Cosmic acceleration is one of the most remarkable cosmological findings of recent years .
Although a dark energy component has usually been invoked as the mechanism for the acceleration , A modification of Friedmann equation from various higher dimensional models provides a feasible alternative .
Cardassian expansion is one of these scenarios , in which the universe is flat , matter ( and radiation ) dominated and accelerating but contains no dark energy component .
This scenario is fully characterized by n , the power index of the so-called Cardassian term in the modified Friedmann equation , and \Omega _ { m } , the matter density parameter of the universe .
In this work , we first consider the constraints on the parameter space from the turnaround redshift , z _ { q = 0 } , at which the universe switches from deceleration to acceleration .
We show that , for every \Omega _ { m } , there exist a unique n _ { peak } ( \Omega _ { m } ) , which makes z _ { q = 0 } reach its maximum value , [ z _ { q = 0 } ] _ { max } = \exp \left [ 1 / ( 2 - 3 n _ { peak } ) \right ] -1 , which is unlinearly inverse to \Omega _ { m } .
If the acceleration happans earlier than z _ { q = 0 } = 0.6 , suggested by Type Ia supernovae measurements , we have \Omega _ { m } < 0.328 
no matter what the power index is , and moreover , for reasonable matter density , \Omega _ { m } \sim 0.3 , it is found n \sim ( -0.45 , 0.25 ) .
We next test this scenario using the Sunyaev-Zeldovich/X-ray data of a sample of 18 galaxy clusters with 0.14 < z < 0.83 compiled by Reese et al .
( 2002 ) .
We determine n and \Omega _ { m } , as well as the Hubble constant H _ { 0 } , using the \chi ^ { 2 } minimization method .
The best fit to the data gives H _ { 0 } = 59.2 { kms } ^ { -1 } { Mpc } ^ { -1 } , n = 0.5 and \Omega _ { m } = \Omega _ { b } ( \Omega _ { b } is the baryonic matter density parameter ) .
However the constraints from the current SZ/X-ray data is weak , though a model with lower matter density is prefered .
A certain range of the model parameters is also consistent with the data .