How to predict the temperature for new points?

Assume that we are going to predict the temperature change of a specific point in a liquid over time. We have the following as our observation:

We have the information about 200 different points in this liquid. Each point has a coordinate(for example $x,y,z$) and we know the temperature change of this point from $t = 0s$ to $t = 100s$. Given a new point $P = (x_0, y_0, z_0)$. How can we predict its temperature change over this period of time?( I mean from $t = 0s$ to $t = 100s$)

Assume that all the points in the liquid have the same starting temperature at $t = 0$

I thought maybe we can use RNNs to do this, but I am not sure how. Does any open-source implementation of this exist?