[trammbar] updating documentation, gitignore, and CHANGELOG

.gitignore

@@ -10,6 +10,8 @@ ext/tram_direct/tram_direct.c
 ext/dtram/dtram.c
 ext/bar/bar.c
 ext/util/util.c
+ext/trammbar/trammbar.c
+ext/trammbar_direct/trammbar_direct.c
 build
 dist
 MANIFEST

CHANGELOG

@@ -92,4 +92,8 @@ Version 0.2.0 (Beta):
     - Use ttrajs, dtrajs instead of the 2D dtraj argument in state_counts() and count_matrices()
     - New feature: the state-continuous TRAM estimator
 
+Version 0.2.1 (Beta):
+
+    - New feature: TRAMMBAR, an extension of the TRAM estimator
+
 Upcoming:

docs/api/api.rst

@@ -8,4 +8,6 @@ API documentation
 
    wham
    dtram
+   tram
+   trammbar
    util

docs/api/tram.rst

@@ -0,0 +1,10 @@
+====
+TRAM
+====
+
+Contents:
+
+.. toctree::
+   :maxdepth: 2
+
+.. automodule:: thermotools.tram

docs/api/trammbar.rst

@@ -0,0 +1,10 @@
+========
+TRAMMBAR
+========
+
+Contents:
+
+.. toctree::
+   :maxdepth: 2
+
+.. automodule:: thermotools.trammbar

ext/tram/tram.pyx

@@ -596,10 +596,13 @@ def estimate(count_matrices, state_counts, bias_energy_sequences, state_sequence
     loglikelihoods : numpy.ndarray(dtype=numpy.float64, ndim=1)
         stored sequence of loglikelihoods
 
-    Termination
-    -----------
+    Note
+    ----
     The self-consitent iteration terminates when
-    .. math:: max\{\max_{i,k}{\Delta \pi_i^k}, \max_k \Delta f^k \}<\mathrm{maxerr}.
+
+    .. math::
+       \max\{\max_{i,k}{\Delta \pi_i^k}, \max_k \Delta f^k \}<\mathrm{maxerr}.
+
     Different termination criteria can be implemented with the callback
     function. Raising `CallbackInterrupt` in the callback will cleanly
     terminate the iteration.

ext/trammbar/trammbar.pyx

@@ -184,9 +184,6 @@ def update_biased_conf_energies(
         return_log_L = True)
     return_log_L : bool
         If true, retrun the TRAM-log-likelihood.
-
-    TRAMMBAR parameters
-    -------------------
     therm_energies : numpy.ndarray(shape=(T), dtype=numpy.float64)
         reduced thermodynamic free energies, must match `biased_conf_energies`
     equilibrium_bias_energy_sequences : list of numpy.ndarray(shape=(X_i, T), dtype=numpy.float64), optional
@@ -281,9 +278,6 @@ def get_log_Ref_K_i(
         scratch array for logsumexp operations
     log_R_K_i : numpy.ndarray(shape=(T, M), dtype=numpy.float64)
         target array for sum of TRAM log pseudo-counts and biased_conf_energies
-
-    TRAMMBAR parameters
-    -------------------
     therm_energies : numpy.ndarray(shape=(T), dtype=numpy.float64)
         reduced thermodynamic free energies, must match `biased_conf_energies`
     equilibrium_therm_state_counts : numpy.ndarray(shape=(T), dtype=numpy.intc)
@@ -330,9 +324,6 @@ def get_conf_energies(
         precomputed sum of TRAM log pseudo-counts and biased_conf_energies
     scratch_T : numpy.ndarray(shape=(T), dtype=numpy.float64)
         scratch array for logsumexp operations
-
-    TRAMMBAR parameters
-    -------------------
     equilibrium_bias_energy_sequences : list of numpy.ndarray(shape=(X_i, T), dtype=numpy.float64), optional
         reduced bias energies in the T thermodynamic states for all X equilibrium samples
     equilibrium_state_sequences : list of numpy.ndarray(shape=(X_i,), dtype=numpy.intc), optional
@@ -474,9 +465,6 @@ def get_pointwise_unbiased_free_energies(
         scratch array for logsumexp operations
     pointwise_unbiased_free_energies : list of numpy.ndarray(shape=(X_i), dtype=numpy.float64)
         target arrays for the pointwise free energies
-
-    TRAMMBAR parameters
-    -------------------
     equilibrium_therm_state_counts : numpy.ndarray(shape=(T), dtype=numpy.intc), optional
         number of equilibrium frames per thermodynamic state, can be zero
     overcounting_factor : double, default = 1.0
@@ -643,9 +631,6 @@ def log_likelihood_lower_bound(
         scratch array for logsumexp operations
     scratch_MM : numpy.ndarray(shape=(M, M), dtype=numpy.float64)
         scratch array for likelihood computation
-
-    TRAMMBAR parameters
-    -------------------
     therm_energies : numpy.ndarray(shape=(T), dtype=numpy.float64)
         reduced thermodynamic free energies, must match `biased_conf_energies`
     equilibrium_bias_energy_sequences : list of numpy.ndarray(shape=(X_i, T), dtype=numpy.float64), optional
@@ -729,9 +714,6 @@ def estimate(count_matrices, state_counts, bias_energy_sequences, state_sequence
         initial guess for the logarithm of the Lagrangian multipliers
     N_dtram_accelerations : int
         not used
-
-    TRAMMBAR parameters
-    -------------------
     equilibrium_therm_state_counts : numpy.ndarray(shape=(T,), dtype=numpy.intc), optional
         number of equilibrium frames per thermodynamic state, can be zero
     equilibrium_bias_energy_sequences : list of numpy.ndarray(shape=(X_i, T), dtype=numpy.float64), optional
@@ -758,10 +740,13 @@ def estimate(count_matrices, state_counts, bias_energy_sequences, state_sequence
     loglikelihoods : numpy.ndarray(dtype=numpy.float64, ndim=1)
         stored sequence of loglikelihoods
 
-    Termination
-    -----------
+    Note
+    ----
     The self-consitent iteration terminates when
-    .. math:: max\{\max_{i,k}{\Delta \pi_i^k}, \max_k \Delta f^k \}<\mathrm{maxerr}.
+
+    .. math::
+       \max\{\max_{i,k}{\Delta \pi_i^k}, \max_k \Delta f^k \}<\mathrm{maxerr}.
+
     Different termination criteria can be implemented with the callback
     function. Raising `CallbackInterrupt` in the callback will cleanly
     terminate the iteration.