TASI 2022 lectures on scattering amplitudes – an

PoS(TASI2022)002
addendum

John Joseph M. Carrasco𝑎,∗

𝑎 Department of Physics and Astronomy
Northwestern University, Evanston, IL 60208, USA
E-mail: [email protected]

This is a short addendum to “TASI 2014 Gauge and Gravity Amplitude Relations,” 1506.00974.

Theoretical Advanced Study Institute (TASI2022)

6 June - 1 July, 2022
University of Colorado, Boulder

∗ Speaker

© Copyright owned by the author(s) under the terms of the Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0). https://pos.sissa.it/
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

1. Introduction

Much of the material presented and discussed in my “Lectures on Scattering Amplitudes” at

the 2022 version of TASI1 are written up in some technical detail in my previous TASI lecture notes,
which are on the preprint archives. Instead, for these proceedings in this addendum, I provide a
quick summary and a very brief update that is meant to contain some information not included in my
original lectures, along with more up-to-date references and the context of my evolving perspective
engaging more directly with higher-derivative operator predictions in the context of effective field
theories. The 2014 TASI lectures were aiming towards students of a scattering focused school with

PoS(TASI2022)002
an idea towards introducing graphical methods useful for multi-loop calculations and introducing the
duality between color and kinematics and associated double-copy. In contrast these 2022 lectures
were intended to be more broadly appreciated by phenomenologists with some familiarity with
Quantum Field Theory but no particular exposure to modern scattering methods and approaches.
The primary takeaway of these 2022 lectures is that even without resorting to dimension-
specific variables (which can indeed drastically simplify calculations – see references below on
spinor-helicity) we really can carry out the idealized (perturbative) S-matrix bootstrap program:
namely constraining the predictions of a wide web of theories by symmetries and principles. As we
learn from quantum field theory, the combination of quantum mechanics and Lorentz invariance
is incredibly powerful. When we restrict ourselves to on-shell kinematics there are only so many
building blocks we can write down. In a guided manner this approach can even be an efficient
approach to calculation–indeed the higher-loop calculations in gauge and gravity theories. Along
the way we can appreciate a universality of building blocks that are recycled to appear in the
predictions of many theories. I elide references to primary references in these notes, but do provide
references to handy lectures and reviews in the concluding section.

2. Unitarity

Loop level integrands, especially those involving external gauge vectors or gravitons can be
laborious to compute using standard Feynman rules. Off-shell Feynman rules carry unphysical
gauge freedom that must cancel only upon consideration of gauge-invariant quantities. On-shell
unitarity methods allow an efficient verification of multiloop integrands when the integrands are
organized in terms of functional maps from graphs. Efficient verification lends itself to natural
construction in the following sense.
We can ask verification questions targeting, at first, individual graphs. Such maximally targeted
cuts, each expressed in terms of a distinct cubic graph2, involve kinematics with all all propagators
taken on-shell – maximal cuts. The contribution associated with the cut graph should involve the
sum over states of a product of trees. The tree-labels come from each vertex of the cut graph, and
all physical states in the theory are allowed to cross the cuts. Any non-vanishing maximal cut must
be associated with the integrand of that individual graph. Associating that information with the
kinematic numerator of such graphs involves an integrand that can only be in error proportional to
inverse-propagators. By releasing cut conditions systematically we gradually allow more graphs
1All TASI 2022 video lectures can be found by searching on [YouTube].
2Cubic graphs involve only three-point internal vertices.

2
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

to contribute to each cut, but any missing information can only be proportional to the inverse
propagators whose cut conditions have been released – representing contact terms that can be either
associated with graphs with higher-order vertices, or assigned to cubic graphs by multiplying and
dividing the contact term by the relevant expanded propagators. We can continue releasing such
cut conditions until we have saturated the potential loop-level powercounting of the theory. While
the 2014 lectures emphasized the use of unitarity methods at the multiloop level (indeed beginning
with a 2-loop example), we can straightforwardly apply exactly the same ideas and techniques to
constraining graph-organized tree-level amplitudes via factorization which is what we treated in the
2022 lecture videos. I will leave much of the mechanics of unitarity based cut-construction to my

