Joel Andersson
Joel Andersson
The user guide is overdue for a more systematic overhaul.
The current Jacobian calculation should be refactored to handle large-scale problems more efficiently. This will allow more compact code generation and allow solving larger problems than now is possible (interesting...
Being able to generate C code with embedded calls to FMUs would be very valuable. The exported code can be linked either statically or dynamically to the FMU DLLs. Identified...
Substeps: - [x] Track complementarity constraint violation - [ ] Signed optimality condition violation in output - [ ] Step for decreasing complementarity constraint violation - [ ] Separation of...
Currently, the output time grid is an *option* for the integrator instances. The reason that this is the case is that it's hard to perform sensitivity analysis with respect to...
The "do_evaluation_dance" code for Simulink is probably incorrect for any FMU with quadratures, or whenever `x` has been changed from it's default value. ```c++ if (do_evaluation_dance_) { // Dummy call...
According to the [FMI 3 specification 2.4.7.4. Variable Attributes](https://fmi-standard.org/docs/3.0.2/#_variable_attributes) > Exactly one variable of an FMU must be defined as [independent](https://fmi-standard.org/docs/3.0.2/#independent) FMI 2 does not require this, but it would...
It looks like FunctionInternal::generate_options is being called from many functions, like FunctionInternal::forward or even XFunction::factory. That means that things like "ad_weight" or even "derivative_of" gets passed on derived functions, which...
``` python from casadi import * x = MX.sym('x') print vertsplit(x, [0, 0, 1]) # [MX(0x1), MX(vertsplit(x){1})] ```
For a function `f: x -> y`, the Jacobian function has the signature: `jac_f: x, out_y -> dy/dx`. This signature makes a lot of sense in some situations. For example,...