Orodruin said:
Given that nature does particle collisions spontaneously all the time by colliding cosmic rays with differents stuff, the number should be huge. However, it is not really a question that has a meaningful answer, there is no way of checking the answer experimentally.
Yea of little faith and creativity. We know that the number is greater than zero, and that it is finite because the age and expanse of the universe is finite. From there, it is simply a matter of improving precision.
We may not have a way of directly checking the answer experimentally, but we can check any of the assumptions that went into our answer experimentally.
For example, inputs in the calculation would include the age of the universe, the size of the universe, the amount of Standard Model particles in the universe (all of which are known to one or two significant digits at least), the mean lifetime of the Higgs boson, and a few key assumptions about Higgs boson production in stellar matter which can be estimated from the very solid ground of Standard Model physics and mainstream theory on stellar structure.
The biggest uncertainties would probably involve questions regarding GUT scale physics, inflation, dark matter interactions with the Higgs, and dark energy interactions with the Higgs. You could get much more accurate if you truncated the first 2% of the age of the universe after the Big Bang, because that is where it is hardest to estimate. If you wanted to estimate the average number of Higgs bosons in existence in the universe over a one year period at any time in the last 13.5 billion years, you could greatly increase your precision. For many purposes, knowing the current average number of Higgs bosons per year per sky radius of 1 degree, for example, might be a more useful number anyway.
Also, for many purposes, the reason you would want to know would be to set thresholds, for example, on the maximum flux of cosmic rays produced by Higgs boson decays in a particular detector so that this background could be compared to the signal of choice and ruled out (or not). In this case, one need only set an order of magnitude upper limit to be useful and don't have to worry about a lower bound, so you could cheat and make gross approximations in cases when your threshold isn't very sensitive to one or more of your assumptions (usually just a few assumptions will dominate the overall calculation).