PoS(TASI2022)002
earlier 2014 lecture notes, but here present my current perspective on the approach and utility of
these on-shell methods when combined with symmetries and principles.
With an eye towards reuse of tree-amplitudes in the service of building multi-loop amplitudes
whose integrals are dimensionally regulated, I stay formally in arbitrary 𝐷 dimensions. This means
that little-group weight are carried by polarization vectors, 𝜀(𝑘) for external vectors of mass 𝑘,
and formal spinors 𝑢(𝑘),
¯ 𝑣(𝑘) for external fermions. How do we sum over states with formal
polarizations and spinors? We simply exploit the following completeness relations as projectors to
sums over physical states:
∑︁ 𝑘 𝜇𝑞𝜈 + 𝑞𝜇 𝑝𝜈
𝜀 𝜇,𝑠 (𝑘)𝜀 𝜈, 𝑠¯ (−𝑘) = 𝜂 𝜇𝜈 − , (1)
𝑠 ∈pols
𝑘·𝑞

and fermionic state sums:

∑︁
¯ 𝑠¯) = /𝑙 + 𝑚 .
𝑣(−𝑙, 𝑠) 𝑢(𝑙, (2)
𝑠 ∈states

where 𝜂 𝜇𝜈 is taken to be the flat spacetime metric, and 𝑞 is an arbitrary reference null-vector
corresponding to freedom of gauge-choice. Any gauge-independent observable such as an ordered
amplitude must therefore ultimately be independent of 𝑞 – so all 𝑞 dependence must cancel out.
When the sewing of trees closes a fermion loop this will result in a trace over the fermionic indices
associated with that loop.
It is worth perhaps seeing such a state-sum on some simple examples. Consider one of the
simplest gauge theories – the covariantized free scalar, i.e. a massive scalar in the adjoint minimally
coupled to Yang-Mills.
1
L = (𝐷𝜙) 2 + 𝑚 2 𝜙2 − Tr(F2 ) (3)
4
To get a feel for the state-sum, let us evaluate the following 𝑠-channel cut with all momenta taken
external:
∑︁
C1 = A (1𝑎𝜙 , 2𝑏𝜙 , 𝑝 𝑒𝐴,𝑠 )A (−𝑝 𝐴𝑒𝑠¯ , 3𝑐𝜙 , 4𝑑𝜙 ) . (4)
𝑠 ∈pols

I use an all out-going convention to specify the momentum labels of each tree so the three-point
amplitude A (1, 2, 𝑝) means that 𝑘 1 + 𝑘 2 + 𝑝 = 0. The subscript of each label specifies the particle
type and thus the on-shell conditions. As such, in C1 , the massive scalar external momenta satisfy
𝑘 12 = 𝑘 22 = 𝑘 32 = 𝑘 42 = 𝑚 2 , and the cut gluon momenta satisfies 𝑝 2 = 0. The superscript index
specifies the color charge of the particle. The color-weight from Feynman rules is straight-forward,

3
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

with each vertex dressed with the anti-symmetric structure constant 𝑓 𝑎𝑏𝑐 ∝ Tr(Ta [Tb , Tc ]) of the
gauge group. Each tree amplitude is given:

A (1𝑎𝜙 , 2𝑏𝜙 , 𝑝 𝐴𝑠 𝑒 ) = 𝑔 𝑓 𝑎𝑏𝑒 (𝑘 1 − 𝑘 2 ) · 𝜀 𝑠 ( 𝑝) (5)

A (3𝑐𝜙 , 4𝑑𝜙 , −𝑝 𝐴𝑠¯ 𝑒 ) = 𝑔𝑓 𝑒𝑐𝑑
(𝑘 3 − 𝑘 4 ) · 𝜀 𝑠¯ (−𝑝) . (6)

Note that these amplitudes are entirely on-shell and should be gauge invariant. On-shell three-point
kinematics means that 𝑘 𝑖 · 𝑝 = 0 for every external leg, so in each case we see that gauge invariance,

