Release 0.7.0

[joseph-gergaud]

1. In OptimalControlSolution type, there is not the value of the variable v ;

2. The plot of the solution of the problem

n=1
m=1
t0=0
tf=1
x0=[0]
xf=[0]
ocp = Model()
state!(ocp, n)
control!(ocp, m)
time!(ocp, [t0, tf])
constraint!(ocp, :initial, x0, :initial_constraint)
#constraint!(ocp, :final, xf, :final_constraint)
constraint!(ocp, :control, -1, 1, :control_constraint)
dynamics!(ocp, (x, u) ->  -x[1] + u[1])
#objective!(ocp, :lagrange, (x, u) -> u[1]*u[1])
objective!(ocp, :mayer, (x0,xf) -> xf)
sol = solve(ocp, grid_size=20, print_level=5)
p1 = plot(sol)

the plot of the control and of the costate is not good (value of -1)

3. It will be better that CTBase as before five function with only vectors as argument. When we'll code the direct shooting method, we'll have to treat the different case scalar, or vector as for de Direct method ...

[ocots]

1. Ok. I will make the update.
2. Did you take a look at sol.control and sol.costate?
3. I guess it is more efficient to give directly a scalar instead of adding some tests inside the functions. It shouldn't be difficult to treat this inside the solvers. If I want to write something efficient for instance for the dynamics, I should write 48 methods instead of the 7. Besides, I should know the dimensions of x, u and v at the creation of the function which is not necessarily what we want.

Remark. You should write x0 = 0 and -x+u in the dynamics.

[ocots]

I have added the field variable: https://github.com/control-toolbox/CTBase.jl/releases/tag/v0.7.1

[jbcaillau]

@joseph-gergaud I don't get what you mean point 3, but apparently @ocots does! Anyway, call me (phone) this Tuesday if you want to discuss this.

[jbcaillau]

@ocots issue resolved?

[ocots]

Not yet, but no answer from @joseph-gergaud

[jbcaillau]

@joseph-gergaud please check whether the problem still exists

2. Did you take a look at sol.control and sol.costate?

[ocots]

Problem solved by #149