solve_ivp problem (scipy 1.0.0)

[AB]

Hello

I don't know why but my answer got lost somehow, so I'm sending it once 
more, hopefully with better luck.

W dniu 2018-02-19 o 11:02, Thomas Jollans pisze:
> On 2018-02-18 14:39, A.Brozi wrote:
>> Hello
>>
>> In new "scipy" (1.0.0) I've found procedure "solve_ivp", which makes it
>> possible to use "events" in order to terminate the ode solution when
>> some condition is satisfied.
>> The precise moment of termination (the value of independent variable) is
>> to be found in "t_events", but I cannot find the dependent variable(s)
>> value(s) at the termination moment.
>> Am I missing something? Or it's simply not possible (hopefully 'not yet')?
>>
>> Regards
>> Andrzej Brozi
> 
> Isn't the solution passed back as ‘y’?

Yes, it is. But the vector "y" contains only values corresponding to the 
time moments given in "t_eval". It doesn't contain the value 
corresponding to the time of event function passing zero.
In this situation a trajectory of a projectile hitting ground will 
terminate at a point above ground.
I've succeeded to find one walk-around for this problem: if the "t_eval" 
vector is not supplied, then the last element of the solution "y" will 
be the desired value.
Using this method it may be necessary to supply some reasonably small 
value of "max_step" (otherwise the solution may contain too few points). 
Unfortunately I don't see a simple way to guarantee the solution not to 
contain too many points (without "t_eval" it will contain all the points 
computed).

Regards
Andrzej Brozi