PoS(TASI2022)002
A (1, 2, 𝑝)| 𝜀→ 𝑝 = 0 = A (3, 4, −𝑝)| 𝜀→− 𝑝 , (7)

is entirely manifest.
Using Eqn. 1 we simply evaluates the sum over states in Eqn. 4 as,
∑︁
C1 = 𝑔 2 𝑓 𝑎𝑏𝑒 𝑓 𝑒𝑐𝑑 (𝑘 1 − 𝑘 2 ) 𝜇 (𝑘 3 − 𝑘 4 ) 𝜈 𝜀 𝑠 ( 𝑝)𝜀 𝑠¯ (−𝑝) (8)
𝑠 ∈pols
 
2 𝑎𝑏𝑒 𝑒𝑐𝑑 𝜇𝜈 𝑝𝜇𝑞𝜈 + 𝑞𝜇 𝑝𝜈
=𝑔 𝑓 𝑓 (𝑘 1 − 𝑘 2 ) 𝜇 (𝑘 3 − 𝑘 4 ) 𝜈 𝜂 − (9)
𝑝·𝑞
= 𝑔 2 𝑓 𝑎𝑏𝑒 𝑓 𝑒𝑐𝑑 (𝑘 1 − 𝑘 2 ) · (𝑘 3 − 𝑘 4 ) . (10)

This is precisely the numerator of the Feynman-rule dressed 𝑠-channel graph evaluated with on-shell
conditions. Note that, following maximal cut-construction, we would assign such a functional dress-
ing to the four-point cubic topology. If we label the s-channel cubic 4-point graph (1𝑎 , 2𝑏 , 3𝑐 , 4𝑑 ),
then we would say:

𝑁 𝑠 = 𝑁 (1𝑎 , 2𝑏 , 3𝑐 , 4𝑑 ) = 𝑔 2 𝑓 𝑎𝑏𝑒 𝑓 𝑒𝑐𝑑 (𝑘 1 − 𝑘 2 ) · (𝑘 3 − 𝑘 4 ) . (11)

Note, in making this assignment, we are no longer assuming maximal cut conditions, that (𝑘 1 +
𝑘 2 ) 2 = 0. The other two channels would follow from simple relabeling of this graph’s dressings:
𝑁𝑡 = 𝑁 (4𝑑 , 1𝑎 , 2𝑏 , 3𝑐 ), and 𝑁𝑢 = 𝑁 (3𝑐 , 1𝑎 , 4𝑑 , 2𝑏 ), with the putative full amplitude given:

𝑁𝑠 𝑁𝑡 𝑁𝑢
A= 2
+ 2
+ . (12)
(𝑘 1 + 𝑘 2 ) (𝑘 2 + 𝑘 3 ) (𝑘 1 + 𝑘 3 ) 2

In this case there are no more cut conditions to release – so we have recovered all information about
this amplitude that is available via unitarity cuts. For this theory we actually have the amplitude,
but we could, in principle, be missing contact terms – four-field operators entirely unconstrained
by factorization probes. Indeed the coefficients of contact terms, including the coefficients of local
higher derivative operators are generically informed in unitarity based amplitude bootstraps by
additional considerations: principles like positivity, symmetry properties, soft behavior, power-
counting, and gauge invariance.
Now let us consider a massless Dirac fermion 𝜓, in the fundamental of some 𝑆𝑈 (𝑁) gauge
group, minimally coupled to glue,

1
L = 𝜓(𝑖 / − Tr(F2 ) .
¯ 𝐷)𝜓 (13)
4

4
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

To illustrate the state projector for fermions it’s convenient to consider again a cut of a four-point
¯
amplitude, A (1𝑖𝑓 , 2̄ 𝑘𝑓¯ , 3𝑎𝐴, 4𝑏𝐴), to two three-point amplitudes, in this case with a fermion crossing
the cut

