How does the Maven dependency resolution handle refined/conflicting version ranges?

Question

I have studied the Maven documentation on poms/dependency mechanism and the "Complete Reference" Book to understand the dependency resolution algorithm and I think that I understand the general approach. Like the idea of mediation, management, scope, exclusions, or optional dependencies.

While it is clear to me what happens when explicit hard/soft versions are used with one particular version (like <version>1.2.3</version>), I could not find information how mediation affects version ranges... or in general, how ranges are being resolved across a dependency set. Assuming I have the following dependency definition for A

A
+-B (1.0.0, 2.0.0)
|-C
  +-B (1.5.0, 2.0.0)

In which A defines a dependency on B, in the version range (1.0.0, 2.0.0), and on C, which also depends on B, but in the range (1.5.0, 2.0.0). Which version of B would be used? Does the dependency algorithm try to unite and solve all these constraints and end up with the newest version that lies in the 1.5.0-2.0.0 range, or does again the closer definition win and completely override the definition in the transitive dependency?

Second example, assuming the two ranges would not overlap (e.g., A depends on B (1.0.0, 1.1.0) and C on B (1.2.0, 1.3.0)), I would expect that (given that both are soft constraints) Maven would just pick the newest version of B that matches the version range defined in A. Or would this crash like a violated hard requirement?

The same two questions can also be asked for dependencyManagement sections or "Bill-of-material" imports ("pom dependency"), but I would expect that they follow the same logic.

I would highly appreciate an explanation or any pointers to documentation that sheds light on this.