Error when reduce time step for integral

[ebe79442114]

my accDT input motion is 0.033 and when I run my model with below command there is no error:

[code]deltaT = $accDt
analyze 2043 $deltaT [expr $deltaT/100] $deltaT 5
[/code]

but when I reduce time step and run my model with below command there is an error:

[code]deltaT = [expr $accDt/2.0] ; # (or even /3.0)
analyze 4086 $deltaT [expr $deltaT/100] $deltaT 5
[/code]
this error occur :

[code]FATAL:second_order_eqn: Complex roots.
when moving active surface.
B2=5.75673e-25 4AC=6.86184e-25
[/code]
I think it's related to my nDMaterial PressureIndependMultiYield material.

Any idea?

please excuse me for my poor english.

[ebe79442114]

When I analyze my model with below command:

accDt = 0.033 ; # dt for input acc
accuracy = 1.0 # number for analyze step related to deltaT/accuracy
deltaT = accDt/accuracy; # time step for analysis

and :
integrator Newmark 0.5 0.25 .4 0 .003 0

the model run correctly and acceleration response is like to experiment,like below pic:

freeimagehosting.net/uploads/bf6daffeb5.jpg

but with below command (accuracy changed to 2):
accDt = 0.033 ; # dt for input acc
accuracy = 2.0 # number for analyze step related to deltaT/accuracy
deltaT = accDt/accuracy; # time step for analysis

and :
integrator Newmark 0.5 0.25 .4 0 .003 0

this error occurred:

FATAL:second_order_eqn: Complex roots.
when moving active surface.
B2=4.8805e-28 4AC=5.4822e-28

and with this below commands model run but the acceleration response is not true (rayleigh coefficient changed):

accDt = 0.033 ; # dt for input acc
accuracy = 2.0 # number for analyze step related to deltaT/accuracy
deltaT = accDt/accuracy; # time step for analysis

and :
integrator Newmark 0.5 0.25 .04 0 .0003 0

[e.roknabadi]

please help me.

why when I reduce time step, the analysis fail?