𝑗¯
∑︁ ¯
𝑗
C2 = A (1𝑖𝑓 , 3𝑎𝐴, − 𝑝¯ 𝑠¯ )A ( 𝑝 𝑠 , 4𝑏𝐴, 2̄ 𝑘𝑓¯ ) (14)
𝑠 ∈states
∑︁
= (𝑔𝑇𝑖𝑎𝑗¯ 𝑢¯ 1 𝜖/3 𝑣 − 𝑝, 𝑠¯ ) (𝑔𝑇 𝑗𝑏𝑘¯ 𝑢¯ 𝑝,𝑠 𝜖/4 𝑣 2 ) (15)
𝑠 ∈states
= 𝑔 2𝑇𝑖𝑎𝑗¯𝑇 𝑗𝑏𝑘¯ 𝑢¯ 1 𝜖/3 ( 𝑘/1 + 𝑘/3 )/𝜖 4 𝑣 2 . (16)

PoS(TASI2022)002
Again we recover precisely the Feynman-rule dressed numerator of the cut graph evaluated under
cut-conditions. This is again manifestly gauge-invariant under such restricted cut-kinematics.
The upside of such maximal cut construction is that the entire calculation at every stage is
expressible in a set of entirely gauge-invariant observables. Have we evaded the factorial complexity
of the number of Feynman graphs (including contact terms) that one must construct traditionally?
By no means – for brute construction we must maximally consider exactly the number of cuts as
equivalent to those allowed by Feynman rules, again including contact graphs. The advantage as
described above is simply in the compactness of resulting expressions, the systematic verifiability,
and the ability to bootstrap – recycling lower-order in coupling prediction to build higher-order in
coupling prediction. Indeed we can interpret the above approach as an efficient means of executing
a Feynman-rule based calculation.
We can supplement such brute construction with insightful guesses. We call such guesses
ansatze, and we can invoke a variety of physical principles and symmetries to constrain such
ansatze so that even at relatively high loop order in symmetric theories the size of the calculation
can still be manageable. This gains tremendous power from the notion of spanning cuts. If cut
𝐴 is describable by imposing additional cut conditions on cut 𝐵, than we say cut 𝐵 spans cut 𝐴.
If cut 𝐵 is satisfied by a set of mappings from graphs to dressings, than every cut spanned by 𝐵
will also be satisfied by that set of mappings. So verification of a dressing can proceed on a very
small number of cuts – a minimal set of spanning cuts. This means that once an ansatz has been
constrained on easy to preform near-maximal cuts, only a few large cuts are necessary to verify
that we have reproduced the amplitude required for the theory. This is what allows maximally
supersymmetric gauge and gravitational calculations to ascend to the loop order they have – at four
points the five loop correction has been completed for the maximally supersymmetric gravitational
theory at four-points, and six loops for the maximally supersymmetric gauge theory.

3. Duality between Color and Kinematics and the Double Copy

In 2008, Zvi Bern, Henrik Johansson, and I (BCJ) discovered a particularly powerful set of
constraints for gauge theories at tree-level, which we generalized to multi-loop integrands in 2010.
For many theories in the adjoint we can require the kinematic numerator weights of graphs to act
in concordance or duality with the color weights – obeying the same algebraic relations between
distinct graphs. The covariantized simple scalar four-point amplitude discussed above is given in

5
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

such a representation. We can write 𝑁 (1𝑎 , 2𝑏 , 3𝑐 , 4𝑑 ) = 𝑐(1𝑎 , 2𝑏 , 3𝑐 , 4𝑑 )𝑛(1, 2, 3, 4), where

𝑛(1, 2, 3, 4) = 𝑔 2 (𝑘 1 − 𝑘 2 ) · (𝑘 3 − 𝑘 4 ) (17)
𝑎 𝑏 𝑐 𝑑 𝑎𝑏𝑒 𝑒𝑐𝑑
𝑐(1 , 2 , 3 , 4 ) = 𝑓 𝑓 . (18)

Note the antisymmetry of the color-factors is mirrored in 𝑛(1, 2, 3, 4) = −𝑛(2, 1, 3, 4). More
surprisingly and far reaching is the fact that the color Jacobi relations are also mirrored:

𝑐 𝑠 = 𝑐𝑡 + 𝑐𝑢 (19)

PoS(TASI2022)002
𝑛 𝑠 = 𝑛𝑡 + 𝑛𝑢 . (20)

Here I followed the same relabeling conventions of the 𝑁𝑖 above.

