refactor: make RNG independent of algorithm order

[andiwand]

after discussion with paul I though it would be beneficial to remove the dependence of the algorithm order of the RNG in the examples framework. changing the order of reader/writer/algorithms used to result in different outputs which might be unexpected and hard to debug

I believe the impact on the randomness in the simulation is negligible and should not result in funny correlations

[codecov]

Codecov Report

Merging #1905 (6ef3324) into main (3847c73) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main    #1905   +/-   ##
=======================================
  Coverage   49.44%   49.44%           
=======================================
  Files         408      408           
  Lines       22704    22704           
  Branches    10371    10371           
=======================================
  Hits        11226    11226           
  Misses       4266     4266           
  Partials     7212     7212           

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[paulgessinger]

/cc @HadrienG2

Do we think it's a problem if all algorithms get the same random number sequence?

[HadrienG2]

That would result in correlations, which would be directly visible if two copies of the same (or similar) algorithms are run, but I don't think we currently do that often. On different algorithms, I expect the effect to be more subtle.

[github-actions]

📊 Physics performance monitoring for 6ef3324

Full report
Seeding: seeded, truth estimated, orthogonal
CKF: seeded, truth smeared, truth estimated, orthogonal
IVF: seeded, truth smeared, truth estimated, orthogonal
Ambiguity resolution: seeded, orthogonal
Truth tracking
Truth tracking (GSF)

Vertexing

Vertexing vs. mu

IVF seeded

IVF truth_smeared

IVF truth_estimated

IVF orthogonal

Seeding

Seeding seeded

Seeding truth_estimated

Seeding orthogonal

CKF

CKF seeded

CKF truth_smeared

CKF truth_estimated

CKF orthogonal

Ambiguity resolution

seeded

Truth tracking (Kalman Filter)

Truth tracking

Truth tracking (GSF)

Truth tracking

[paulgessinger]

I think we should probably be ok.
@andiwand do you want to update the changed references?

[andiwand]

will do 👍 at the same time we should then get a 1:1 in your branch if we merge this first

[paulgessinger]

Docker error?

[andiwand]

something is off with the CI and I just got more conflicts with main 😄

[paulgessinger]

Ok I think the CI is back now.

[paulgessinger]

test_full_chain_odd_example_pythia_geant4 still fails.

[benjaminhuth]

Just one thought: Would it be useful to make this optional and off by default?
In case someone really wants to run actually two simulations for some reason...

[andiwand]

Just one thought: Would it be useful to make this optional and off by default? In case someone really wants to run actually two simulations for some reason...

I think it would be easier to just pass in two different RNG to the algorithms in this case

@benjaminhuth digging a little bit into the code this seems to be possible. you can give each algorithm a different seed

[andiwand]

test_full_chain_odd_example_pythia_geant4 still fails.

oh no... we have another propagation error with the changed seed

10:21:12    Propagator     ERROR     Propagation reached the step count limit of 10000 (did 10000 steps)
10:21:12    CKF            ERROR     Propapation failed: PropagatorError:3 Propagation reached the configured maximum number of steps with the initial parameters: 
 8.31541
-26.5624
-2.46935
0.751848
-1.74987
  18.233
10:21:12    TrackFinding   WARNING   Track finding failed for seed 10638 with errorPropagatorError:3

[andiwand]

increasing the iteration limit again seems to fix the problem. lets see if this passes this time

[paulgessinger]

lgtm