the step. The actual time step size
h
is also provided, as this may be needed by some
interpolation methods.
The interpolation uses the current parametric velocities, and possibly
the current parametric positions as well. Hence this method should be
called before either of these are changed using setParametricPosState()
or setParametricVelState()
.
u
- vector returning the parametric target velocitess
- specifies time relative to the current time steph
- time step sizevoid getParametricVelState(VectorNd u)
u
(which will be sized appropriately by
the system).u
- vector in which state is storedvoid setParametricVelState(VectorNd u)
u
, whose size should be greater than
or equal to the value returned by getParametricVelStateSize()
.u
- vector supplying the state informationvoid setParametricForces(VectorNd f)
f
, whose size should be greater or equal to
the value returned by getParametricVelStateSize()
.f
- vector supplying the force informationvoid getParametricForces(VectorNd f)
f
(which will be sized appropriately by the
system).f
- vector in which to return the force informationvoid getActivePosDerivative(VectorNd dxdt, double t)
dxdt
(which will be
sized appropriately by the system).dxdt
- vector in which the derivative is storedt
- current time valuevoid getActiveForces(VectorNd f)
f
(which will be
sized appropriately by the system).f
- vector in which the forces are storedvoid setActiveForces(VectorNd f)
f
, whose size
should be greater or equal to the value returned by
getActiveVelStateSize()
.f
- vector specifying the forces to be setboolean buildMassMatrix(SparseBlockMatrix M)
M
. On input,
M
should be empty with zero size; it will be sized
appropriately by the system.
This method returns true
if the mass matrix is constant;
i.e., does not vary with time. It does not place actual values in the
matrix; that must be done by calling getMassMatrix()
.
A new mass matrix should be built whenever the system's structure
version (as returned by getStructureVersion()
) changes.
M
- matrix in which the mass matrix will be builtvoid getMassMatrix(SparseBlockMatrix M, VectorNd f, double t)
M
to the current value of the mass matrix for this
system, evaluated at time t
. M
should
have been previously created with a call to
buildMassMatrix()
. The current mass forces
are returned in the vector f
, which should have a
size greater or equal to the size of M
.
The mass forces (also known as the fictitious forces)
are given by
- dM/dt uwhere
u
is the current system velocity.
f
will be sized appropriately by the system.M
- returns the mass matrixf
- returns the mass forcest
- current timevoid getInverseMassMatrix(SparseBlockMatrix Minv, SparseBlockMatrix M)
Minv
to the inverse of the mass matrix M
.
Minv
should have been previously created with a call to
buildMassMatrix()
.
This method assumes that M
is block diagonal and hence
Minv
has the same block structure. Although it is possible
to compute Minv
by simply inverting each block of
M
, the special structure of each mass block may permit
the system to do this in a highly optimized way.
Minv
- returns the inverted mass matrixM
- mass matrix to invertvoid mulInverseMass(SparseBlockMatrix M, VectorNd a, VectorNd f)
void buildSolveMatrix(SparseNumberedBlockMatrix S)
S
. On input,
S
should be empty with zero size; it will be sized
appropriately by the system. The resulting matrix should have all the
blocks required to store any combination of the mass matrix, the
force-position Jacobian, and the force-velocity Jacobian.
This method does not place actual values in the matrix; that must be done
by adding a mass matrix to it, or by calling addVelJacobian()
or
addPosJacobian()
.
A new solve matrix should be built whenever the system's structure
version (as returned by getStructureVersion()
) changes.
S
- matrix in which the solve matrix will be builtint getSolveMatrixType()
SYMMETRIC
or POSITIVE_DEFINITE
,
which aid in determining the best way to solve the matrix.int numActiveComponents()
getStructureVersion()
remains unchanged.int numParametricComponents()
getStructureVersion()
remains unchanged.void addVelJacobian(SparseNumberedBlockMatrix S, VectorNd f, double h)
h
, to
the matrix S
, which should have been previously created with
a call to buildSolveMatrix()
.
Addition fictitious forces associated
with the Jacobian can be optionally returned in the vector
f
, which will be sized appropriately by the system.S
- matrix to which scaled Jacobian is to be addedf
- if non-null, returns fictitious forces associated with the Jacobianh
- scale factor for the Jacobianvoid addPosJacobian(SparseNumberedBlockMatrix S, VectorNd f, double h)
h
, to
the matrix S
, which should have been previously created with
a call to buildSolveMatrix()
. Addition
fictitious forces associated with the Jacobian can be optionally returned
in the vector f
, which will be sized appropriately by the
system.S
- matrix to which scaled Jacobian is to be addedf
- if non-null, returns fictitious forces associated with the Jacobianh
- scale factor for the Jacobianint getBilateralConstraints(SparseBlockMatrix GT, VectorNd dg)
GT
. On input,
GT
should be empty with zero size; it will be sized
appropriately by the system. The derivative term is returned
in dg
; this is given by
dG/dt uwhere
u
is the current system velocity.
dg
will also be sized appropriately by the system.