The adjoint color-factors for all distinct (2𝑛 − 5)!! cubic graphs contributing to any 𝑛-point
amplitudes can be reduced via Jacobi to the color-weights of (𝑛 − 2)! graphs – the so called
half-ladder graphs where the two-outermost legs are held fixed (say legs 1 and 𝑛) and the rest
are permuted. The same reduction holds for the kinematic graphs. But when we make our graphs
functional via an ansatze we are dressing every half-ladder topology with the same function – simply
with permuted arguments. This means that we only need to give a single ansatz to the half-ladder
at any multiplicity, and color-dual kinematic relations propagate it to all graph topologies. This is
called making manifest the duality between color and kinematics. This not only drastically reduces
the number of graphs we must dress with an ansatz, but allows the resulting amplitude to participate
in double-copy construction – recycling kinematic building blocks to generate predictions in a wide
web of theories from a relatively small number of distinct primary theories.
Let me introduce the following setup for discussing adjoint double-copy theories. Consider
a gauge theory scattering amplitude expressed in terms of a sum over all distinct external labels
permutations of cubic-graphs Γ3 :
∫ ∑︁
1 𝑛𝑔 𝑐 𝑔
A= . (21)
𝑔 ∈Γ
𝑆𝑔 𝑑𝑔
3

Here again 𝑛𝑔 are called the kinematic numerator weights and contain all kinematic dependence
outside of the denominator weights 𝑑 𝑔 which encode the standard propagator structure, and the
𝑐 𝑔 are the graphs dressed with structure constants 𝑓 𝑎𝑏𝑐 at each vertex. The 𝑆 𝑔 represent the
internal symmetry factors of the graphs. The integral is over all internal loop momenta. It is
worth emphasizing that as a result of Jacobi relations the 𝑐 𝑔 are not independent. That means that
individual 𝑛𝑔 are not required by any means to be gauge-invariant. In general gauge invariance for
vector theories will occur in conspiracy between graphs as a result of the algebraic identities obeyed
by the 𝑐 𝑔 .
If we have also identified a set of kinematic dressing 𝑛˜ 𝑔 that obey the same algebraic relations
as arbitrary adjoint 𝑐 𝑔 , then we can replace the 𝑐 𝑔 with 𝑛˜ 𝑔 without violating the gauge invariance
of the full amplitude:

1 𝑛𝑔 𝑛˜ 𝑔
∫ ∑︁
𝑛⊗ 𝑛˜
A = . (22)
𝑔 ∈Γ
𝑆𝑔 𝑑𝑔
3

6
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

If both 𝑛 and 𝑛˜ were kinematic weights of vector gauge theories than the resulting double-copy
constructed amplitude will be a gravitational theory. The linearized gauge invariance of the vector
theories is promoted through the double-copy to linearized diffeomorphism invariance.
Are all gauge theories adjoint color-dual? The answer perhaps not so surprisingly is no but for
pure vectors to find an example we must consider higher-derivative operators.

4. EFT, the double-copy, and the emergence of string theory

Consider the higher-derivative operator 𝑇𝑟 (𝐹 4 ) compatible with supersymmetry. The four-

PoS(TASI2022)002
point amplitude associated with this operator insertion goes as:
4
A 𝐹 = 𝑑 𝑎𝑏𝑐𝑑 A BI (23)

where A BI is the four-point permutation-invariant Born-Infeld photon amplitude, and 𝑑 𝑎𝑏𝑐𝑑 is the
completely permutation invariant normalized sum over all distinct four-point color-traces. While
this may be color-dual in a permutation-invariant sense, it is not color-dual in any adjoint sense. In-
triguingly the amplitudes in this theory can be written as an adjoint double-copy because Born-Infeld
amplitudes are given as an adjoint double-copy between the kinematics of Yang-Mills amplitudes
and NLSM pion amplitudes.
This just begins to scratch the surface of the web of theories, but there is an important hint
that provides the key to recovering both open and closed string theory amplitudes as field theory
adjoint double-copies. Instead of rigidly considering double-copy only between kinematic weights
𝑛, allow mixed color and kinematic functions 𝑧 𝑔 that obey anti-symmetry and Jacobi. For example,
consider the following function:

