Both of the approaches you are describing are reductionist, as in both cases, you are assuming that statistical physics (SP) reduces to particle physics. Given this, you are really asking two separate questions.
Question: What is the name of these two approaches in epistemology?
On the difference between the top-down approach (from SP to PP) vs the bottom-up approach (from PP to SP), what is at play here is not whether one approach is reductionist and the other isn't, as I mentioned above, they are both reductionist.
Instead what is at play here is the logic of discovery used:
- The top-down case, starting with SP and deriving PP, is an A Priori, Rationalist approach to discovery: In this approach, your starting point is a pre-existing theory of statistical physics, and you are making some assumptions about particle physics. The challenge is then to reconcile these pre-existing theories by trying to find the bridge laws (or bridge principles) that tie the two theories together. It is a priori (Latin for beforehand) because you are assuming that statistical physics is true before hand, now you just seek to confirm it. For the same reason it is rationalist, in the sense that assumes the truth of a fundamental idea that is independent of reason, namely that statistical physics must be in some way reconciled with particle physics, and the challenge lies in confirming this idea experimentally. This approach to scientific discovery was supported, among others, by William Whewell, see the SEP and Wikipedia articles on him for more on this approach.
- The bottom-up approach you describe is closer to an Empiricist approach to scientific discovery: An empiricist would try to make no assumptions at all, or as little as possible about theories, and instead would be guided by only the experimental data that is available. The bottom up approach you describe is more empiricist because in this case the scientists makes no assumptions about what his/her statistical physics theory should look like. Instead, having confirmed the basic laws of particle physics, they would try to derive the laws of statistical physics strictly through logic and experimental verification. It is essentially an a posteriori (Latin for after the fact) method.
To summarize, the top-down approach would be a priori and closer to rationalist and idealist conceptions of scientific discovery, and the bottom-up approach would be a posteriori, and closer to empiricist conceptions of scientific discovery.
What is the opposite of the reductionist approach?
The opposite of reductionism is emergentism or emergence, the view that some physical and natural properties can never be reduces to more fundamental laws. For example, some hold the view that psychology can never be reduced to neuro-science, or that biology can never be reduced to chemistry. This is independent of the direction of the explanation one is trying to develop: Whether you start from the laws of chemistry and then try to integrate them into laws of biology, or whether you start with the laws of biology and try to break them down to the laws of chemistry doesn't matter. An emergentist believes that you will never be able to explain one in terms of the other. The whole is different from the sum of the parts.