UMICH-2015: Inflation Summary
what would detection of r mean?
- What if we detect r > 0.001?:
- Does it tell us the energy scale of inflation? Conclusion (from Scott Watson): yes due to unseen NG
- Sets bounds on graviton mass
- We know inflation traversed super-Planckian field space distance (large field) which implies some symmetry to protect against corrections
- What if we find r < 0.001?:
- Exclude Starobinsky/Higgs inflation and monomial -- only models where ns -1 ~ 1/N (not as an accident)
Natural targets / thresholds?
- r: 0.01 (monomial) and 0.003 (Starobinsky) are both natural targets motivating sensitivity threshold of sigma(r) = 0.001. Such sensitivity would disfavor two large classes of inflationary models, or detect a new signal. Note that assumed value would affect strategy.
- ns: ns not equal to 1 is only natural target of which we are aware. This has already been passed, but we can reduce model dependence. This is a parameter that is difficult to improve dramatically beyond current uncertainties.
- Consistency condition: Single-field slow roll with canonical kinetic term leads to nt = -r/8. Almost certainly too small to be seen by CMB-S4. Post-BICEP2 r=0.2 paper by Scott Dodelson provides relevant forecasts. Yes, we need to think big. Is this too big?
- nt: nt > -2 is a target. And we can get at nt > 0.2 due to contribution to Neff.
- running: Both monomial and Starobinsky lead to running too small to see. For CMB-S4 expect roughly sigma(alpha) = 0.002. Hard to improve on with LSS. Want polarization out to high ell.
- Features: Generically expect features from some interesting models, but they could easily be undetectably small. No natural thresholds.
- Three-point function in scalars: Can get fNL of order unity from multi-field models and single-field models w/ non-trivial dynamics. There may be some interesting targets here from factor of ~few reduction in error beyond what Planck has done, since "order unity" might mean as big as five for some models. LSS expected to do better. Combination of CMB-S4 and LSS may probe scale dependence of fNL. No natural target known of here.
- Four-point function: tauNL ~ fNL squared is natural target: out of reach
- Non-Gaussianity in tensors?: no known natural targets. Obviously highly dependent on r.
- Isocurvature: No known threshold
- Topological defects: ? Interaction with isocurvature?
- Primordial magnetic fields: may drive requirements on polarization angle calibration (because makes EB not 0)
Anomalies: leading hyp
Bottom line for survey/ experiment design: r drives design one direction, Mnu and Neff another. All inflation stuff besides r drives things in similar ways as Neff and Mnu, not r. The exception is anomalies. Large-angle polarization may provide key clues for understanding large-scale temperature anomalies.
Delensing and Foregrounds:
- Blake: need high sensitivity lensing map.Plot shows residual delensed B-mode noise on large scales vs. noise of the high-res lens-reconstruction-experiment (right figure from Smith et al. 2010).