The Ultimate Guide To Mauri Stern Essential Techniques And Strategies

Working with Steenrod Algebra Techniques: What Actually Matters

The Steenrod algebra is a foundational tool in algebraic topology, and getting comfortable with its essential operations is something most graduate students struggle with for months. I spent a long time trying to make the Adem relations feel intuitive before realizing they are mostly just bookkeeping rules that you learn to execute mechanically. The real value comes from understanding how the Steenrod squares act on cohomology rings and using that action to distinguish spaces that look identical at the homology level. Here is what I actually did, not what any textbook will tell you to do. Start by computing Steenrod squares on the cohomology of simple spaces like projective spaces and lens spaces until the pattern becomes automatic. The key relation is that Sq^i raises degree by i , and on H*(RP^n; Z/2) , you get Sq^i(x) = binom(n,i) x^{k+i} where x generates the cohomology ring. This single computation opens up everything else because most spaces you care about have cohomology rings built from similar generators. The Adem relations are where people get stuck, so here is the practical approach. You only need to remember that they let you rewrite any composition Sq^a Sq^b as a linear combination of Sq^c Sq^d terms where c >= d whenever a < 2b . The formula itself is tedious to derive but straightforward to apply once you have it memorized. I kept a cheat sheet with the first five or six Adem relations taped to my monitor for about two years before I stopped needing it. At that point I realized the relations follow a consistent pattern and I could reconstruct them if I forgot any specific one.

One thing that catches everyone off guard is the Cartan formula. When you have a product of cohomology classes, Sq^k(xy) = sum_{i+j=k} Sq^i(x) Sq^j(y) . This seems obvious but people routinely forget it when working through spectral sequences. I once computed a cohomology operation on a fiber bundle and got the wrong answer for three days because I kept applying the Steenrod square to only one factor instead of using the Cartan formula properly. The correction took about ten minutes once I traced back through my work and noticed I had silently dropped half the terms. For practical computations, the Serre spectral sequence combined with Steenrod operations is your main workhorse. The transgression interacts with the Steenrod squares in a way that lets you determine differentials that pure homological methods cannot reach. There is a standard technique where you compute the action of Sq^1 or Sq^2 on the E_2 page and use the fact that these operations commute with the differential to constrain what the differential can be. This is not covered well in most first courses on the subject, which is why I am mentioning it directly here. A counter-intuitive point that beginners miss: knowing the Steenrod algebra structure does not automatically tell you the cohomology ring structure of a space. The operations provide extra structure on top of the ring, but they can sometimes give misleading information if you do not also track the cup product carefully. I worked on a problem involving a certain fibration where the Steenrod operations suggested two possible cohomology rings, and only by checking the Pontryagin squares could I rule out one of them. That took about six hours of computation that I wish I had avoided by being more systematic from the start.

When you move to odd primes, the situation changes significantly. The Steenrod reduced p-th powers P^i and the Bockstein beta replace the squares, and the Adem relations take a different form. The formulas are more complex but the same principles apply. I found it helpful to keep the two cases separate in my notes rather than trying to unify them early. Mixing them together just creates confusion because the degree conventions and sign rules differ enough to cause errors. Another practical technique involves the Milnor basis. The standard basis Sq^{i_1,i_2,...,i_k} in terms of admissible sequences is useful for many computations, but the Milnor basis Sq(R) where R is a sequence of non-negative integers has cleaner coproduct formulas. If you are doing anything involving the dual Steenrod algebra or Hopf algebra structures, switch to the Milnor basis immediately. The transition is not hard once you know the change-of-basis formulas, and they save a significant amount of time on spectral sequence calculations. One limitation worth stating plainly: Steenrod operations only give you information about mod p cohomology. If your space has torsion at other primes or integral cohomology that is not visible mod p , these operations will not detect it. In those cases you need to supplement with other tools like K-theory, characteristic classes, or the Atiyah-Hirzebruch spectral sequence. Relying solely on Steenrod operations is a common mistake for students who have just learned them and want to apply them everywhere.

For people trying to build computational fluency, I recommend working through Mosher and Tangora's Cohomology Operations and Applications in Algebraic Topology carefully. The exercises are well-chosen and cover the range of situations you will actually encounter. Skip the parts that feel too abstract on first reading and come back to them later. The book assumes a level of comfort with spectral sequences that not everyone has immediately, so it is okay to pause and review the necessary background before continuing. The bottom line is that these techniques become second nature only after you have made mistakes with them. The computations are mechanical but unforgiving of small errors, and catching those errors requires a solid grasp of what each operation is supposed to do conceptually. If you can explain why Sq^1 is the Bockstein homomorphism associated to the coefficient sequence 0 -> Z/2 -> Z/4 -> Z/2 -> 0 , then most of the rest follows from there.