ensembles . Stewart

A new scheme is proposed which will permit electron spin resonance pulse techniques to be used to realize a quantum computer with a 100 qbits, or more. The computation is performed on effective pure states which correspond to off-diagonal blocks of the density matrix. Described is a scheme which very efficiently performs the preparation stage and which permits " pseudo-projective measurement " to be made on the output. With such measurements all members of the ensemble remain coherent. The demonstration by Shor (1) that a quantum computer can factor large numbers in a time which goes as a polynomial rather than the exponential of the number of digits has electrified interest in quantum computers. In a quantum calculation an initial state |0 >= |0. .. 0 > is acted upon by a unitary transformation U which reflects the calculation and which gives a result |R >= U |0 >. In order that the calculation make sense |0 > must be a perfectly specified state and it must be possible to determine |R >. The designer of a quantum computer is immediately faced by three non-trivial problems. First, since |0 > is some particular state, the entropy is zero and the system must be cooled to absolute zero. Second, the fundamental tenants of quantum mechanics insist a given quantum state |R > cannot be determined accurately by a single measurement. Repeated calculation cycles are necessary even before concerns of the signal to noise are addressed. Third, it is necessary to read the individual qbits. Cirac and Zoller (2) have suggested a physical realization of a quantum computer based on linear ion traps, however even supporters (3) of such devices as a proof-of-principle for quantum computing concede that these will never make a useful computer. Ingenious schemes for performing computations using standard liquid room temperature nuclear magnetic resonance (NMR) techniques has been introduced by Gershenfeld and Chuang (4) and Cory et al (5). Again there is a serious scaling problem and it is difficult to imagine 1