You are currently browsing the tag archive for the ‘epsilon regularisation’ tag.
Title: Give yourself an epsilon of room.
Quick description: You want to prove some statement about some object (which could be a number, a point, a function, a set, etc.). To do so, pick a small , and first prove a weaker statement (which allows for “losses” which go to zero as ) about some perturbed object . Then, take limits . Provided that the dependency and continuity of the weaker conclusion on are sufficiently controlled, and is converging to in an appropriately strong sense, you will recover the original statement.
One can of course play a similar game when proving a statement about some object , by first proving a weaker statement on some approximation to for some large parameter N, and then send at the end.
General discussion: Here are some typical examples of a target statement , and the approximating statements that would converge to :
|for some independent of|
|is finite||is bounded uniformly in|
|for all (i.e. maximises f)||for all (i.e. nearly maximises f)|
|converges as||fluctuates by at most o(1) for sufficiently large n|
|is a measurable function||is a measurable function converging pointwise to|
|is a continuous function||is an equicontinuous family of functions converging pointwise to OR is continuous and converges (locally) uniformly to|
|The event holds almost surely||The event holds with probability 1-o(1)|
|The statement holds for almost every x||The statement holds for x outside of a set of measure o(1)|
Of course, to justify the convergence of to , it is necessary that converge to (or converge to , etc.) in a suitably strong sense. (But for the purposes of proving just upper bounds, such as , one can often get by with quite weak forms of convergence, thanks to tools such as Fatou’s lemma or the weak closure of the unit ball.) Similarly, we need some continuity (or at least semi-continuity) hypotheses on the functions f, g appearing above.
It is also necessary in many cases that the control on the approximating object is somehow “uniform in “, although for “-closed” conclusions, such as measurability, this is not required. [It is important to note that it is only the final conclusion on that needs to have this uniformity in ; one is permitted to have some intermediate stages in the derivation of that depend on in a non-uniform manner, so long as these non-uniformities cancel out or otherwise disappear at the end of the argument.]
By giving oneself an epsilon of room, one can evade a lot of familiar issues in soft analysis. For instance, by replacing “rough”, “infinite-complexity”, “continuous”, “global”, or otherwise “infinitary” objects with “smooth”, “finite-complexity”, “discrete”, “local”, or otherwise “finitary” approximants , one can finesse most issues regarding the justification of various formal operations (e.g. exchanging limits, sums, derivatives, and integrals). [It is important to be aware, though, that any quantitative measure on how smooth, discrete, finite, etc. should be expected to degrade in the limit , and so one should take extreme caution in using such quantitative measures to derive estimates that are uniform in .] Similarly, issues such as whether the supremum of a function on a set is actually attained by some maximiser become moot if one is willing to settle instead for an almost-maximiser , e.g. one which comes within an epsilon of that supremum M (or which is larger than , if M turns out to be infinite). Last, but not least, one can use the epsilon room to avoid degenerate solutions, for instance by perturbing a non-negative function to be strictly positive, perturbing a non-strictly monotone function to be strictly monotone, and so forth.
To summarise: one can view the epsilon regularisation argument as a “loan” in which one borrows an epsilon here and there in order to be able to ignore soft analysis difficulties, and can temporarily be able to utilise estimates which are non-uniform in epsilon, but at the end of the day one needs to “pay back” the loan by establishing a final “hard analysis” estimate which is uniform in epsilon (or whose error terms decay to zero as epsilon goes to zero).
A variant: It may seem that the epsilon regularisation trick is useless if one is already in “hard analysis” situations when all objects are already “finitary”, and all formal computations easily justified. However, there is an important variant of this trick which applies in this case: namely, instead of sending the epsilon parameter to zero, choose epsilon to be a sufficiently small (but not infinitesimally small) quantity, depending on other parameters in the problem, so that one can eventually neglect various error terms and to obtain a useful bound at the end of the day. (For instance, any result proven using the Szemerédi regularity lemma is likely to be of this type.) Since one is not sending epsilon to zero, not every term in the final bound needs to be uniform in epsilon, though for quantitative applications one still would like the dependencies on such parameters to be as favourable as possible.
Prerequisites: Graduate real analysis. (Actually, this isn’t so much a prerequisite as it is a corequisite: the limiting argument plays a central role in many fundamental results in real analysis.) Some examples also require some exposure to PDE.