𝑧 𝑠(2) = 𝑧(1𝑎 , 2𝑏 , 3𝑐 , 4𝑑 ) = 𝑑 𝑎𝑏𝑐𝑑 𝑛 𝜋 (1, 2, 3, 4) (24)

where  
𝑛 𝜋 (1, 2, 3, 4) = (𝑘 1 + 𝑘 2 ) 2 (𝑘 1 + 𝑘 3 ) 2 − (𝑘 1 + 𝑘 4 ) 2 /3 (25)

Define 𝑧 𝑡(2) and 𝑧 𝑢(2) by the same relabeling as above. We see that 𝑧 𝑠(2) = 𝑧 𝑡(2) + 𝑧 𝑢(2) – the permutation
invariant color-weight 𝑑 𝑎𝑏𝑐𝑑 doesn’t affect the algebraic relations obeyed by the kinematics. It’s
not hard to see that:
𝑐𝑠 𝑧𝑠 𝑐𝑡 𝑧𝑡 𝑐𝑢 𝑧𝑢
𝑑 𝑎𝑏𝑐𝑑 A NLSM = 2
+ 2
+ . (26)
(𝑘 1 + 𝑘 2 ) (𝑘 2 + 𝑘 3 ) (𝑘 1 + 𝑘 3 ) 2
4
The construction of A 𝐹 as a double-copy follows from replacing the 𝑐 𝑔 on the RHS with the
kinematic weights of Yang-Mills.
We can consider amplitudes in a very general bi-colored adjoint color-dual scalar theory with
mass scale 1/𝛼 0 and all order in 𝛼 0 color-dual corrections:
∞
∑︁ ∑︁ 𝑧 𝑔(𝑖) 𝑐 𝑔
0𝑖
Z= 𝛼 𝑊𝑖 . (27)
𝑔 ∈Γ3 𝑖=0
𝑑𝑔

Here the 𝑐 𝑔 are in the adjoint, but the 𝑧 𝑔 are allowed to depend on arbitrary color-traces as well
as kinematics – the only constraint is that they satisfy adjoint color-kinematics duality and the

7
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

amplitudes support factorization. With the appropriate Wilson coefficients 𝑊𝑖 , these amplitudes
are known as 𝑍-theory amplitudes and double-copy with supersymmetric Yang-Mills to build the
tree-level amplitudes of the open superstring to all multiplicity.
It turns out the construction of the 𝑧 𝑔(𝑖) in Z, at least at low multiplicity, is not as daunting a task
as it might seem. One can identify functional compositions for algebraic relations such that if 𝑥 𝑔 and
𝑦 𝑔 both satisfy algebraic relations, a 𝑧 𝑔 = compalg (𝑥, 𝑦)𝑔 will obey such algebraic relations. This
allows one to start with an algebraic weight linear in Mandelstam invariants, and via composition
climb to the UV. Four points provides a very simple example. Let’s define at four-points,

PoS(TASI2022)002
compadj (𝑥, 𝑦) 𝑠 = 𝑥 𝑡 𝑦 𝑡 − 𝑥 𝑢 𝑦 𝑢 (28)

It is not hard to see that we can build the pion numerator from the composition between covariatized
scalar numerators given above,
𝑛𝑠𝜋 ∝ 𝑛2𝑡 − 𝑛2𝑢 . (29)
One could continue climbing the ladder to higher mass dimension scalar adjoint numerator weights,
but there a fantastic surprise awaits us – all higher mass-dimension adjoint scalar numerators can
be expressed in terms of powers of the two scalar permutation invariants 𝜎2 = 𝑠2 + 𝑡 2 + 𝑢 2 and
𝜎3 = 𝑠𝑡𝑢 and a basis of 𝑛𝑔 and 𝑛 𝜋 . As we span all distinct color-traces at four-points with 𝑑 𝑎𝑏𝑐𝑑
and 𝑐 𝑠 , 𝑐 𝑡 , and 𝑐 𝑢 – one can span all 𝑧 (𝑖) to any mass dimension 𝑖 with a very small number of
building blocks indeed. What physical principles beyond unitarity and good UV behavior constrain
the Wilson coefficients 𝑊𝑖 appearing in the open superstring remains very much an open question.

