Could the cosmological constant be due to vacuum fluctuations in a box, i.e., in a finite universe? Assumption: If the universe were a finite box whose boundary is the cosmological horizon, then there would be a zero-point energy inside that box.
Consequence 1: This zero-point energy would be given by the size of the box. The calculated energy value is very similar to the measured cosmological constant. 
Consequence 2: The zero-point energy would have been larger when the universe was smaller. The cosmological constant would not be a constant, but decay in time. 
Question: Could that be the case?
 A: If quantum fields are restricted to a finite box of dimension $L$, then this changes the computation of the zero point energy because the fluctuations exist at discrete frequencies, 
$$\omega_n = \frac{2\pi n}{L}$$
Following Ford (in the case of a 1-dimensional 'box') in https://journals.aps.org/prd/pdf/10.1103/PhysRevD.11.3370, the zero-point energy with a frequency cutoff is
$$E_0 = \sum_n \omega_n \exp(-\alpha \omega_n)= \frac{L}{2\pi \alpha^2} -\frac{\pi}{6L} + {\rm positive\, powers\, of\,} \alpha$$
Following the Casimir prescription, the idea is to subtract the vacuum density appropriate to the unconstrained topology from this energy (which includes it),
$$\tilde{E}_0 = E_0 - E_0(L \rightarrow \infty) = -\frac{\pi}{6L}$$
Note that the energy is inversely related to the dimension of the box, and is negative.  
Ford goes on to explore the more realistic case of a closed, expanding universe and finds that the zero-point energy is infinite, even after subtracting the flat-space energy momentum tensor.
A: The cosmological constant $\lambda$ is the observed vacuum energy density. 
The zero point energy predicted by Quantum Field Theory is much large than $\lambda$ - it can be $120$ orders of magnitude larger depending on the assumptions. 
