We perform three-dimensional vertically-stratified local shearing-box ideal MHD simulations of the magnetorotational instability ( MRI ) that include a net vertical magnetic flux , which is characterized by midplane plasma \beta _ { 0 } ( ratio of gas to magnetic pressure ) .
We have considered \beta _ { 0 } = 10 ^ { 2 } , 10 ^ { 3 } and 10 ^ { 4 } and in the first two cases the most unstable linear MRI modes are well resolved in the simulations .
We find that the behavior of the MRI turbulence strongly depends on \beta _ { 0 } : The radial transport of angular momentum increases with net vertical flux , achieving \alpha \sim 0.08 for \beta = 10 ^ { 4 } and \alpha \gtrsim 1.0 for \beta _ { 0 } = 100 , where \alpha is the height-integrated and mass-weighted Shakura-Sunyaev parameter .
A critical value lies at \beta _ { 0 } \sim 10 ^ { 3 } : For \beta _ { 0 } \gtrsim 10 ^ { 3 } , the disk consists of a gas pressure dominated midplane and a magnetically dominated corona .
The turbulent strength increases with net flux , and angular momentum transport is dominated by turbulent fluctuations .
The magnetic dynamo that leads to cyclic flips of large-scale fields still exists , but becomes more sporadic as net flux increases .
For \beta _ { 0 } \lesssim 10 ^ { 3 } , the entire disk becomes magnetic dominated .
The turbulent strength saturates , and the magnetic dynamo is fully quenched .
Stronger large-scale fields are generated with increasing net flux , which dominates angular momentum transport .
A strong outflow is launched from the disk by the magnetocentrifugal mechanism , and the mass flux increases linearly with net vertical flux and shows sign of saturation at \beta _ { 0 } \lesssim 10 ^ { 2 } .
However , the outflow is unlikely to be directly connected to a global wind : for \beta _ { 0 } \gtrsim 10 ^ { 3 } , the large-scale field has no permanent bending direction due to dynamo activities , while for \beta _ { 0 } \lesssim 10 ^ { 3 } , the outflows from the top and bottom sides of the disk bend towards opposite directions , inconsistent with a physical disk wind geometry .
Global simulations are needed to address the fate of the outflow .