Additional general questions regarding Chrono-DEME interoperability

[Gianni Curti]

Hi Ruochun,

my questions are:

1) When I build the program there is this warning appearing due to the fact that I am using GNUplot to produce graphs that shows the temporal trends of the object's position and velocity in Chrono. Is this harmful in some way? How can I suppress it?

2) At the point where the Chrono and DEME bodies interact, you find the torque by multiplying the diagonal inertia matrix of the ball for the angular acceleration. In my case, I have an unsymmetrical body, and as you told me I used the functions InformCentroidPrincipal and SetMOI to fully define principal inertia axis direction, principal inertia moments and CoG position. Thus my question is, in this case do I have to calculate tor as shown in line 1652 (where inertia is the vector containing the three principal inertia moments) or follow the process in line 1653 (and so use an inertia matrix which is not diagonal therefore I explicitely performed the product between the matrix and the vector)?

Thank you again for your time.

Gianni

[Ruochun Zhang]

Hi Gianni,

For question 1 I can't comment too much on it, as I never used this functionality. But since it's an unused variable warning, I don't imagine any harm if the logic of the code is correct.

For question 2, you should use the diagonal inertia matrix that is consistent with the one you supplied to DEME. Note that if this object's motion (velocity and acceleration) is completely managed by Chrono instead of DEME, and the only thing you care about is the force and torque, then the inertia matrix does not even have to be "correct" (you can choose to not set it at all, and let it be the default 1.0), it needs only to be consistent.

Thank you,

Ruochun

[Gianni Curti]

Hi Ruochun,

thank you for your answer. 

Another question regards the computational speed. In this case the terrain is completely managed by the the GPU, but regarding the Chrono part is it possible to have any speed boost (maybe by setting the number of threads used by the processor)? It is just a curiosity since maybe it does not even impact that much.

Thank you again,

Gianni

[Ruochun Zhang]

Hi Gianni,

You are probably asking about using Chrono Multicore. It's the best that you start a Multicore-specific thread to let people with experience of that answer this question.

As a rule of thumb, if the Chrono part is about a small number of objects then no; If it's about a moderate number of clearly divisible groups of objects, like several ground vehicles, then yes. I wouldn't dismiss the possibility that the Chrono part of the simulation can have an impact on the simulation, even if the DEM part involves a million times more entities since they are also computationally lightweight and run on GPUs. However, I think it's more reasonable to ensure the Chrono part is implemented correctly and efficiently using appropriate MBD models, before thinking about running it in parallel.

Thank you,

Ruochun

[Gianni Curti]

Hi Ruochun,

yes indeed I was referring to Chrono Multicore. Since I have only a few objects for now I will avoid to use parallel computing as you said.

Thank you again for everything.

Gianni