|
SuperDex Physics C++ API
|
[Experimental] Evaluation settings common to the full scene. More...
#include <eval_params.h>
Public Member Functions | |
| bool | operator== (ExperimentalEvalParams const &) const =default |
Public Attributes | |
| bool | consistencyResNorm = false |
| Validate that the residual is the gradient of the objective by a directional finite-difference consistency check. | |
| real | consistencyResNormStep = 1e-4_r |
| Step size for finite-difference consistency check. | |
| bool | explicitNormals = false |
| Use explicit normals (from stage-start kinematics) for the evaluation of friction. | |
| bool | fadeFriction = true |
| Fade friction coefficient based on normal alignment. | |
| SaturationHessianParams | fittedSaturationHessian |
| Controls whether force-saturation terms use a fitted quadratic Hessian (more stable) or the exact analytical Hessian (faster convergence but less stable), independently for each saturation pathway. | |
| CoulombFrictionModel | frictionModel = CoulombFrictionModel::Default |
| Selects which Coulomb friction smoothing model to use. | |
| bool | implicitNormalForceForDissipation = false |
| Treat the normal contact force implicitly (if true) or explicitly (if false) for dissipative contact terms. | |
[Experimental] Evaluation settings common to the full scene.
They tune the evaluation of internal models (contact, constraints, etc).
Definition at line 73 of file eval_params.h.
|
default |
| bool superdex::ExperimentalEvalParams::consistencyResNorm = false |
Validate that the residual is the gradient of the objective by a directional finite-difference consistency check.
Definition at line 145 of file eval_params.h.
| real superdex::ExperimentalEvalParams::consistencyResNormStep = 1e-4_r |
Step size for finite-difference consistency check.
Definition at line 150 of file eval_params.h.
| bool superdex::ExperimentalEvalParams::explicitNormals = false |
Use explicit normals (from stage-start kinematics) for the evaluation of friction.
Whether to treat the colliding and collider normals explicitly (using stage-start kinematics) or implicitly (using current kinematics) for the evaluation of alignment and the friction plane.
Definition at line 87 of file eval_params.h.
| bool superdex::ExperimentalEvalParams::fadeFriction = true |
Fade friction coefficient based on normal alignment.
When enabled, friction is scaled by (maxAlignmentNormals - alignment) / (maxAlignmentNormals + 1), where "alignment" is defined as the dot product between colliding and collider normals.
Definition at line 103 of file eval_params.h.
| SaturationHessianParams superdex::ExperimentalEvalParams::fittedSaturationHessian |
Controls whether force-saturation terms use a fitted quadratic Hessian (more stable) or the exact analytical Hessian (faster convergence but less stable), independently for each saturation pathway.
Definition at line 129 of file eval_params.h.
| CoulombFrictionModel superdex::ExperimentalEvalParams::frictionModel = CoulombFrictionModel::Default |
Selects which Coulomb friction smoothing model to use.
C1Regularized (default) preserves existing behavior exactly. CinfRegularized is an additional model that can be selected at runtime without affecting C1Regularized users.
Definition at line 137 of file eval_params.h.
| bool superdex::ExperimentalEvalParams::implicitNormalForceForDissipation = false |
Treat the normal contact force implicitly (if true) or explicitly (if false) for dissipative contact terms.
Definition at line 119 of file eval_params.h.