Some regression test fails

'Quad-Simple-Test-1' and 'Distribution-Binomial-1' regression tests fails

http://amas.web.psi.ch/opal/regressionTests/master/results_2020-07-09.xml

added Bug label

changed title from regression test fails to Some regression test fails

changed the description

Likely these tests fail due to src!385 (merged) or src!394 (merged)

Thanks for reporting. Most likely src!385 (merged) I would say.

cc: @albajacas_a @kraus

src!394 (merged) seems very unlikely. It seems like the energy is wrong.

However src!385 (merged) is so central that I would expect more tests to fail.

Quad-Simple-Test-1:

Run regression test Quad-Simple-Test-1
Test rms_x(last) failed: 0.00012649639419667642 (eps=2e-08) 
Test rms_y(last) failed: 0.0005071332592516626 (eps=2e-08) 
Test rms_s(last) failed: 0.005897682619635395 (eps=2e-08) 
Test emit_x(last) failed: 0.00016994908743218497 (eps=2e-08) 
Test emit_y(last) failed: 0.00012497973087850295 (eps=2e-08) 
Test emit_s(last) failed: 5.752641555120503e-07 (eps=2e-08) 

Distribution-Binomial-1:

Run regression test Distribution-Binomial-1
Test rms_x(last) failed: nan (eps=1e-12) 
Test rms_y(last) failed: nan (eps=1e-12) 
Test rms_s(last) failed: nan (eps=1e-12) 

For Binomial:

reference:

OPAL{+0}> * Distribution type: BINOMIAL
OPAL{+0}> 
OPAL{+0}> * SIGMAX   = 0.00215 [m]
OPAL{+0}> * SIGMAY   = 0.0115 [m]
OPAL{+0}> * SIGMAZ   = 0.1 [m]
OPAL{+0}> * SIGMAPX  = 4.71216e-08 [Beta Gamma]
OPAL{+0}> * SIGMAPY  = 0.000244303 [Beta Gamma]
OPAL{+0}> * SIGMAPZ  = 0.000159667 [Beta Gamma]

now:

OPAL{0}> * SIGMAX   = 2.150000e-03 [m]
OPAL{0}> * SIGMAY   = 1.150000e-02 [m]
OPAL{0}> * SIGMAZ   = 1.000000e-01 [m]
OPAL{0}> * SIGMAPX  = 1.065789e-15 [Beta Gamma]
OPAL{0}> * SIGMAPY  = 2.984209e-08 [Beta Gamma]
OPAL{0}> * SIGMAPZ  = 1.274684e-08 [Beta Gamma]

I'll investigate more.

p.s. Output seems now in a different format. I like the original better.

So this comes directly from src!385 (merged), where the meaning of SIGMAPX (when INPUTOUNITS= EV) was changed from kinetic energy to momentum. Easiest fix would be to change the SIGMAPs of the input file.

You mean this snippet

double Distribution::converteVToBetaGamma(double valueIneV, double massIneV) {
    double value = std::copysign(std::sqrt(std::pow(std::abs(valueIneV) / massIneV + 1.0, 2) - 1.0), valueIneV);
    if (std::abs(value) < std::numeric_limits<double>::epsilon())
        value = std::copysign(std::sqrt(2 * std::abs(valueIneV) / massIneV), valueIneV);

    return value;
}

which didn't convert from eV to beta gamma.

yes exactly. This is now changed in

value = massIneV / valueIneV

which explains the difference.

I managed to pass the regression tests by updating SIGMAPX, SIGMAPY, SIGMAPZ.

Thanks for looking into it. This fix makes sense to me

@snuverink_j you mean

value = valueIneV / massIneV

right?

@kraus yes

mentioned in issue src#563 (closed)

created merge request !19 (merged) to address this issue

mentioned in merge request !19 (merged)

closed via merge request !19 (merged)

closed via commit d4b06eb4

assigned to @snuverink_j

mentioned in issue #90 (closed)