5. Further reading

• For a pedagogic discussion of graph-based methods in amplitudes, including unitarity meth-

ods, color-kinematics and the double copy, see my earlier 2014 TASI lecture notes [1].

• For a gentle (non-technical) introduction to double-copy and applications to particle-physics,

cosmology, and gravitational wave astrophysics, written for the broader particle physics
community, please see the Snowmass white-paper [2].

• For applications of similar and related ideas to a Cosmological Bootstrap, there is a gentle
introduction to the ideas and literature via Snowmass in ref. [3].

• For a slightly more technical invitation to learn more about double-copy and amplitudes
and the web of color-dual theories, I recommend the SAGEX review [4]. Indeed the entire
SAGEX compilation [5] provides an excellent overview of many research directions in the
amplitudes community.

• In 2019 we presented a comprehensive, at the time, technical review of double-copy and it’s
applications in ref. [6]. In this review we work out some low-multiplicity bootstrapping of
NLSM (Chiral Lagrangian pions) and YM in Chapter 3.

• For lecture notes in the form of a mathematica notebook, please see my Amplitudes 2017
Summer School lectures 3 of [7]. Uses some public graph manipulation code I have on github
3Video for the lectures are linked to at https://prettyquestions.com/lectures/2017_Amplitudes_School_Lectures/.

8
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

based more or less on how I talk about things in my TASI 2014 lectures. This goes into
some details on cut verification, cut construction, and double-copy to gravity amplitudes at
2-loops.

• For bootstrapping massive scalar QCD in arbitrary representations with color-kinematics

see ref. [8] and extracting out pure Einstein-Hilbert gravity mediated massive scalars via
double-copy and projection [9]. Similar bootstrap for actual QCD with quarks through 4-
point 1-loop in the fundamental given in ref. [10]. This later reference provides some nice
pedagogic discussion of functional fermionic ansatze.

PoS(TASI2022)002
• For bootstrapping (higher)-spin massive amplitudes using massive spinor helicity, including
imposing Majorana vs Dirac conditions, see Henrik Johanson and Alex Ochirov’s nice paper
[11].

• For detailed discussion of color-dual construction of higher-derivative EFT operator predic-

tions in gauge and gravity theories I would take a look at refs. [12–16].

For broader context and discussion specializing to four-dimensions, I would absolutely recom-
mend the following:
• For bootstrapping massless theories with spinor helicity at 3-points see Clifford Cheung’s
TASI notes [17] and and Paolo Benincasa and Freddy Cachazo’s seminal paper [18]. It’s an
excellent exercise to carry out this exercise in 𝐷-dimensions at three-points both for massive
and massless matter.

• Definitely check out Lance Dixon teaching amplitudes methods and ideas to phenomenol-
ogists, in TASI-1995 [19] – which is where I learned spinor helicity, and TASI-2013 [20].
I believe Lance lives in mostly minus convention, so it’s good to compare with Henriette
Elvang and Yu-tin Huang’s awesome textbook [21] which is in mostly-plus.

6. Acknowledgements

I am absolutely delighted to thank the organizers for the opportunity to lecture at TASI once
again and for generating a fantastic environment as always. I also want to thank the students — and
the other lecturers — for all the great questions and inspiring discussions – what a wonderful return
post-pandemic! This work is supported in part by the DOE under contract DE-SC0015910 and by
the Alfred P. Sloan Foundation.

References

[1] J.J.M. Carrasco, Gauge and Gravity Amplitude Relations, in Theoretical Advanced Study
Institute in Elementary Particle Physics: Journeys Through the Precision Frontier:
Amplitudes for Colliders, pp. 477–557, WSP, 2015, DOI [1506.00974].

[2] T. Adamo, J.J.M. Carrasco, M. Carrillo-González, M. Chiodaroli, H. Elvang, H. Johansson

et al., Snowmass White Paper: the Double Copy and its Applications, in 2022 Snowmass
Summer Study, 4, 2022 [2204.06547].

