Added CFD balancing

Author: ngreifer

NEWS.md

@@ -1,6 +1,10 @@
 WeightIt News and Updates
 ======
 
+# `WeightIt` (development version)
+
+* Added `method = "cfd"` for characteristic function distance balancing as described by [Santra, Chen, and Park (2026)](http://arxiv.org/abs/2601.15449). Energy balancing is a special case of this method.
+
 # `WeightIt` 1.5.1
 
 * `calibrate()` with `method = "isoreg"` can now be used with sampling weights.

R/weightit.R

@@ -119,6 +119,7 @@
 #' | [`"super"`][method_super] | Propensity score weighting using SuperLearner |
 #' | [`"bart"`][method_bart] | Propensity score weighting using Bayesian additive regression trees (BART) |
 #' | [`"energy"`][method_energy] | Energy balancing |
+#' | [`"cfd"`][method_cfd] | Characteristic function distance balancing |
 #'
 #' `method` can also be supplied as a user-defined function; see [`method_user`]
 #' for instructions and examples. Setting `method = NULL` computes unit weights.

R/weightit2cfd.R

@@ -232,5 +232,23 @@
     density_ok = TRUE,
     stabilize_ok = TRUE,
     plot.weightit_ok = FALSE
+  ),
+  "cfd" = list(
+    treat_type = c("binary", "multinomial"),
+    estimand = c("ATE", "ATT", "ATC"),
+    alias = c("cfd", "kernel"),
+    description = "characteristic function distance balancing",
+    ps = FALSE,
+    msm_valid = FALSE,
+    msm_method_available = FALSE,
+    subclass_ok = FALSE,
+    packages_needed = "osqp",
+    s.weights_ok = TRUE,
+    missing = "ind",
+    moments_int_ok = TRUE,
+    moments_default = 0,
+    density_ok = FALSE,
+    stabilize_ok = FALSE,
+    plot.weightit_ok = FALSE
   )
 )}

R/weightit_methods.R

@@ -10,7 +10,7 @@ knitr::opts_chunk$set(
   warning = FALSE,
   message = FALSE,
   tidy = FALSE,
-  fig.align='center',
+  fig.align = 'center',
   comment = "##",
   fig.path = "man/figures/README-"
 )
@@ -94,6 +94,7 @@ The tables below contain the available methods in *WeightIt* for estimating weig
  SuperLearner PS | [`"super"`](https://ngreifer.github.io/WeightIt/reference/method_super.html)
  Bayesian additive regression trees PS | [`"bart"`](https://ngreifer.github.io/WeightIt/reference/method_bart.html)
  Energy balancing | [`"energy"`](https://ngreifer.github.io/WeightIt/reference/method_energy.html)
+ Characteristic function distance balancing | [`"cfd"`](https://ngreifer.github.io/WeightIt/reference/method_cfd.html)
 
 #### Multi-Category Treatments
 
@@ -109,7 +110,8 @@ The tables below contain the available methods in *WeightIt* for estimating weig
  SuperLearner PS | [`"super"`](https://ngreifer.github.io/WeightIt/reference/method_super.html)
  Bayesian additive regression trees PS | [`"bart"`](https://ngreifer.github.io/WeightIt/reference/method_bart.html)
  Energy balancing | [`"energy"`](https://ngreifer.github.io/WeightIt/reference/method_energy.html)
- 
+ Characteristic function distance balancing | [`"cfd"`](https://ngreifer.github.io/WeightIt/reference/method_cfd.html)
+
 #### Continuous Treatments
 
  Method               | `method`

README.Rmd

@@ -169,6 +169,7 @@ install them.
 | SuperLearner PS | [`"super"`](https://ngreifer.github.io/WeightIt/reference/method_super.html) |
 | Bayesian additive regression trees PS | [`"bart"`](https://ngreifer.github.io/WeightIt/reference/method_bart.html) |
 | Energy balancing | [`"energy"`](https://ngreifer.github.io/WeightIt/reference/method_energy.html) |
+| Characteristic function distance balancing | [`"cfd"`](https://ngreifer.github.io/WeightIt/reference/method_cfd.html) |
 
 #### Multi-Category Treatments
 
@@ -184,6 +185,7 @@ install them.
 | SuperLearner PS | [`"super"`](https://ngreifer.github.io/WeightIt/reference/method_super.html) |
 | Bayesian additive regression trees PS | [`"bart"`](https://ngreifer.github.io/WeightIt/reference/method_bart.html) |
 | Energy balancing | [`"energy"`](https://ngreifer.github.io/WeightIt/reference/method_energy.html) |
+| Characteristic function distance balancing | [`"cfd"`](https://ngreifer.github.io/WeightIt/reference/method_cfd.html) |
 
 #### Continuous Treatments