I used flux-simulator to generate 1,000,000 pair-e...

I used flux-simulator to generate 1,000,000 pair-end reads of length 75bp. However, I realized that the when I map the reads back onto the reference genome(excluding reads that are mapped to multiple locations), the reads distribution are kind of sparse.

For example, in a gene A of length 1000 nt, and I partition it into 4 bins of 250 base-pair long. What I observe was:
Bin 1 : 5 reads
Bin 2 : 0 read
Bin 3 : 1 read
Bin 4 : 5 reads

As I was expecting at least a fews reads in every bins..

My question is do I need a larger number of reads to generate a better coverage(distribution) of the reads onto genes?

Also, does the number of molecules parameter causes this observation?

The attached is my par file.

Thank you.