The method returns a version number for the constraint structure. This number should be incremented whenever the block structure of the constraint matrix changes.
GT
- returns the transpose of Gdg
- returns the derivative term for Gvoid getBilateralInfo(MechSystem.ConstraintInfo[] ginfo)
getBilateralConstraints()
.
The information is returned in the array ginfo
,
which should contain preallocated
ConstraintInfo
structures and should have a length greater or equal to
the column size of GT
returned by
getBilateralConstraints()
.ginfo
- Array of
ConstraintInfo
objects used to return the constraint information.void setBilateralImpulses(VectorNd lam, double h)
lam
, which
should have a size greater or equal to the column size of
GT
returned by getBilateralConstraints()
.lam
- Bilateral constraint impulses being supplied to the system.h
- Time interval associated with the impulses. Dividing by this
should give the average constraint forces.void getBilateralImpulses(VectorNd lam)
lam
, which should
have a size greater or equal to the column size of GT
returned by getBilateralConstraints()
.
For constraints which
where present in the previous solve step, the impulse values should equal
those which were set by the prevous call to setBilateralImpulses()
.
Otherwise, values can be estimated from
previous impulse values (where appropriate), or supplied as 0.lam
- Bilateral constraint impulses being returned from the system.void getUnilateralConstraints(SparseBlockMatrix NT, VectorNd dn)
NT
. On input,
NT
should be empty with zero size; it will be sized
appropriately by the system. The derivative term is returned
in dn
; this is given by
dn/dt uwhere
u
is the current system velocity.
dn
will also be sized appropriately by the system.NT
- returns the transpose of Ndn
- returns the derivative term for Nvoid getUnilateralInfo(MechSystem.ConstraintInfo[] ninfo)
getUnilateralConstraints()
.
The information is returned in the array ninfo
,
which should contain preallocated
ConstraintInfo
structures and should have a length greater or equal to
the column size of NT
returned by
getUnilateralConstraints()
.ninfo
- Array of
ConstraintInfo
objects used to return the constraint information.void setUnilateralImpulses(VectorNd the, double h)
the
, which
should have a size greater or equal to the column size of
NT
returned by getUnilateralConstraints()
.the
- Unilateral constraint impulses being supplied to the system.h
- Time interval associated with the impulses. Dividing by this
should give the average constraint forces.void getUnilateralImpulses(VectorNd the)
the
, which should
have a size greater or equal to the column size of NT
returned by getUnilateralConstraints()
.
For constraints which where present in the previous solve step, the
impulse values should equal those which were set by the prevous call to
setUnilateralImpulses()
. Otherwise,
values can be estimated from previous impulse values (where appropriate),
or supplied as 0.the
- Unilateral constraint impulses being returned from the system.int maxFrictionConstraintSets()
getFrictionConstraints()
.
This is used to size the finfo
array supplied to that
method.void getFrictionConstraints(SparseBlockMatrix DT, MechSystem.FrictionInfo[] finfo)
DT
. On input,
DT
should be empty with zero size; it will be sized
appropriately by the system.
Each column block in DT
describes a friction constraint
set associated with one unilateral or bilateral constraint. Information
about each friction constraint set is returned in the array
finfo
, which should contain preallocated
FrictionInfo
structures and should have a length greater or
equal to the value returned by
FrictionConstraintSets()
.
DT
- returns the transpose of Dfinfo
- returns information for each friction constraint setvoid addActivePosImpulse(VectorNd q, double h, VectorNd u)
q
) by applying
a velocity u
for time
h
. Where velocity equals position derivative,
this corresponds to computing
q += h u.In other situations, such as where position is orientation expressed as a quaternion and velocity is angular velocity, the system should perform the analagous computation.
q
should have a size greater or equal to the value
returned by getActivePosStateSize()
, and
u
should have a size greater or equal to the value
returned by getActiveVelStateSize()
.q
- positions to be adjustedh
- length of time to apply velocityu
- velocity to be appliedvoid updateConstraints(double t, StepAdjustment stepAdjust, int flags)
getBilateralConstraints()
.
Because contact computations are expensive, the constraints associated
with contact should only be computed if the flags COMPUTE_CONTACTS
or UPDATE_CONTACTS
are specified. The former
calls for contact information to be computed from scratch, while the
latter calls for contact information to be modified to reflect changes in
body positions, while preserving the general contact state between
bodies.
In the process of updating the constraints, the system may determine
that a smaller step size is needed. This is particulary true when
contact calculations show an unacceptable level of interpenetration.
A smaller step size can be recommended If so, it can indicate this through
the optional argument stepAdjust
, if present.
t
- current timestepAdjust
- (optional) can be used to indicate whether the current advance
should be redone with a smaller step size.flags
- information flagsvoid updateForces(double t)
t
. This method should be called between any change to
position or velocity values (such as through
setActivePosState()
or
setActiveVelState()
,
and any call to getActiveForces()
.
In the process of updating the forces, the system may determine that a
smaller step size is needed. If so, it can indicate this through the
argument optional stepAdjust
, if present.
t
- current time