9
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

[3] D. Baumann, D. Green, A. Joyce, E. Pajer, G.L. Pimentel, C. Sleight et al., Snowmass White
Paper: The Cosmological Bootstrap, in Snowmass 2021, 3, 2022 [2203.08121].

[4] Z. Bern, J.J. Carrasco, M. Chiodaroli, H. Johansson and R. Roiban, The SAGEX Review on
Scattering Amplitudes, Chapter 2: An Invitation to Color-Kinematics Duality and the
Double Copy, 2203.13013.

[5] G. Travaglini et al., The SAGEX review on scattering amplitudes, J. Phys. A 55 (2022)
443001 [2203.13011].

PoS(TASI2022)002
[6] Z. Bern, J.J. Carrasco, M. Chiodaroli, H. Johansson and R. Roiban, The Duality Between
Color and Kinematics and its Applications, 1909.01358.

[7] J.J. Carrasco, Graphical Methods for Sharp Predictions. "Live" Lecture Notes - Amplitudes
2017 Summer School, https://doi.org/10.6084/m9.figshare.5197213.v2.

[8] J.J.M. Carrasco and I.A. Vazquez-Holm, Loop-Level Double-Copy for Massive Quantum
Particles, Phys. Rev. D 103 (2021) 045002 [2010.13435].

[9] J.J.M. Carrasco and I.A. Vazquez-Holm, Extracting Einstein from the loop-level
double-copy, JHEP 11 (2021) 088 [2108.06798].

[10] J.J.M. Carrasco and A. Seifi, Loop-level double-copy for massive fermions in the
fundamental, JHEP 05 (2023) 217 [2302.14861].

[11] H. Johansson and A. Ochirov, Double copy for massive quantum particles with spin, JHEP
09 (2019) 040 [1906.12292].

[12] J.J.M. Carrasco, L. Rodina, Z. Yin and S. Zekioglu, Simple encoding of higher derivative
gauge and gravity counterterms, Phys. Rev. Lett. 125 (2020) 251602 [1910.12850].

[13] J.J.M. Carrasco, L. Rodina and S. Zekioglu, Composing effective prediction at five points,
JHEP 06 (2021) 169 [2104.08370].

[14] J.J.M. Carrasco, M. Lewandowski and N.H. Pavao, The color-dual fates of 𝐹 3 , 𝑅 3 , and
N = 4 supergravity, 2203.03592.

[15] J.J.M. Carrasco and N.H. Pavao, Virtues of a symmetric-structure double copy, Phys. Rev. D
107 (2023) 065005 [2211.04431].

[16] J.J.M. Carrasco, M. Lewandowski and N.H. Pavao, Double-copy towards supergravity
inflation with 𝛼-attractor models, JHEP 02 (2023) 015 [2211.04441].

[17] C. Cheung, TASI Lectures on Scattering Amplitudes, in Proceedings, Theoretical Advanced

Study Institute in Elementary Particle Physics : Anticipating the Next Discoveries in Particle
Physics (TASI 2016): Boulder, CO, USA, June 6-July 1, 2016, R. Essig and I. Low, eds.,
pp. 571–623 (2018), DOI [1708.03872].

10
TASI 2022 lectures on scattering amplitudes – an addendum John Joseph M. Carrasco

[18] P. Benincasa and F. Cachazo, Consistency Conditions on the S-Matrix of Massless Particles,
0705.4305.

[19] L.J. Dixon, Calculating scattering amplitudes efficiently, in Theoretical Advanced Study
Institute in Elementary Particle Physics (TASI 95): QCD and Beyond, pp. 539–584, 1, 1996
[hep-ph/9601359].

[20] L.J. Dixon, A brief introduction to modern amplitude methods, in Theoretical Advanced
Study Institute in Elementary Particle Physics: Particle Physics: The Higgs Boson and
Beyond, pp. 31–67, 2014, DOI [1310.5353].

PoS(TASI2022)002
[21] H. Elvang and Y.-t. Huang, Scattering Amplitudes, 1308.1697.

11