Contraction hierarchies

The contraction hierarchies (CH) algorithm is a two-phase approach to the shortest path problem consisting of a preprocessing phase and a query phase. As road networks change rather infrequently, more time (seconds to hours) can be used to once precompute some calculations before queries are to be answered. Using this precomputed data many queries can be answered taking very little time (microseconds) each.[1][3] CHs rely on shortcuts to achieve this speedup. A shortcut connects two vertices ${\displaystyle u}$ and ${\displaystyle v}$ not adjacent in the original graph. Its edge weight is the sum of the edge weights on the shortest ${\displaystyle u}$-${\displaystyle v}$ path.

Consider two large cities connected by a highway. Between these two cities, there is a multitude of junctions leading to small villages and suburbs. Most drivers want to get from one city to the other – maybe as part of a larger route – and not take one of the exits on the way. In the graph representing this road layout, each intersection is represented by a node and edges are created between neighboring intersections. To calculate the distance between these two cities, the algorithm has to traverse all the edges along the way, adding up their length. Precomputing this distance once and storing it in an additional edge created between the two large cities will save calculations each time this highway has to be evaluated in a query. This additional edge is called a "shortcut" and has no counterpart in the real world. The contraction hierarchies algorithm has no knowledge about road types but is able to determine which shortcuts have to be created using the graph alone as input.

To find a path from ${\displaystyle s}$ to ${\displaystyle t}$ the algorithm can skip over the grey vertices and use the dashed shortcut instead. This reduces the number of vertices the algorithm has to look at. The edge weight of the shortcut from ${\displaystyle u}$ to ${\displaystyle v}$ is the sum of the edge weights of the shortest ${\displaystyle u}$-${\displaystyle v}$ path.

Preprocessing phase

The CH algorithm relies on shortcuts created in the preprocessing phase to reduce the search space – that is the number of vertices CH has to look at, at query time. To achieve this, iterative vertex contractions are performed. When contracting a vertex ${\displaystyle v}$ it is temporarily removed from the graph ${\displaystyle G}$, and a shortcut is created between each pair ${\displaystyle \{u,w\}}$ of neighboring vertices if the shortest path from ${\textstyle u}$ to ${\textstyle w}$ contains ${\displaystyle v}$.[2] The process of determining if the shortest path between ${\textstyle u}$ and ${\textstyle w}$ contains ${\displaystyle v}$ is called witness search. It can be performed for example by computing a path from ${\displaystyle u}$ to ${\displaystyle w}$ using a forward search using only not yet contracted nodes.[3]

The original graph is the line ${\displaystyle (a,b,c,d,e,f)}$(solid). Dashed edges represent shortcuts, grey arrows show which two edges are combined to form the respective shortcut. Vertices have been drawn to represent the node order in which the vertices are being contracted, bottom-to-top. Contracting vertex ${\displaystyle c}$ introduces a shortcut between ${\displaystyle b}$ and ${\displaystyle d}$ with ${\displaystyle \mathrm {dist} (b,d)=\mathrm {dist} (b,c)+\mathrm {dist} (c,d)}$. Contractions of the vertices ${\displaystyle e}$ and ${\displaystyle d}$ introduce one shortcut respectively. Contractions of ${\displaystyle a}$, ${\displaystyle b}$ and ${\displaystyle f}$ do not introduce any shortcuts and are therefore not shown.

Query phase

In the query phase, a bidirectional search is performed starting from the starting node ${\displaystyle s}$ and the target node ${\displaystyle t}$ on the original graph augmented by the shortcuts created in the preprocessing phase.[2] The most important vertex on the shortest path between ${\displaystyle s}$ and ${\displaystyle t}$ will be either ${\displaystyle s}$ or ${\displaystyle t}$ themselves or more important than both ${\displaystyle s}$ and ${\displaystyle t}$. Therefore the vertex ${\displaystyle u}$ minimizing ${\displaystyle \mathrm {dist} (s,u)+\mathrm {dist} (u,t)}$ is on the shortest ${\displaystyle s-t}$ path in the original graph and ${\displaystyle \mathrm {dist} (s,u)+\mathrm {dist} (u,t)=\mathrm {dist} (s,t)}$ holds.[2] This, in combination with how shortcuts are created, means that both forward and backward search only need to relax edges leading to more important nodes (upwards) in the hierarchy which keeps the search space small.[3] In all up-(down-up)-down paths, the inner (down-up) can be skipped, because a shortcut has been created in the preprocessing stage.

