Post
Conversation
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
This wooden bear puzzle shows your parents how their love created a beautiful family. Each bear custom engraved with names - a gift they'll treasure forever 
Shop now 
clapandhug.com/bear-family
Rate proposed Community Notes
I haven’t read the paper yet. I’m curious to see how it works empirically, there are a lot of different kinds of graphs. I’m guessing it’s got good (and neutral) performance relative to old algos on certain kinds of graphs at certain scales.
Interest experiment: current LLMs don’t have this result in their training data, are any of them capable of independently discovering it?
A major result, sure. But it’s been proven Dijkstra’s algorithm is the optimal general method to produce a complete vertex order for arbitrary graphs. Shortest path for random vertices is a more applicable problem, in a sense, but inherently prey to overfit algorithms. Planarity
Resource constrained environments often produce more innovative solutions. The boundaries force optimization rather than brute force approaches.
Win a 610hp Lamborghini Huracán Sterrato—only 1,499 exist and this is the last naturally-aspirated V10 Lambo ever made. Or choose a 558hp Hennessey VelociRaptor that hauls family and embarrasses sports cars.
Get 4x Entries on the Limited Edition Sweepstakes Exclusives.
Cheaper deliveries? 
assuming the companies using the algorithm won’t just stash the delta
assuming the companies using the algorithm won’t just stash the delta 
It is very much used! Even the microchips powering your phone were designed using Dijkstra’s algorithm (for planning wires inside the chip)
Always blows my mind seeing CS research directly enable real-world efficiencies like smoother deliveries. Huge kudos to the team!
True - but navigation, network routing, energy optimization, and chip wiring all use Dijkstra’s today, so faster algorithms can still bring real benefits!
This result is both exciting and disappointing to me.
It's disappointing because it's just combining 2 known solutions, not something entirely new.
It's exciting because it manages to decide when to switch with a tiny compute budget: less than log(n) per node.
It's useless
The algorithm’s practical impact may be limited by graph density, as a 1997 SIGACT News study on single-source shortest paths suggests memory overhead could negate gains in dense networks
41 years later… Dijkstra finally gets some competition 
huge win for both theory *and* everything with a map
huge win for both theory *and* everything with a map
This is not true. You can sort in log log n, giving you a trivial O(m log log m). Which is faster.
I believe we did linear time shortest path back in the 90s. It was a home assignment in my class.
Can you share the paper? I just solved a problem yesterday using dikstra’s algorithm
This might only get taught in a graduate algorithms course though realistically.
can you explain this too me, especially the O calculation
convert pseudo code in image into functional C code, and explain how it works
I did graph theory for my phd so this result was very exciting to me. I wrote more about it here:
Quote
Nil
@Nil053
Someone found a faster algorithm than Dijkstra's after 69 years. And that's absolutely wild.
Time to drop some graph theory lore.
Dijkstra invented his single-source shortest-path algorithm in 1956, and it is remarkable for how long it has stood as the best deterministic
Show more
This won't lead to faster anything, real graphs are not large enough and the large ones are not sparse enough to profit from this. And revamping CS corriculum removing an elegant algorithm that is easy to reason about and apply to many situations for this wouls be a mistake
par rapport à Dijkstra le gain est de combien en terme très simple ? Imaginons sur un trajet de 100km qui prends 1h.
"less traffic, cheaper deliveries" seems a bit of a stretch. Sow now one can compute route 60x faster? So 1 second compute instead of a minute? You still have to drive same route as before, and drive duration does not change.
explain the new algorithm to me like I'm ten, guve the example and comparison to Dijktra's algorithm
Where are the placeholder functions (BaseCase, FindPivots, D_Initialize, D.Insert, D.Pull, D.BatchPrepend)?
can you explain faster traffic , cheaper delivery , shorter paths? Shorter path algorithm has nothing to do with traffic or delivery cost. there are many other factors that desides the route.
Still no algorithm that caters to multiple flyovers on real road. We need algorithm that support 3d vertex space
But can it be applied in real life use cases? Or would it be one among many algorithms, as has been the case till now?
It only beats Dijkstra asymptotically on sparse graphs, for today's practical speedup, you need advanced data structures.
is ur m the edge, and n the node? dijkstra can be done in E + Vlog(V) with fib heap, when the graph is dense this is faster than E * (log(V))^(2/3). ur thinking about the usual min heap method E * log(V), but fib heap is faster.
Did you pay for electric service in East Lansing between April 2, 2019, and April 30, 2025?
Click here to learn more and file a claim for a cash payment.
For road networks m is usually about 4n, so it collapses to O(n log n) again.
That means nothing of what you mentioned.
Compute is not a bottleneck for any of what you mentioned 
NEW Gummies, crafted for consistency, convenience, and taste.
Lineup: Original, Sour, Lemonade, Tropical and Sleep.
Exclusively at STIIIZY stores near you!
why are they using coding language and also math language interchangeably.
Neither discipline can read this shit.
Faster routes are exciting, but rethinking data structures might yield even greater leaps.
Hey help Dorsa to understand her claims about cheaper and faster don't apply to Dijkstra's algorithms as A* has been already outpacing it in most situations
