Talk:Monge Problem: Difference between revisions
Jump to navigation
Jump to search
No edit summary |
|||
Line 2: | Line 2: | ||
* Perhaps replace the first sentence by something that would give the intuitive idea of the Monge problem to a non-mathematician. You can point out that the key difference between the Monge problem and the Kantorovich problem is that Monge doesn't allow mass to be split | * Perhaps replace the first sentence by something that would give the intuitive idea of the Monge problem to a non-mathematician. You can point out that the key difference between the Monge problem and the Kantorovich problem is that Monge doesn't allow mass to be split | ||
* Change the phrase ``Kantorovich Problem`` to be a link to the corresponding page on the OT wiki | * Change the phrase ``Kantorovich Problem`` to be a link to the corresponding page on the OT wiki | ||
==Background and Statement== | |||
* type-o ``the best way to move piles of dirt from its...`` | |||
* ``without separating any individual pile`` is a bit confusing -- perhaps somethign like ``all dirt that starts at the same initial location must be kept together and sent to the same final location``? | |||
* ``a solution to the Monge Problem is a TRANSPORT map that specifies how to rearrange the dirt with minimal cost'' | |||
* ``let X and Y be metric spaces'', not vector spaces | |||
* mu and nu must be probability measures | |||
* ``called a transport map FROM MU TO NU`` | |||
* ``infimum is taken over all transport maps SENDING MU TO NU`` | |||
==Issues with Formulation== | |||
* Replace the first part of the first sentence with``It took many years to prove existence of a solution to the Monge problem due to the constraint of only considering transport maps.`` | |||
* ``existence or uniqueness of AN OPTIMAL TRANSPORT MAP'' | |||
==Examples== | |||
*This section is empty. Please either add an example or remove this section. | |||
==Relation with Kantorovich== | |||
* ``the Kantorovich problem is framed in terms of transport plans, which allow mass starting at the same initial location to be split and sent to various final locations.`` | |||
==References Section== | ==References Section== | ||
* There is a broken reference to Monge's original paper | * There is a broken reference to Monge's original paper |
Revision as of 21:49, 13 May 2020
Introduction
- Perhaps replace the first sentence by something that would give the intuitive idea of the Monge problem to a non-mathematician. You can point out that the key difference between the Monge problem and the Kantorovich problem is that Monge doesn't allow mass to be split
- Change the phrase ``Kantorovich Problem`` to be a link to the corresponding page on the OT wiki
Background and Statement
- type-o ``the best way to move piles of dirt from its...``
- ``without separating any individual pile`` is a bit confusing -- perhaps somethign like ``all dirt that starts at the same initial location must be kept together and sent to the same final location``?
- ``a solution to the Monge Problem is a TRANSPORT map that specifies how to rearrange the dirt with minimal cost
- ``let X and Y be metric spaces, not vector spaces
- mu and nu must be probability measures
- ``called a transport map FROM MU TO NU``
- ``infimum is taken over all transport maps SENDING MU TO NU``
Issues with Formulation
- Replace the first part of the first sentence with``It took many years to prove existence of a solution to the Monge problem due to the constraint of only considering transport maps.``
- ``existence or uniqueness of AN OPTIMAL TRANSPORT MAP
Examples
- This section is empty. Please either add an example or remove this section.
Relation with Kantorovich
- ``the Kantorovich problem is framed in terms of transport plans, which allow mass starting at the same initial location to be split and sent to various final locations.``
References Section
- There is a broken reference to Monge's original paper