When computing the shortest path from ${\displaystyle s}$ to ${\displaystyle t}$, forward (orange) and backward (blue) search only need to follow edges going upwards in the hierarchy. The found path marked in red and uses one shortcut (dashed).

If the edge weights are changed more often than the network topology, CH can be extended to a three-phase approach by including a customization phase between the preprocessing and query phase. This can be used for example to switch between shortest distance and shortest time or include current traffic information as well as user preferences like avoiding certain types of roads (ferries, highways, ...). In the preprocessing phase, most of the runtime is spent on computing the order in which the nodes are contracted.[3] This sequence of contraction operations in the preprocessing phase can be saved for when they are later needed in the customization phase. Each time the metric is customized, the contractions can then be efficiently applied in the stored order using the custom metric.[2] Additionally, depending on the new edge weights it may be necessary to recompute some shortcuts.[3] For this to work, the contraction order has to be computed using metric-independent nested dissections.[1]

CHs as described above search for a shortest path from one starting node to one target node. This is called one-to-one shortest path and is used for example in car-navigation systems. Other applications include matching GPS traces to road segments and speeding up traffic simulators which have to consider the likely routes taken by all drivers in a network. In route prediction one tries to estimate where a vehicle is likely headed by calculating how well its current and past positions agree with a shortest path from its starting point to any possible target. This can be efficiently done using CHs.[2]

In one-to-many scenarios, a starting node ${\displaystyle s}$and a set of target nodes ${\displaystyle T}$ are given and the distance ${\displaystyle \mathrm {dist} (s,t)}$ for all ${\displaystyle t\in T}$ has to be computed. The most prominent application for one-to-many queries are point-of-interest searches. Typical examples include finding the closest gas station, restaurant or post office using actual travel time instead of geographical distance as metric.[2]

In the many-to-many shortest path scenario, a set of starting nodes ${\displaystyle S}$ and a set of target nodes ${\displaystyle T}$ are given and the distance ${\displaystyle \mathrm {dist} (s_{i},t_{i})}$ for all ${\displaystyle (s_{i},t_{j})\in S\times T}$ has to be computed. This is used for example in logistic applications.[2] CHs can be extended to many-to-many queries in the following manner: First, perform a backward upward search from each ${\displaystyle t_{j}\in T}$. For each vertex ${\displaystyle u}$ scanned during this search, one stores ${\displaystyle \mathrm {dist} (t_{j},u)}$ in a bucket ${\displaystyle \beta (u)}$. Then, one runs a forward upward search from each ${\displaystyle s_{i}\in S}$, checking for each non-empty bucket, whether the route over the corresponding vertex improves any best distance. That is, if ${\displaystyle \mathrm {dist} (s_{i},u)+\mathrm {dist} (u,t_{j})<\mathrm {dist} (s_{i},t_{j})}$ for any ${\displaystyle (s_{i},t_{j})\in S\times T}$.[2][3]

Some applications even require one-to-all computations, i.e., finding the distances from a source vertex ${\displaystyle s}$ to all other vertices in the graph. As Dijkstra’s algorithm visits each edge exactly once and therefore runs in linear time it is theoretically optimal. Dijkstra’s algorithm, however, is hard to parallelize and is not cache-optimal because of its bad locality. CHs can be used for a more cache-optimal implementation. For this, a forward upward search from ${\displaystyle s}$ followed by a downward scan over alle nodes in the shortcut-enriched graph is performed. The later operation scans through memory in a linear fashion, as the nodes are processed in decreasing order of importance and can therefore be placed in memory accordingly.[14] Note, that this is possible because the order in which the nodes are processed in the second phase is independent of the source node ${\displaystyle s}$.[2]

In production, car-navigation systems should be able to compute fastest travel routes using predicted traffic information and display alternative routes. Both can be done using CHs.[2] The former is called routing with time-dependent networks where the travel time of a given edge is no longer constant but rather a function of the time of day when entering the edge. Alternative routes need to be smooth looking, significantly different from the shortest path but not significantly longer.[2]

CHs can be extended to optimize multiple metrics at the same time; this is called multi-criteria route planning. For example one could minimize both travel cost and time. Another example are electric vehicles for which the available battery charge constrains the valid routes as the battery may not run empty.[2]