Package index • femtograd

All functions

abs(<value>): Absolute value for value objects
anova(<femtofit>): Analysis of variance for femtofit models
backward(): Generic function for the Backward pass for automatic differentiation (finds the gradient of every sub-node in the computational graph with respect to e). In other words, it is responsible for computing the gradient with respect to e.
backward(<default>): Default implementation does not propagate gradients. For instance, if we have a constant, then the partial of the constant is not meaningful.
backward(<value>): Backward pass for value objects
bfgs(): BFGS quasi-Newton optimizer
bootstrap: Bootstrap Inference
bootstrap_fit(): Bootstrap standard errors and confidence intervals
bounded(): Transform to bounded interval
check_convergence(): Check convergence diagnostics
check_hessian(): Check Hessian properties
compare(): Compare multiple fitted models
confint(<bootstrap_result>): Confidence intervals from bootstrap
confint_mle(): Compute confidence intervals from MLE results
confint_profile(): Profile confidence intervals
cos(<value>): Cosine function for value objects
diagnostics(): Model Diagnostics
digamma(<value>): Digamma (psi) function for value objects
distributions: Log-likelihood functions for exponential family distributions
div_safe(): Safe division (handles division by zero)
dof(): Extract degrees of freedom from hypothesis test
`-`(<value>): Subtraction for value objects
dual: dual R6 class for forward-mode automatic differentiation
dual_num(): Create a dual number
exp(<value>): Exponential function for value objects
exp_safe(): Stable exp function (with overflow protection)
femtofit() coef(<femtofit>) vcov(<femtofit>) confint(<femtofit>) logLik(<femtofit>) nobs(<femtofit>) print(<femtofit>) summary(<femtofit>) print(<summary.femtofit>): Constructor for femtofit objects
find_mle(): Find MLE with standard errors
fisher_information(): Compute observed Fisher information matrix
fisher_scoring(): Fisher scoring optimizer
fit(): Fit a model via maximum likelihood
fitting: Statistical model fitting with automatic differentiation
get_data() `get_data<-`(): Retrieve the data stored by an object
grad(): Gradient of x with respect to e in backward(e), e.g., dx/de. (applies the chain rule)
grad(<default>): Default gradient is zero matrix
grad(<value>): Gradient of a value object x with respect to e in backward(e), e.g., dx/de. (applies the chain rule)
gradient(): Compute gradient as a numeric vector
gradient_ascent(): Gradient ascent/descent optimizer
gradient_descent(): Gradient descent (minimize)
hessian(): Compute Hessian matrix via forward-over-reverse automatic differentiation
hypothesis_tests: Hypothesis Testing for Fitted Models
inv_bounded(): Inverse of bounded transform
inv_positive(): Inverse of positive transform
inv_probability(): Inverse of probability transform
inverse_transforms: Inverse transforms for recovering original scale
is_dual(): Check if object is a dual number
is_significant_at(): Check if test is significant at given level
is_value(): Check if an object is of class value
lbfgs(): L-BFGS optimizer (limited memory BFGS)
lgamma(<value>): Log-gamma function for value objects
line_search(): Backtracking line search (Armijo condition)
log(<value>): Natural logarithm for value objects
log1p(<value>): Log(1+x) for value objects
log1p_safe(): Log1p with underflow protection
log_safe(): Safe logarithm (handles zeros)
log_sigmoid(): Log-sigmoid (numerically stable)
logit(): Logit function for value objects
loglik_bernoulli(): Bernoulli distribution log-likelihood
loglik_beta(): Beta distribution log-likelihood
loglik_binomial(): Binomial distribution log-likelihood
loglik_exponential(): Exponential distribution log-likelihood
loglik_gamma(): Gamma distribution log-likelihood
loglik_logistic(): Logistic regression log-likelihood (binary)
loglik_negbinom(): Negative binomial log-likelihood
loglik_normal(): Normal (Gaussian) log-likelihood
loglik_pareto(): Pareto distribution log-likelihood
loglik_poisson(): Poisson distribution log-likelihood
loglik_weibull(): Weibull distribution log-likelihood
logsumexp(): Log-Sum-Exp (numerically stable)
lower_bounded(): Transform to lower-bounded interval
lrt(): Likelihood Ratio Test
mean(<value>): Mean for value objects
newton_raphson(): Newton-Raphson optimizer
observed_info(): Observed Fisher information matrix
optimization: Optimization routines for maximum likelihood estimation
plot(<profile_likelihood>): Plot profile likelihood
`+`(<value>): Addition for value objects
positive(): Transform to positive values
`^`(<value>): Power operation for value objects.
predict(<femtofit>): Predictions from a fitted model
primal(): Extract primal from dual or return value unchanged
print(<anova.femtofit>): Print anova table for femtofit
print(<bootstrap_result>): Print method for bootstrap results
print(<hessian_check>) print(<convergence_check>) print(<model_diagnostics>): Print methods for diagnostic objects
print(<likelihood_ratio_test>): Print method for likelihood ratio test
print(<model_comparison>): Print model comparison
print(<profile_likelihood>): Print method for profile likelihood
print(<value>): Print value object and its computational graph
print(<wald_test>): Print method for Wald test
probability(): Transform to probability values
profile_likelihood: Profile Likelihood
profile_loglik(): Compute profile likelihood for a parameter
pval(): Extract p-value from hypothesis test
relu(): ReLU activation function for value objects
se(): Standard errors from a fitted model
se_reliable(): Check if standard errors are reliable
sigmoid(): Sigmoid activation function for value objects
sigmoid_stable(): Stable sigmoid function
sin(<value>): Sine function for value objects
`/`(<value>): Division for value objects
softmax(): Softmax function (numerically stable)
softplus(): Softplus function for value objects
sqrt(<value>): Square root for value objects
stability: Numerical stability utilities for automatic differentiation
std_errors(): Compute standard errors from Hessian
sum(<dual>): Sum for dual numbers
sum(<value>): Summation for value objects
summary(<bootstrap_result>): Summary for bootstrap results
tangent(): Extract tangent from dual or return 0
tanh(<value>): Hyperbolic tangent for value objects
test_stat(): Extract test statistic from hypothesis test
`*`(<value>): Multiplication for value objects
transforms: Parameter Transformation Helpers
trigamma(<value>): Trigamma function for value objects
upper_bounded(): Transform to upper-bounded interval
val(): value object constructor
value: value R6 class
vcov_matrix(): Compute variance-covariance matrix from Hessian
wald_test(): Wald Test for Model Parameters
zero_grad(): Reset gradients to zero